Quantitative Analysis of Passenger and Baggage Security Screening at Airports

Transcription

1 Journal of Advanced Transportation, Vol. 41, No. 2, pp Quantitative Analysis of Passenger and Baggage Security Screening at Airports Alexandre G. de Barros David D. Tomber The terrorist attacks in the United States in 2001 opened a new era in air transportation. The realization that civil aircraft can be used as powerful weapons of mass destruction by a small group of people has drastically increased the need for security screening procedures to protect civil flights. Serving as the interface between the air and land transportation modes, airports have become the main focus in the implementation of those procedures. The need to more thoroughly screen passengers and baggage, and the consequent increase in processing time, has created the need for more space for security checkpoints and baggage screening inside passenger terminal buildings space that is costly and very difficult to find in existing buildings. This paper evaluates the impact those measures have had on the planning and operation of airport passenger terminals. Quantification of those impacts is performed with the use of discrete-event simulation and spreadsheet models. 1. Introduction In the 80 s and 90 s, the rapid growth in air traffic and the consequent need for investments in infrastructure captured almost all the attention and resources of individuals and organizations involved in air transportation planning. During that period, air traffic grew at annual rates over 5%. Unfortunately, the expansion of the aeronautical infrastructure airports and air traffic control did not keep up with the growth in air traffic, leading to large imbalances between supply and demand, causing congestion in several airport systems around the world. A review of the literature on air transportation during the 80 s and 90 s Alexandre G. de Barros, University of Calgary, Department of Civil Engineering, Calgary AB, Canada David D. Tomber, Port of Seattle/Aviation Planning, Seattle-Tacoma International Airport, Seattle, Washington, USA Received: January 2007 Accepted: January 2007

2 172 A. G. de Barros and D. D. Tomber reveals the community s obsession with this imbalance (Tosic, 1992; Brunetta et al, 1999). The issue of safety, albeit important, was relegated to a lesser priority. This picture, however, changed drastically on September 11, Attacks on civil aircraft did not start with the XXI century. The loss of a large civil aircraft, besides the inestimable value of lost human lives, also incurs enormous financial loss. These financial losses include: the market value of the aircraft which can reach hundreds of millions of US dollars; the loss of direct revenue generated by that aircraft; and the reduction in traffic caused by the fear of further attacks. Since perpetrating such attacks is not as difficult as it should be, civil aircraft have become a favorite target for terrorists. Before 2001, three types of attacks on civil aircraft were the most common: Hijacking: the seizure of an aircraft with the intent of obtaining gains in exchange for its release. Bombing: an attack on an aircraft with the sole intent of destroying it and the lives aboard it. The most common technique used in aircraft bombings was the placing of a bomb in a bag that either traveled unaccompanied or with a person who did not know about it. In general, the perpetrators did not travel on the aircraft targeted. Robbing: in this case, the perpetrators goal is to steal valuables that are being transported aboard the aircraft. Note that, in all three types of attacks, the civil aircraft, with people and goods aboard, was the only target. That supported the argument that aircraft security was mainly the responsibility of the airline. As a result, little was done to improve the security of civil aircraft outside of the private sector, especially in countries with highly deregulated markets such as the ones in North America and, more recently, Western Europe. The terrorist attacks of 2001 in the United States changed that picture dramatically. Those acts introduced a new type of attack, in which the civil aircraft is not only a target but also a weapon used for the destruction of high-profile targets on land. They also introduced the concept of suicidal terrorism to civil aviation, since such attacks require the perpetrator s physical presence aboard the aircraft at the time of its destruction. Given the increasing activity of suicidal terrorists in the world (Jane s Airport Review, 2000), the need to better prevent against external attacks has become evident, not only for the protection of the airlines and their passengers, but also for the security of nations

3 Quantitative Analysis 173 and their citizens. Air transportation security has become a major public concern. The planning of terrorist attacks begin far away and months or even years in advance to the actual attacks. Evidently the best way to prevent such attacks is through the actions of intelligence services that allow the frustration of terrorist plans before their realization. Intelligence, however, is subject to failure. Therefore, it is at the airport the interface between aircraft and land transportation that we find the last opportunity to deny boarding to any persons or objects representing a threat to the integrity of the aircraft and the people it carries. This is the reason for the intensification of passenger and baggage inspection procedures that have been established by governments of many if not all countries. These new security measures at airports come with a high price. In the United States, the processing rate at the passenger security checkpoints dropped from passengers/h/lane up to September 2001, to passengers/h/lane immediately after. This drastic reduction in processing capacity has caused immense queues and long hours of waiting for passengers. It has also created the need for expansion of the inspection areas, which cannot always be performed due to tight space constraints. The requirement to screen 100% of checked bags has also created the need to procure and install new equipment that is expensive, large, heavy and slow i.e. it is required in large numbers at a given airport. These two improved security processes passenger and baggage screening have the most significant effects on the planning and design of airport passenger buildings. This paper evaluates the impact the new security measures have had on the planning and operation of airport passenger terminals. Quantification of those impacts is performed with the use of discreteevent simulation and spreadsheet models. A methodology for developing a spreadsheet model using deterministic queuing theory is presented and applied in the preliminary analysis of both passenger and baggage security screening systems. The paper also presents a novel framework outlining a rational process for baggage system planning and design. 2. Passenger Screening

4 174 A. G. de Barros and D. D. Tomber The pre-boarding screening of passengers is not a recent innovation: it has been in use in North America and Europe since the 60 s and 70 s. The purpose of this screening has always been to catch and confiscate objects that could be used as a weapon, a bomb or a bomb making material or otherwise pose a threat to flight security. In order to achieve that goal, it is necessary to inspect both the passenger and his/her carryon baggage for such objects. This inspection, however, must be as little intrusive as possible to avoid causing a high level of stress and discomfort to the passenger. In addition, the screening process must be quick in order to avoid a bottleneck in the passenger flow. The prohibition in 2006 of liquids carried by passengers, which caused severe delays at airports around the world, is an excellent example of the importance of an efficient screening process. In the last two decades, the model that has become the standard for passenger security checkpoints in most airports is the security channels. Each channel or lane is equipped with an arch-shaped magnetometer and an X-ray machine. Passengers are required to walk through the magnetometer which will sound an alarm if a metal object is detected. Meanwhile, the passenger s carry-on items such as purses, laptop computers and small bags are scanned by the X-ray machine. Passenger and carry-on baggage are reconciled at the end of the X-ray belt. In the United States, Canada and some other countries, secondary screening of carry-on items may be performed on selected items using explosive trace detection machines. The channels are laid out at points of access to the boarding gates, creating a secure area where all boarding gates are located and to which only people who have been screened have access. Figure 1 shows a conceptual example of the secure boarding area and the location of the security checkpoint. With the drastic reduction of up to 83% in the passenger screening processing rates, airports set out to investigate ways to improve the screening procedure and make it more efficient.

5 Quantitative Analysis 175 Figure 1. Secure boarding area 3. Baggage Screening Like passenger screening, baggage screening is not a new idea. Even before 9/11, several cases of attacks to civilian aircraft with explosives placed in checked baggage had prompted authorities of several countries set up in-line screening systems. The US Federal Aviation Administration (FAA) was already committed to 100% checked baggage screening by 2010 (Federal Aviation Administration, 2003). Airport baggage handling systems are complex systems even without baggage screening. They can account for a significant portion of the cost of building and operating an airport passenger terminal. Most airport terminals had already expanded their baggage systems to their limits and had no space for additional equipment. The sudden need to introduce Explosive Detection Systems (EDS) quickly as required by the Aviation and Transportation Security Act (ATSA), passed by US Congress on November 19, 2001, posed an enormous challenge to airport planners. The screening model adopted in the US, shown in Figure 2 is quite different and much less efficient than the 5-level one used in European

6 176 A. G. de Barros and D. D. Tomber From check-in LEVEL 1 CT Screening LEVEL 2 Explosive Trace Detecition LEVEL 3 Manual search Unscreened bags Cleared bags Suspicious bags To baggage make-up Figure 2. Baggage screening model used in the US From check-in LEVEL 1 AT Screening LEVEL 2 Image inspection LEVEL 3 CT Screening LEVEL 4 Manual search LEVEL 5 Controlled explosion Unscreened bags Cleared bags To baggage make-up Suspicious bags Figure 3. Baggage screening model used in Europe and Canada

7 Quantitative Analysis 177 and Canadian airports and illustrated in Figure 3. These models use an advanced X-ray technology known as AT scanning for the primary inspection of baggage. AT machines have a high throughput rate: bags/hour. Only alarmed bags will be examined by more sophisticated EDS machines, even so only after the AT image has been inspected and rejected by an officer. ATSA, however, requires the use of EDS machines for all bags, due in part to the fact that AT machines have not been certified by the FAA. EDS machines have considerably lower throughput rates demonstrated in-line performance ranges between bags/hour. They are also bigger, heavier and more expensive than AT machines. Existing baggage handling systems had to be retrofitted and, in many cases, completely redesigned to accommodate those machines. 4. Quantitative Analysis of Passenger and Baggage Screening Systems There are basically three types of tools available for the analysis of passenger screening systems: analytical, spreadsheet and simulation models. The choice of which tool to use will depend mainly on the level of detail required and on the time available for the analysis. Following is a discussion on the advantages and disadvantages of each type of tool Analytical models Analytical modeling of passenger and baggage screening processes consists in developing mathematical relationships between the main variables involved. Queuing theory is widely used for the analysis of queues in stochastic processes. Formulae relating waiting time, queue length, arrivals rate and processing rate have been developed based on simplifying assumptions for a number of queuing systems (Newell, 1982). Such formulae are simple to understand and use. In practice, however, the underlying assumptions made to simplify the formulation with queuing theory almost never hold for airport systems. Passenger arrival rates vary according to a flight schedule and a passenger arrivals profile. Processing rates depend on staff and equipment schedules. Furthermore, airlines and airport authorities are often more concerned with an estimate of the maximum queue length and waiting time.

8 178 A. G. de Barros and D. D. Tomber Deterministic queuing theory (Newell, 1982) can be more effective for airport planning purposes. A passenger arrivals profile can be obtained from a survey or derived from a flight schedule. It is possible to derive formulae for the maximum queue lengths and total waiting times for certain peak shapes. Figure 4 illustrates this process. As an example, if the peak between times t A and t B can be approximated to a symmetric quadratic parabola (Bandara and Wirasinghe, 1990), the total waiting time W is 3/ 2 2 t ( λ µ ) = B t A W (1) 3 λ λ µ 0 where µ = processing rate; λ = maximum arrival rate; λ 0 = arrival rate at t A ; t A = time of beginning of peak period; t B = time of ending of peak period. λ µ Q max Passengers/hour λ 0 t A t B Figure 4. Graphical representation of maximum queue length Time

9 Quantitative Analysis 179 This technique is very useful to quickly determine the system capacity necessary to satisfy a maximum waiting time requirement, usually between 10 and 20 minutes Spreadsheet models Spreadsheet models can be a mix of analytical models and simulation models. Using basic deterministic analytical formulas, it is possible to model multiple processes and quickly obtain maximum queue lengths and waiting times with a deterministic approach. Spreadsheet models are quick to build, simple to understand and easily expandable (de Neufville et al, 2002). They can be used to model both single and multiple sequential processes. Cumulative arrival and departure profiles, as shown in Figure 5, can be approximated to a discrete process with arrivals and departures measured for sequences of time intervals i of size τ.. In this case, the queue at the end of time interval i, Q i, is Q i = A i - D i (2) where A i and D i are the cumulative number of passengers that have arrived at the process and the number of departures at interval i, respectively. Passengers Arrivals Q (t ) 1 µ Departures t Time Figure 5. Cumulative arrivals and departures

10 180 A. G. de Barros and D. D. Tomber Passenger arrivals at interval i will depend on the flight schedule with flight departure times and passenger loads and the passenger earliness of arrival (EOA) profiles. A common simplifying assumption made in airport passenger flow analysis is to use one single EOA profile throughout the day. Profiles are commonly assumed to vary with type of airline and type of flight e.g. domestic or international but variation in time is not modeled. However, variation in the EOA profile does occur within the day. Passengers flying on early morning flights tend to arrive at the airport much closer to flight departure time than those departing on mid- and late-day flights. Figure 6 demonstrates this trend with data collected at Seattle/Tacoma International Airport. The use of accurate EOA profiles is critical to the accuracy of the analysis, whatever method is used. As part of this research, the EOA profile for flights departing at interval j are modeled as the proportions p j,k where k = 1, 2, m represents the number of intervals in advance to flight departure. For instance, with 5-minute intervals (τ = 5) where the first interval represents the period between 0:00 and 0:05, p 90,12 = 8% means that 8% of passengers departing between 7:25 and 7:30 the 90 th interval in the day will arrive between 6:25 and 6:30 the 12 th interval counted backwards from 7:25-7:30. Note that having one profile for each interval may be quite cumbersome to model, as the number of time intervals in the model may be quite large. It is preferable then to define profiles for each relevant period of the day. PDF Before 7:00 7:01-10:00 After 10: Time to departure (minutes) 60 0 Figure 6. Temporal variation of earliness-of-arrival proflies

11 Quantitative Analysis 181 In this work, the flight loads and departure times are modeled as numbers of scheduled passenger departures at times j, denoted S j. This will correspond to the sum of all flight loads scheduled to depart during time period j. Using this notation, the cumulative number of passengers reporting at the check-in counter at the end of time interval i, A i, is i + m A = + i Ai S j p j, i j j= i 1 (3) D i, the cumulative number of departures from the check-in counters at the end of time interval i, is calculated as D i = min(a i, D i-1 + µ τ) (4) Equation 2 can then be used to calculate the queue at the end of interval i. An estimate of the maximum waiting time experienced by the last passenger in queue can be obtained by dividing Q i by the processing rate µ. Figure 7 shows an example of a Microsoft Excel spreadsheet used to calculate queue lengths and waiting times every five minutes (τ = 5) at a security checkpoint Simulation models Discrete-event simulation models have greatly evolved from the ones described by Mumayiz (1991) and Tosic (1992). They have become very popular among airport terminal planners and are now widely used to support the design of new and expanded passenger and baggage facilities. Figure 8 shows an example of a simulation of a security checkpoint. Simulation models have the ability to model processes ranging from simple to very complex. They are also easy to communicate to decisionmakers, including the capability of producing schematic animations. These characteristics have made simulation a favorite among airport engineers. However, the amount of data and computer modeling necessary to build highly detailed models cannot be overlooked. Spreadsheet models may be easier to build and faster to produce results if a low level of detail is required.

12 182 A. G. de Barros and D. D. Tomber Figure 7. A spreadsheet model of a security checkpoint 4.4. Framework for quantitative analytical planning and design of baggage systems Tomber & de Barros (2004) researched industry best practices in baggage system planning and design and developed a framework outlining a rational process for baggage system planning and design. As part of this research, this framework was refined and is shown in Figure 9. The proposed framework is based on the following steps: 1. Assumptions: a variety of modeling inputs that influence results. 2. Static analysis: using spreadsheet models and process flow and logic control diagrams, identify the problem to be solved with concept layouts. 3. Initial concept layout: physical configuration that meets stated requirements. 4. Dynamic analysis: using discrete-event simulation models. 5. Refined concept layout: final layout of physical configuration based on dynamic analysis. 6. Emulation: enables software controls to manage a detailed simulation model of an automatic baggage handling system.

13 Quantitative Analysis 183 Figure 8. Simulation of a security checkpoint 5. Case Study: Seattle-Tacoma International Airport Seattle-Tacoma (Sea-Tac) International Airport is located in the Northwestern United States. The airport is an important northwest gateway, having handled 27 million passengers in In the peak hour, 6,400 passengers depart on airplanes from SeaTac. The airport is managed and operated by the Port of Seattle.

14 184 A. G. de Barros and D. D. Tomber Assumptions Static Analysis Spreadsheet Models Initial Concept Layouts Planning Horizon Percent of Annual Peaks Method Surged Schedule Airline Cruise Forecast Growth EOA Profiles Load Factor Originating Factor Bags per Passenger Out of gauge bags Check-In Splits Check-In Capacity Process Rates/Protocols Check-In Screening Equipment Staffing Bag Close-out Time Redundancy 1 Total Hourly Bag Peak 2 Split by Feed Line Minute Peaks 3 Initial Equipment Needs Level 1 Level 2 4 Process Flow Diagram Logic/Controls/Equip Loads through System Dynamic Analysis Simulation Refined Concept Layouts Emulation System Performance Time through System Wait Times Queue Lengths Bags Missing Flights Operational Scenarios Refined Equipment Needs Criteria to Refine Layouts 1 Physical Configuration 2 Control Logic High-Level Low-Level Figure 9. Framework for planning and designing baggage systems 5.1. Passenger screening The passenger building has four security checkpoints. The drastic reduction in processing capacity after 9/11 from 600 pax/h to 120 pax/h and the lack of space for expansion required an investigation of the screening process to find ways to increase the processing capacity (Tomber and de Barros, 2004). A number of ideas were investigated, including: Divesting and repacking tables Common use wanders/bag searchers Holding pen and queue before wanding (secondary screening)

15 Quantitative Analysis 185 Shoe x-ray (lack contributes to higher wanding time) Length of exit roller table X-ray delay time vs. continuous operation Policy on number of passenger carry-on items Condensing bins/plastic bags Bag retrieval Walk-through metal detector alarm rate Ratio of walk-through metal detectors to x-ray lanes Dual x-ray machines at each lane Integrated lanes processing selectees and non-selectees The airport has used several spreadsheet models for high-level analysis or security checkpoint processing capacity and staffing levels. Due to the complexity of the screening process and the high level of detail required in the analysis, simulation was the tool chosen to evaluate the aforementioned ideas. A simulation model using the software Arena was built using inputs measured in the field and validated through comparison of the baseline throughput rates with the throughput measured at the security checkpoints, as well as observation of the model behavior using Arena s animation feature. Seven scenarios were simulated: 1. Baseline: 2.8 X-ray items per passenger, no pre-screen tables, one dedicated wander per lane, lane stops if a passenger sets off the arch magnetometer alarm until screened by a wander with a manual metal detector. 2. Use of pre-screen tables: provision of tables prior to the X-ray machine, where passengers can unload their lose belonging into trays in preparation for the screening. 3. Pre-screen tables and reduction in the number of items per passenger: in this scenario, it is assumed that this average will be reduced to 1.8 items per passenger. The objective of this scenario is to test the effect of policies to encourage or force people to bring fewer carry-on items. 4. Pre-screen tables and common-use secondary metal detector screeners (wanders): the baseline scenario considers the use of one dedicated wander per lane. In this scenario, wanders will move between lanes as necessary, i.e., if a passenger sets off the arch

16 186 A. G. de Barros and D. D. Tomber magnetometer alarm, he/she will be screened by the first wander available. 5. Pre-screen tables and holding pens for secondary screening: passengers who set off the alarm at the arch magnetometer are held in a fenced area until a wander is available. The passenger flow through the arch magnetometer is not interrupted. 6. Pre-screen tables and staff schedules: staffing levels vary according to the peak periods. 7. Scenarios 4, 5 and 6 combined. Figure 8 shows a snapshot of the model animation. Table 1 shows the scenarios simulated using a flight schedule for the year 2000 the highest demand level before implementation of new security procedures - and the respective results in terms of queue length and waiting time. The results of the analysis prove that it is possible to drastically reduce waiting times with changes in the screening procedures. The simple addition of pre-screen tables helped reduce the waiting time by two thirds. The use of common-use wanders and holding pens can further reduce the line. Table 1. Options to increase security checkpoint performance Scenario Max queue time (minutes) Max queue length (passengers) 1- Baseline Pre-screen tables Pre-screen tables and less X-ray item per passenger 4- Pre-screen tables and common-use wanders 5- Pre-screen tables and holding pens 6- Pre-screen tables and staff schedules 7- Scenarios 4, 5, 6 combined

17 Quantitative Analysis 187 By far, the most effective measure would be to reduce the number of X-ray items by an average of one item per passenger. Clearly, the X-ray is a significant bottleneck in the screening process. In response to these results, SeaTac has begun experimenting with a novel procedure to reduce the number of X-ray items. The procedure consists in handling large plastic bags to passengers at the end of the line. Passengers can then place all their small and mid-sized belongings in the bag and place the bag in the X-ray. Pre-screening divesting time is greatly reduced, and recollection of the belongings post-screening is also sped up Baggage screening Sea-Tac is primarily an origin and destination airport, which places heavy loads on its baggage systems. The airport is served by over 25 airlines and prior to 9/11 each airline had a separate system that it used exclusively. After 9/11, with the need to add expensive baggage screening equipment, systems were redesigned for shared use by several airlines to minimize initial capital costs, as well as ongoing costs for operation and maintenance. The 25 exclusive use baggage systems were consolidated into 6 shared used baggage systems. Shared use systems also offered the added advantage of reducing equipment requirements due to a reduced demand load since airlines peaked at different times. In addition to cost considerations, there was a limitation on the number of TSA screeners allocated to Sea-Tac due to the US Congressional mandate that fixed the number of screeners nationwide at 45,000. Several factors thus incentivized Sea-Tac, TSA, and airline partners to optimize capital and operational efficiencies by creating a small number of shared use systems in lieu of a larger number of exclusive use systems. These constraints required an investigation of the baggage screening system to optimize capacity (Tomber & de Barros, 2004). A number of ideas were investigated, including: Explosives detection screening (EDS) equipment allocation logic (waterfall versus roundrobin). Merge control logic and correcting conveyor merge point problems. Window reservation logic for baggage input lines at check-in counter. Balancing loads between main feed lines. Overflow and recirculation logic. System redundancy and resilience.

18 188 A. G. de Barros and D. D. Tomber Increased buffering capabilities in conveyor system to lower impact of stops and diebacks. Decoupling check-in counter allocation and chute allocation by load sharing. Avoiding blending suspect, clean, and overflow bags. Identification and prioritization of time sensitive bags. Benefit of fully automated versus manual screening systems. Increased flexibility in system by implementation of small sorter loops or similar multiplexer constructs. Implementation of redundancy in manual encoding stations and SCADA (supervisory control and data acquisition) system. Tracking late bags and automatic chute allocation closures relative to scheduled time of departure. Reduction in the number of locations a bag can be sorted to in make-up. Optimization of labor in make-up cart staging, recycling, and downstream station sortation. Figure 10 shows a snapshot of the model animation for a baggage system at Sea-Tac, using a flight schedule for the year The original design shown in Figure 10 had two separate lines feeding into three EDS machines, with two line-dedicated machines and one common use machine. With that design, only two machines could be used by any one line. A plow merge was available to divert bags from one line to the other when one of the dedicated machines is down. Figure 10. Simulation of a baggage system

19 Quantitative Analysis 189 An initial analysis was done using the spreadsheet model technique described earlier in this paper. Table 2 shows the EOA profiles used in the analysis. Figure 11 shows the resulting baggage flows every 15 minutes on each line. It can be seen that the flows on Line 2 are significantly higher than on Line 1. The peak flow of 725 bags/hour on Line 2 far exceeds the combined capacity of the two EDS machines that are accessible from Line 2. This showed the need to have both lines feeding into all three EDS machines. A simulation model was then built using Arena and run to quantify the gains that could be obtained with such change in the design. Figure 10 illustrates the system as modeled. The actual model is not shown in this paper due to its complexity. Table 3 summarizes the results of the three scenarios tested: 1- the system operating as originally designed with dedicated machines; 2- the new operation with both lines feeding into all machines, using the plow merge in the original design to dynamically move bags between the two lines; 3- same as 2, but with a high-speed diverter replacing the plow merge. Table 2. Earliness of arrival profile for the baggage flow analysis Time to STD (minutes) Interval (k) Domestic (p 1k ) International (p 2k ) % 0% % 0% % 1% % 4% % 13% % 21% % 23% % 19% % 11% % 5% % 2% % 1% % 0% % 0% % 0% % 0%

20 190 A. G. de Barros and D. D. Tomber Line 1 Line 2 Bags / hour :15 1:15 2:15 3:15 4:15 5:15 6:15 7:15 8:15 9:15 10:15 11:15 12:15 13:15 14:15 Figure 11. Baggage flows every 15 minutes Table 3. Options to increase baggage system performance Scenario Die back to ticket counter Number of bags missing flights 1- Baseline, plow merge stationary 2- Plow merger operational as low speed diverter 3- High speed diverter in lieu of plow merge 15:15 16:15 17:15 18:15 19:15 20:15 21:15 22:15 23:15 Time in system (minutes) Yes No 1 20 No 0 18 The results showed that replacement of a plow merge in Scenarios 1 and 3 with a high speed diverter in Scenario 3 that can dynamically balance loads is clearly the best. The stationary plow merge in Scenario 1 actually resulted in catastrophic system breakdown, or gridlock. In this catastrophic condition many bags did not reach their intended flight, a very costly situation for airlines. Scenarios 1 and 2, which utilize a plow merge, suffered from several drawbacks. The plow merge is not designed

21 Quantitative Analysis 191 to move very frequently, and consequently a single point of failure for the system in the event of a breakdown. The simulation determined that the equipment selected to balance loads needed to make over 1,000 movements each day. The high speed diverters were able to meet this requirement by dynamically balancing loads to 3 EDS screening machines. Simulation provided a distinct advantage over static spreadsheet analysis by quantifying the performance of changes to both physical configuration as well as high-level operational software control logic. Operational software controls are as important to system performance as physical configuration. Simulation also demonstrated several things that were not possible to ascertain through static spreadsheet analysis, such as: number of bags missing flights, time in system, and whether bag die back extended to the ticket counter. 6. Conclusions The heightened security in air transportation has greatly reduced the processing capacity at pre-boarding security checkpoints and baggage handling. This has created the need to investigate changes in the passenger and baggage screening procedures to improve the process and reduce the need for expensive equipment, personnel and building space. The choice of the appropriate tool for the analysis will depend on the level of detail required. Analytical models can be used for quick analyses of queuing systems. These models can help decision-makers and designers to quickly determine the design parameters in terms of number of processors and queuing space requirements. Spreadsheet models may be used to model systems with sequential processors with simple referring rules. Simulation has the ability to model more complex systems and, with the recent advances in computer software and hardware, can also produce excellent results within a reasonable timeframe. The Sea-Tac case study has shown that it is possible to greatly improve the passenger and baggage screening processes with the adoption of simple procedures. The X-ray screening of passenger carryons has proved to be the main bottleneck in the security checkpoint system. Policies and methods to reduce the number of carry-on items to be screened in the X-ray machine have proved to be the most efficient and important measures.

22 192 A. G. de Barros and D. D. Tomber 7. Acknowledgements The original limited length version of this paper was published as de Barros (2005). The authors thank the Hong Kong Society for Transportation Studies for giving permission to publish this extended version. This research was supported in part by the Natural Sciences and Engineering Research Council of Canada. 8. References Bandara, S. & S.C. Wirasinghe (1990) Airport gate position estimation for minimum total cost approximate closed form solution. Transportation Research B 24, Brunetta, L., Righi, L. and Andreatta, G. (1999) An operations research model for the evaluation of an airport terminal: SLAM (simple landside aggregate model). Journal of Air Transport Management 5, De Barros, A.G. (2005) New Security Procedures at Airports and their Impact on the Planning and Operation of Passenger Terminals. In Transportation and the Economy Proceedings of the 10 th HKSTS Conference, eds. W.H.K. Lam and J. Yan, p Hong Kong Society for Transportation Studies, Hong Kong. De Neufville, R., de Barros, A.G. and Belin,S. (2002) Optimal configurations of airport passenger buildings for travelers. Journal of Transportation Engineering 121, Federal Aviation Administration (2003) Committee recommendations on FY2002 budget. Research, Engineering and Development Advisory Committee. Washington DC, USA. Jane s Airport Review (2000) Suicide terrorism: a global threat, r001020_1_n.shtml, accessed on 3/4/2004. Mumayiz, S.A. (1991) Overview of airport terminal simulation models. Transportation Research Record 1273, Newell, G.F. (1982) Applications of queuing theory. 2 nd edition, Chapman and Hall, New York, USA. Tomber, D.D. & de Barros, A.G. (2004) Optimisation of security checkpoint and baggage screening at Sea-Tac International Airport.

23 Quantitative Analysis 193 Passenger Terminal Expo th International Conference for Terminal Design, Geneva, Switzerland. Tosic, V. (1992) A review of airport passenger terminal operations analysis and modeling. Transportation Research A 26, 3-26.