Journal of Air Transport Management Volume 8 Issue 4 2002 (Doi 10.1016 - s0969-6997 (02) 00003-0) Graham Francis Ian Humphreys Jackie Fry - The Benchmarking of Airport Performance
Cynthia Barnhart Peter Belobaba Amedeo R. Odoni Massachusetts Institute of Technology, Cambridge, Massachusetts 02139 cbarnhart@mit.edu belobaba@mit.edu arodoni@mit.edu T his paper presents an overview of several important areas of operations research appli- cations in the air transport industry. Specic areas covered are: the various stages of aircraft and crew schedule planning; revenue management, including overbooking and leg- based and network-based seat inventory management; and the planning and operations of aviation infrastructure (airports and air trafc management). For each of these areas, the paper provides a historical perspective on OR contributions, as well as a brief summary of the state of the art. It also identies some of the main challenges for future research. 1. Introduction During the one hundred years since the rst ight of Orville and Wilbur Wright, the air transport indus- try has grown into a major sector of the global econ- omy. Even more importantly, it has become essential to developing and maintaining cultural and economic links among countries and peoples. The airlines alone generated more than $300 billion in revenues in 2002, a lean year, and carried about 1.6 billion passengers, a number expected to grow at an annual rate of 4%5% over the next 20 years according to most forecasts. According to the industry air transport provides 28 million direct, indirect, and induced jobs world- wide and carries over 40% of the world trade of goods, by value (Collaborative Forum 2003). After spending roughly its rst 40 years trying to get off the ground, literally at times, the air trans- port industry has grown by leaps and bounds dur- ing the last 60, especially since the advent of the jet age in the late 1950s. Throughout that second period, operations research (OR) has played a critical role in helping the airline industry and its infrastructure sus- tain high growth rates and make the transition from a novelty that catered to an elite clientele to a service industry for the masses. More than 100 airlines and air transport associations are currently represented in AGIFORS, the Airline Group of Operational Research Societies, which has been active since 1961. Indeed, it is difcult to think of any single sector, other than perhaps military operations, with which operations research has been linked more closely. One of the rea- sons is that airline operations and, more generally, the air transport environment provide natural con- texts for the application of OR techniques and models. A second is that the airline industry has consistently been a leader in the use of information technology and has relied heavily on the intensive use of com- puters over the years. The objective of this paper is to present a histor- ical perspective on the contributions of operations research to the air transport industry, as well as to offer an assessment of some of the challenges that will be confronted next. Any reasonably thorough cover- age of this subject would probably require an entire issue of this journal because the number of OR papers published on air transport easily exceeds 1,000 over the last 50 years. In view of the severe constraints on its length, the scope of the paper will instead be conned to a selected subset of air transport-related topics, where operations research has made some Transportation Science 2003 INFORMS Vol. 37, No. 4, November 2003, pp. 368391 0041-1655/03/3704/0368 1526-5447 electronic ISSN BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry of its most signicant contributions to date. Exam- ples of important topics that are either not covered at all or are touched on peripherally include: avia- tion safety and security, airline eet planning, airline stafng, airline maintenance planning, aircraft load- ing, and decision support tools for the management of airport operations (e.g., gate assignments). More- over, the specic topics and contributions that are highlighted are presented in nonquantitative terms and largely reect the authors own interests. In addi- tion to the bibliographic references associated with these contributions, other survey papers, which pro- vide additional details and references, are cited when- ever possible. Section 2 of the paper deals with the classical problems of scheduling, routing, and crew assign- ment in the airline industry. This is a context that is perfectly suited to the use of large-scale, discrete optimization approaches and, indeed, has motivated several methodological and computational develop- ments in this vibrant area of OR over the years. Sec- tion 3 covers airline revenue management, includ- ing overbooking, ight leg yield management and network revenue maximization. Through a combina- tion of stochastic and optimization models, OR work in this area has generated signicant additional rev- enues for the airlines ever since the late 1980s. More- over, revenue management continues to be a eld in which airlines are vying intensively for competitive advantage. Section 4 surveys selected applications of OR to the study, planning, and design of the two major pieces of aviation infrastructure, the airports system and the air trafc management (ATM) system. Historically the emphasis here has been on stochas- tic models, as the questions addressed have focused on capacity, delays, and safety under conditions in which the probabilistic characteristics of the input parameters play a dominant role. However, optimiza- tion models, both deterministic and stochastic, have found use in the intensive recent research on air traf- c ow management, a topic also reviewed briey in 4. Finally, 5 summarizes the main conclusions regarding the fundamental challenges faced by future research. 2. Aircraft and Crew Schedule Planning Schedule planning involves designing future aircraft and crew schedules to maximize airline protabil- ity. This problem poses daunting challenges because it is characterized by numerous complexities, includ- ing a network of ights, differing aircraft types, gate, airport slot and air trafc control restrictions, noise curfews, maintenance requirements, crew work rules, and competitive, dynamic environments in which passenger demands are uncertain and pricing strate- gies are complex. Not surprisingly, no single opti- mization model has been solved, or even formulated, to address this complex design task in its entirety. The problems unmanageable size and complexity has resulted in the decomposition of the overall problem into a set of subproblems, often dened as follows: (1) Schedule design: Dening which markets to serve and with what frequency, and how to schedule ights to meet these frequencies. (2) Fleet assignment: Specifying what size aircraft to assign to each ight. (3) Aircraft maintenance routing: Determining how to route aircraft to ensure satisfaction of maintenance requirements. (4) Crew scheduling: Selecting which crews to assign to each ight to minimize crew costs. Suboptimal, yet feasible aircraft and crew plans are constructed by solving the subproblems in order, con- straining the solutions to subsequent problems based on the solutions to preceding problems. Although smaller and simpler than the overall problem, these subproblems are still large-scale and rich in complex- ity. In fact, OR theoreticians and practitioners have been developing models and algorithms to solve them for decades and, in so doing, have had signicant suc- cesses and impacts. In the late 1960s and early 1970s, United Airlines, American Airlines, British Airways, and Air France, recognizing the competitive advantage that decision technologies could provide, formed OR groups. These groups grew rapidly, developing decision support tools for a variety of airline applications, including aircraft and crew schedule planning, and in some cases began offering their services to other airlines Transportation Science/Vol. 37, No. 4, November 2003 369 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry as well. According to one executive, the schedule planning system developed at American Airlines and Sabre has generated over $500 million in incremen- tal prots annually (Cook 2000). Most major airlines around the world have formed similar groups by now. Their name and size, as well as their placement in the corporate structure, vary considerably from air- line to airline. Many consulting companies are also offering products and services in this general area. Other testimonials on the impact of optimization in the airline industry can be found in Yu et al. (2003), Butchers et al. (2001), Wiper et al. (1994), Smith et al. (1992), and Patty et al. (1991). Beginning in the 1950s and 1960s, early research on airline schedule planning focused on approaches to assign aircraft to routes (Ferguson and Dantzig 1956a) and crews to trips (Arabeyre et al. 1969). Since then, researchers have continued to work on these prob- lems, rening and expanding the models to better rep- resent the actual problem faced, or developing more sophisticated algorithms to improve the quality of the solutions generated. In the following sections, we provide an overview of progress, highlighting some of the major accom- plishments and describing some of the remaining challenges. 2.1. Schedule Design The ight schedule, specifying the ight legs to be own and the departure time of each ight leg, largely denes the competitive position of an air- line and is thus a key determinant of airline prof- itability. Designing a prot maximizing ight sched- ule, however, is extremely complex. It affects and is affected by essentially all aircraft and crew schedul- ing decisions of the airline, and competing airlines as well. No single model has captured all these inter- dependencies, and even if such a model were for- mulated, it surely would be intractable. Moreover, its input data requirements are impractical, requiring, for example, accurate estimates of itinerary-specic pas- senger demands, spill costs, and recapture rates. Notwithstanding this complexity, ight schedule design and variants of the problem have been of interest to researchers for many years, with Simp- son (1966), Chan (1972), Soumis et al. (1980), and Etschmaier and Mathaisel (1984) describing early work. Nonetheless, because of the inability of opti- mization models to adequately capture the scope of the design problem, the typical airline practice today is to build ight schedules manually, with limited optimization. With recent research advances, how- ever, this trend is reversing and optimization is begin- ning to play a role. Success has been achieved by dening a simplied design problem involving only incremental changes to existing ight schedules. Berge (1994), Marsten et al. (1996), and Lohatepanont and Barnhart (2001) develop models and algorithms that select from a subset of candidate ights legs, those that will be added to or removed from a given (often existing) ight schedule. Their approaches are incremental in that the changes from one published ight sched- ule to the next are limited. The reported impacts, however, are signicant. Lohatepanont and Barnhart solve problems at one major airline that contain about 800 potential ight legs, 65,000 itineraries, and 165 aircraft, resulting in formulations with 30,00060,000 rows and 50,00065,000 columns, and solution times ranging from 12 hours to more than 3 days. They report potential improvements in aircraft utilization and signicant increases in revenue, with an esti- mated impact exceeding $200 million at that airline. A nonincremental, clean-slate approach to airline schedule design is described in Armacost et al. (2002). They present models and algorithms to generate (near-) optimal ight network designs for express par- cel delivery, and report savings of about 7% in oper- ating costs and potential reductions of 10% in the required eet size. Another schedule design application involving a charter airline is described in Erdmann et al. (1999). By exploiting the special characteristics of the prob- lem, they are able to achieve near-optimal solutions in minutes. Schedule generation, with its important strategic and nancial implications, represents an important area for future research, one rich in opportunity and challenge. The successes to date are just rst steps in addressing the myriad of questions surrounding schedule design. Future research is needed to capture the critical interactions among the various resources of the airline, its competitors, and airports. 370 Transportation Science/Vol. 37, No. 4, November 2003 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry 2.2. Fleet Assignment With the ight schedule determined, the eet assign- ment problem is to nd the cost-minimizing assign- ment of aircraft types to legs in the ight network. Fleeting costs are comprised of: (1) Operating costs: Specied for each ight leg aircraft type pair, representing the cost of ying that ight leg with that aircraft type. (2) Spill costs: Measuring the revenue lost when passenger demand for a ight leg exceeds the assigned aircrafts seating capacity. One of the earliest research efforts on airline scheduling was that of Ferguson and Dantzig (1956a). They developed a linear programming model to allo- cate aircraft to roundtrip routes to minimize operat- ing plus spill costs. In a follow-up paper (Ferguson and Dantzig 1956a), they expanded their approach to handle uncertain demands. These rst models, while applied to simplied ight networks and tiny prob- lems by todays standards, identied key problem attributes and launched decades of research on eet assignment. As a result, some 30 years later, researchers devel- oped the capabilities to solve eeting problems rep- resentative, both in complexity and size, of those truly faced by airlines. Abara (1989) and Hane et al. (1995) formulated the eet assignment problem as a multicommodity network ow problem with side constraints. The underlying network (depicted in Figure 1) has: (1) Nodes: Representing the times and locations of ight leg departures and arrivals. (2) Flight arcs and ground arcs: Each ight arc corre- sponds to a ight leg with its arrival time adjusted to include the minimum amount of time required on the ground for disembarking and embarking pas- sengers, unloading and loading baggage, and refuel- ing. Ground arcs represent idle aircraft on the ground between ights. The objective is to ow commodities, that is, air- craft types, through the network feasibly and with minimum cost. Side constraints enforce the following requirements and restrictions: (1) Cover: Each ight leg is assigned to exactly one aircraft type. airport A airport B count line flight arc ground arc Figure 1 A Timeline Network Involving Two Airports (2) Aircraft count: Only available aircraft are assigned to the network. (3) Balance: Each aircraft type is assigned to the same number of ight legs arriving at a station as departing that station. Each possible assignment of an aircraft type to a ight leg is represented by a binary variable, with value 1 if the aircraft type is assigned to that ight leg, and 0 otherwise. The resulting formulation for the eet assignment problem of a major airline contains about 20,000 rows and 30,000 columns and can be solved typically within minutes. 2.2.1. Impacts and Challenges. Multicommodity ow-based eet assignment models are widely used by the industry and are credited with achieving signicant cost savings, measuring in the millions of dollars annually. For example, Rushmeier and Kontogiorgis (1997) report realized savings of at least $15 million annually at USAir. Using eet assignment models, Wiper et al. (1994) report annual savings of $100 million at Delta Airlines, and Abara (1989) reports a 1.4% improvement in operating margins at American Airlines. Beyond providing economic benets, research on eet assignment problems has led to advanced tech- niques for solving general linear programs. For exam- ple, the node consolidation idea introduced in Hane et al. (1995), which reduced formulation size by more than 40% for the problems of one major airline, has been generalized and incorporated into commercial solvers, allowing more efcient solution of large-scale optimization problems. Transportation Science/Vol. 37, No. 4, November 2003 371 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry These impressive results notwithstanding, there remain several critical challenges in eet assign- ment. Many of these challenges stem from modeling assumptions that include: (1) Many eet assignment models assume that the ight schedules repeat daily, even though most air- lines operate different schedules on the weekend. (2) Most eeting models assume ight leg demand is known and does not vary by day of week, but his- torical data show that day-to-day demand variations are present. (3) Flying times and ground times are typically assumed to be deterministic in eet assignment mod- els; however, congestion on the ground and in the air, weather conditions, and new security practices pro- duce large variations in ight and ground times. (4) Most eet assignment models assume that the number of spilled passengers, and their associated spill costs, can be computed at a ight-leg level. In fact, passenger demand, spill, and the revenue asso- ciated with each passenger are itinerary specic, not ight-leg specic. As a result, it is possible to estimate leg-specic spill costs only approximately. To improve on leg-based models, researchers intro- duceditinerary- or origin-destination-(O-D) basedeet assignment approaches (Jacobs et al. 1999; Barnhart et al. 2002a, b). These approaches consider passen- ger fares, demand, spill, and recapture to be itinerary, not ight-leg, specic. Inaccuracies resulting from the allocation of fares and demands to ight legs are thus not introduced into these enhanced eeting approaches. Using their O-D-based eet assignment approach, Jacobs et al. (1999) report annual improve- ments of 0.54%0.77% in revenue compared to those obtained with a ight-leg-based model. Similarly, in a case study involving a major U.S. airline, Barn- hart et al. (2002a) report that itinerary-based eet assignment improves on leg-based eeting signi- cantly, with estimated savings ranging from 30 mil- lion to over 100 million dollars annually. In a rst step at integrating eet assignment and schedule design decisions, Rexing et al. (2000) simul- taneously assign an aircraft type to each ight leg and select each ight legs departure time, allowing retimings of 520 minutes from the current schedule. Departure retiming allows additional aircraft assign- ments that can lead to better matches between ight leg demand and assigned capacity. The result, accord- ing to Rexing et al. (2000), is reduced operating costs and improved revenue capture, with savings for one major airline of $20$50 million annually. 2.3. Aircraft Maintenance Routing With schedule design and eet assignment decisions made, the ight network decomposes into subnet- works, each one associated with aircraft of a single type. The assignment of individual aircraft to ight legs in a subnetwork occurs in the aircraft mainte- nance routing step. The goal is to determine routings, or rotations, for each aircraft in a eet. A routing is a sequence of ight legs, with the destination of one ight leg the same as the origin of the next leg in the sequence. A rotation is a routing that starts and ends at the same location. Each aircrafts rotation vis- its maintenance stations at regular intervals. More details on the maintenance routing problem are con- tained in Feo and Bard (1989), Gopalan and Talluri (1998), and Clarke et al. (1996b). For restricted mainte- nance routings of three or four days, Gopalan and Tal- luri (1998) and Talluri (1998) describe graph-theoretic approaches to maintenance routing. In general, the aircraft maintenance routing prob- lem can be modeled as a network circulation problem with side constraints. The decision variables corre- spond to sequences (strings) of ight legs, with each sequence beginning and ending at maintenance sta- tions and satisfying the rules governing the maximum time between maintenance. If a string is included in the solution, a single aircraft ies each ight in the sequence and then undergoes maintenance. Side con- straints include cover constraints and count constraints. Cover constraints ensure that each ight leg is con- tained in exactly one selected string, and count con- straints limit the number of assigned aircraft to the number available. Additional details are provided in Barnhart et al. (1998a). 2.3.1. Impacts and Challenges. Solving the eet assignment problem rst and then the resulting aircraft routing problems can lead to violations of aircraft maintenance requirements. To guaran- tee feasible solutions, particularly in low-frequency, 372 Transportation Science/Vol. 37, No. 4, November 2003 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry point-to-point networks, researchers have integrated eet assignment and aircraft routing models, as in Desaulniers et al. (1997) and Barnhart et al. (1998a). This has the effect of substantially expanding the air- craft routing model to include multiple eet types, instead of a single aircraft type. The number of con- straints for large airlines is often less than a few thou- sand, but there are many billions of possible variables. For hub-and-spoke networks, however, the increased size and complexity of this integrated model is rarely warranted. Instead, feasible solutions are typically generated using a slightly modied sequential solu- tion approach in which an altered eeting model is rst solved and then the routing model is solved. The modied eet assignment model contains pseudo- maintenance constraints to ensure that sufcient num- bers of aircraft of each type are located at mainte- nance stations periodically. In hub-and-spoke networks with banks containing many aircraft together on the ground at the same time, these maintenance con- straints are often sufcient to ensure that the resulting eet assignment has associated feasible maintenance routings. In addition to the possibility of generating infea- sible solutions, a disadvantage of the sequential solution approach to aircraft and crew planning is that aircraft routing solutions limit possible crew scheduling opportunities, potentially causing crew costs to increase signicantly. The linkage between aircraft routing and crew scheduling occurs because a crewmember can connect between two ight legs separated by less than the minimum required connec- tion time only if the same aircraft is assigned to both legs. To account for this, Klabjan et al. (2002) swap the order of the problems and solve the crew pair- ing problem before the maintenance routing problem. This approach has the advantage of generating opti- mal crew solutions, but it does not ensure that for the optimized crew solution there is a correspond- ing maintenance-feasible solution. To achieve crew optimality and maintenance feasibility, Cordeau et al. (2000) and Cohn and Barnhart (2003) integrate the basic maintenance routing and crew pairing models. 2.4. Crew Scheduling Crew scheduling problems with numerous, com- plex rules, and well-dened costs are particularly amenable to optimization. With so many possible decisions, it is difcult to nd feasiblelet alone optimalsolutions manually. Moreover, crews repre- sent the airlines second highest operating cost after fuel, so even slight improvements in their utilization can translate into signicant savings. For these reasons, airline crew scheduling has gar- nered considerable attention, with research spanning decades. Arabeyre et al. (1969) published a survey of early research activities on the topic. At that time, because of large problem size and lack of advanced techniques and sufcient computing capa- bilities, heuristics were employed to nd improved crew solutions. Since then, researchers have continued to work on the crew scheduling problem, seeking more efcient solution approaches, expanding models to capture more of the intricacies of the crew scheduling prob- lem, and integrating the crew problem with other airline scheduling problems. Recent detailed descrip- tions of the airline crew scheduling problem are included in the survey papers by Desaulniers et al. (1998), Clarke and Smith (2000) and Barnhart et al. (2003). Even today, the crew scheduling problem is typ- ically broken into two sequentially solved subprob- lems, the crew pairing problem and the crew assignment problem: (1) The crew pairing problem. The problem generates minimum-cost, multiple-day work schedules, called pairings. Regulatory agencies and collective bargain- ing agreements specify the many work rules that dene how ight legs can be combined to create fea- sible schedules. Work-rule restrictions include limits on the maximum number of hours worked in a day, the minimum number of hours of rest between work periods, and the maximum time the crew may be away from their home base. Even with these limi- tations, the number of feasible pairings measures in the billions for major U.S. airlines. The cost structure of pairings adds further complexity, with cost typ- ically represented as a nonlinear function of ying time, total elapsed work time, and total time away from base. (2) The crew assignment problem. This problem com- bines these pairings into equitable and efcient Transportation Science/Vol. 37, No. 4, November 2003 373 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry month-long crew schedules, called bidlines or rosters, assigning them to individual crewmembers. With ros- tering, schedules are constructed for and assigned to specic individuals, taking into account their par- ticular needs and requests. With bidline generation (a practice commonly used in the United States), a generic set of schedules is constructed and individ- ual employees reveal their relative preferences for these schedules through a bidding process. The spe- cic assignment of schedules to employees is based on seniority, with the more senior crewmembers most often assigned their preferred schedules. More details on the bidline and rostering problems can be found in Kohl and Karisch (2003), Cappanera and Gallo (2001), Caprara et al. (1998), Christou et al. (1999), Dawid et al. (2001), Day and Ryan (1997), and Gamache et al. (1998). The crew pairing, bidline, and rostering problems can all be cast as set partitioning problems. While sim- ple in form, the set partitioning model is powerful in this context. To illustrate, consider the crew pairing problem. The crew pairing problem has one binary deci- sion variable for each possible pairing, and its objec- tive is to minimize the cost of the selected pairings such that for each ight, exactly one pairing con- taining that ight is chosen. By dening variables as sequences of ight legs, only feasible pairings are considered and explicit formulation of complicated work rules is unnecessary. Moreover, nonlinear pair- ing costs can be computed ofine and captured as a single value. The crew pairing model, is then a set par- titioning problem, with a linear objective function and binary decision variables. The one drawback to this modeling approach is that there are potentially many billions of possible crew pairings, especially in hub- and-spoke ight networks with numerous aircraft at hubs at the same time, allowing many possible crew connections. 2.4.1. Solving Crew Scheduling Problems. Be- cause of the immense size of airline crew scheduling problems, i.e., thousands of constraints and billions of variables, researchers initially focused on heuristic methods to achieve solutions. Many of these heuris- tics generate solutions by considering only a relatively small number of pairings. Examples of this strategy can be found in Arabeyre et al. (1969), Anbil et al. (1991), Gershkoff (1989), and Hoffman and Padberg (1993). While producing improved solutions compared to those generated earlier, these heuristics still have a drawback; namely, the quality of their solutions can- not be quantied relative to an optimal solution. With advances in optimization theory and comput- ing, however, researchers have built enhanced crew scheduling capabilities and designed optimization- based approaches capable of generating provably optimal or near-optimal crew pairing solutions. To generate crew solutions with known optimal- ity bounds, even when variable enumeration is not practical, many researchers have turned to branch-and- price. Branch-and-price, surveyed in Barnhart et al. (1998b), is a smart enumeration technique in which linear programming bounds are generated at each node of a branch-and-bound tree using column gen- eration. Column generation allows very large LPs to be solved without explicitly enumerating all of the variables, or columns. Rather than directly solving the master problem (the problem with all possible columns), a restricted master problem containing only a subset of the original columns is solved. The dual solution to the restricted master problem is used to identify columns with negative reduced cost. These columns are added to the restricted master problem; the restricted master problem is then resolved, and the process is repeated until no negative reduced cost columns can be found. At that point, the algorithm terminates with an optimal solution to the master problem. In the case of the crew pairing problem, negative reduced cost columns are identied without explic- itly considering each pairing. Instead, all columns are implicitly considered by solving a pricing prob- lem, often formulated as a multilabel, shortest path problem in which some paths correspond to pairings. By exploiting dominance, shortest pathsand hence minimum reduced cost pairingsare identied with- out examining all paths, or pairings. For a detailed exposition of multilabel, shortest path problems, see Desrochers and Soumis (1988). A particular challenge in branch-and-price algo- rithms is that a standard branching rule based on 374 Transportation Science/Vol. 37, No. 4, November 2003 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry variable dichotomy is often difcult to implement. Instead, a branch-on-follow-ons branching rule is typi- cally used (Ryan and Foster 1981). This rule is based on pairs of ights, with one ight immediately follow- ing the other. One branching decision requires sched- ules containing either ight to contain both ights and the other decision disallows this. For applications of this branching strategy to airline crew schedul- ing, see Lavoie et al. (1988), Desrosiers et al. (1991), Ryan (1992), Gamache and Soumis (1993), Barnhart et al. (1994), Vance et al. (1997), and Desaulniers et al. (1998). Even with the effective branch-on-follow-ons rule, the number of branching decisions to prove optimal- ity is typically excessive for crew-scheduling prob- lems. To remedy this, researchers have embedded heuristics within the branch-and-price optimization process. An example is the variable-xing approach described in Marsten (1994) in which variables with fractional values close to one are sequentially xed to one. Marsten shows that by reducing the num- ber of LPs solved in generating integer solutions, improved solutions can be generated with signi- cantly less computing time and memory. Klabjan et al. (2001) integrate the heuristic consideration of sub- sets of columns with an optimization-based pricing approach. They produce improved solutions, com- pared to those achieved with branch-and-price, by initially developing hundreds of millions of columns and applying random selection and pricing tech- niques to prune the number of columns ultimately considered. 2.4.2. Impacts and Challenges. Most major air- lines use optimization tools to partially or fully generate their crew schedules, with signicant eco- nomic impacts. Already in the 1960s, airlines were solving crew pairing problems to reduce costs and automate their burdensome manual planning process (Arabeyre et al. 1969). With advances in optimization and computing, crew pairing solvers became increas- ingly sophisticated and applicable. Using advanced heuristic techniques, Anbil et al. (1991) achieved cost savings at American Airlines of $20 million dollars annually, representing an increase in crew utilization of 1.5%. Motivated by the potential for even further cost savings, researchers focused on the development of optimization-based approaches in the decades span- ning the 1980s and 1990s. Barnhart et al. (2003) reported that using a branch-and-price approach, one major U.S. airline was able to produce near-optimal crew solutions costing $50 million per year less than the solutions generated by a state-of-the-art heuristic. In addition to these economic benets, crew scheduling optimization can be a useful tool in con- tract negotiations, helping airlines to quantify the impacts of proposed changes in cost structures, bene- ts, and work rules. The economic impact of each pro- posed change can be evaluated by changing selected inputs to the crew scheduling model, and rerunning the optimization procedure. While signicant, the effects of optimization on crew scheduling are nonetheless limited by the sequential schedule planning process. With the ight schedule, eet assignment, and aircraft routing deci- sions xed, the range of crew scheduling possibilities is limited. To mitigate the myopic effects of sequential solutions, researchers have developed extended mod- els that incorporate crew considerations into some of the other subproblems solved. For example, Clarke et al. (1996a) and Barnhart et al. (1998c) extend eet assignment models to account for some of the down- stream effects on crews. The eet assignment model in the sequential solution process is then replaced with one of these extended models, resulting in an increase in eet assignment costs that is offset by a greater reduction in crew costs. 2.5. Ongoing and Future Challenges Notwithstanding the substantial progress made in solving aircraft and crew scheduling problems in recent decades, signicant work remains. Research opportunities, beyond those identied for the sched- ule design, eet assignment, aircraft routing, and crew scheduling problems described above, include: Schedule planning: (1) Integrating decisions involv- ing schedule design and aircraft and crew routing and scheduling; (2) expanding schedule planning models to include pricing and revenue management decisions; (3) assessing the systemwide cost and ser- vice impacts of paradigm shifts in airline scheduling, Transportation Science/Vol. 37, No. 4, November 2003 375 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry such as depeaking ight schedules by spreading out aircraft arrival and departure banks; and (4) devel- oping tools, such as the stochastic model and airline operations simulator described in Rosenberger et al. (2002), to evaluate airline plans and recovery policies in a stochastic operating environment. Robust planning: Optimized solutions are rarely exe- cuted as planned. Crew sickness, mechanical failures, and adverse weather result in necessary changes to the plan, often leading to signicantly increased costs. Moreover, a nely tuned, optimized solution achieves increased utilization through the removal of slack, providing crews with less time to connect between ights and aircraft with less time on the ground between ying. Less slack time, although economical in theory, can translate in practice into less robust- ness and increased costs. To address these issues, researchers have begun to investigate a different plan- ning paradigm, one that considers unplanned, dis- ruptive events and attempts to minimize realized, and not planned, costs. Current research, such as that of Ageeva (2000), Rosenberger et al. (2001a), Schae- fer et al. (2001), and Chebalov and Klabjan (2001), attempts to achieve robustness by isolating causes of disruption and/or downstream effects. In Lan (2003), robust aircraft routes are generated to place slack judi- ciously, that is, where it is needed to minimize the disruptive effects of delays on passengers. Operations recovery: Given a plan and one or more disruptions, the goal of operations recovery is to return the airline to its plan quickly (often requir- ing decisions within minutes) and cost effectively. Detailed descriptions of recovery models and algo- rithms are included in Jarrah et al. (1993), Clarke (1997), Thengvall et al. (2000), Rosenberger et al. (2001b), Stojkovic et al. (2002), Bratu (2003), and Yu et al. (2003). Although these problems typically involve multiple airports and many of the airlines resources, most of the work to date focuses on recov- ering a particular resource, either aircraft, crews, or passengers. For example, in their Edelman Award- winning work, Yu et al. (2003) report that Conti- nental Airlines saved approximately $40 million in 2001 using their crew recovery system. The challenge moving forward is to enhance current capabilities by considering integrated passenger, crew, and aircraft recovery decisions, while rapidly determining low- cost solutions. 3. Airline Revenue Management Even with an optimized eet assignment and sched- ule of operations, some ight departures will have empty seats while others will experience more pas- senger demand than capacity. In an effort to better match the demand for each ight with its capacity and to increase total revenues, airlines practice differ- ential pricing by offering a variety of fare products at different price levels for the same ight. Revenue man- agement is the practice of determining the number of seats on each ight to be made available at each fare level, limiting low-fare seats and protecting seats for later-booking, higher-fare passengers. Given that the operating costs of a scheduled ight departure are in large part xed in the very short run, the goal of rev- enue management is to ll each ight with the maxi- mum possible revenue to maximize operating prot. This section provides a brief review of the role of operations research in the development of airline rev- enue management (RM) models, with an emphasis on the works that have most inuenced the state of the practice in the industry. A much more comprehensive survey of OR literature dealing with revenue manage- ment and related problems can be found in McGill and van Ryzin (1999). In addition, Weatherford and Bodily (1992) developed a categorization of perish- able asset revenue management problems, of which the airline revenue management problem is the best- known example. This review begins with an introduction to the functions of typical airline RM systems, followed by descriptions of the types of OR models employed to perform three of the principal techniques of revenue managementoverbooking, fare class mix, and O-D control. Therefore, the focus of this discussion is on the seat inventory control component of airline rev- enue maximization, as virtually all airline RM systems assume that the fare structure is determined exoge- nously by a separate airline pricing function. 3.1. The Evolution of Airline RM Systems The sheer size and complexity of the airline seat inventory control problem led to the development 376 Transportation Science/Vol. 37, No. 4, November 2003 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry of computerized RM systems. A medium-sized air- line might operate 1,000 ight legs per day, using 10 booking (fare) classes in its reservations system, and accepting bookings up to 330 days prior to each departure. At any point in time, this airlines seat inventory includes over three million booking limits, which can change with each booking that is accepted. Airline RM systems have evolved in both their computer database and mathematical modeling capa- bilities over the past 20 years. The rst RM systems, developed in the early 1980s, were designed to collect and store data extracts from computer reservations systems (CRS). By the mid-1980s, several RM sys- tems offered additional monitoring capabilities that allowed actual ight bookings to be tracked dynam- ically, relative to an expected or threshold book- ing curve for the ight. By the late 1980s, the more advanced airlines began to implement RM systems that could perform forecasting and optimization by booking class for each future ight leg departure, in addition to having all of the database and booking monitoring capabilities of previous systems. It was into this third generation of RM systems that OR models, some of which had been developed a decade or more earlier, began to be integrated. The major components of a typical third-generation RM system are illustrated in Figure 2. Historical book- ing data for the same ight leg and day of the week are combined with actual booking information for each future ight departure to generate a forecast of total demand by booking class for that departure. REVENUE DATA ACTUAL BOOKINGS HISTORICAL BOOKING DATA OVERBOOKING MODEL NO-SHOW DATA FORECASTING MODEL RECOMMENDED BOOKING LIMITS OPTIMIZATION MODEL Figure 2 Third-Generation Airline RM System These forecasts, together with estimates of the rev- enue value of each booking class, are then fed into an optimization model that calculates the recommended booking limits for the ight departure in question. At the same time, the demand forecasts are fed into an overbooking model, which also makes use of his- torical information about passenger no-show rates to calculate an optimal overbooking level for the future ight departure. Both the booking class limits and overbooking levels calculated by the mathematical models are then presented as recommendations to the RM analyst. The demand forecasts and booking limits are reviewed by the RM systems at regular intervals during the ight booking process, as often as daily in some cases. Should unexpected booking activity occur, the system reforecasts demand and reoptimizes its booking limit recommendations. A substantial pro- portion of the revenue gain attributable to fare mix optimization comes from this dynamic revision of booking limits. Most large and medium-sized airlines throughout the world have implemented third-generation RM systems, and the benets of such systems have been well documented. Effective use of techniques for over- booking and fare class mix alone have been estimated to generate revenue increases of as much as 4%6% compared to situations in which no seat inventory control tools were applied (Belobaba 1992b, Smith et al. 1992). 3.2. Overbooking Models Airlines have been accepting reservations in excess of aircraft capacity for more than two decades, in an effort to reduce the revenue losses associated with no- shows. With the development of RM systems, over- booking was incorporated into the seat inventory control functions of these systems. The objective of the ight overbooking component of revenue man- agement is to determine the maximum number of bookings to accept for any given future ight depar- ture, trading off the risks and costs of denied board- ings against the potential revenue loss from unsold or spoiled seats. Operations research work on the airline overbook- ing problem can be traced to the 1960s, well before Transportation Science/Vol. 37, No. 4, November 2003 377 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry overbooking became integrated into RM systems (and before RM systems were even required). Notable early published works include those of Simon (1968) and Vickrey (1972), as well as articles by Rothstein (1971, 1985). Rothsteins 1985 article is a survey of previous OR literature dealing with the airline overbooking problem, and includes a discussion of the customer service impacts of inaccurate overbooking and the role of government in regulating denied boarding penalties. Statistical overbooking models typically represent no-show rates as Gaussian random variables. The objective is to nd the maximum authorized number of bookings (or AU) that will keep denied board- ings below some airline-specied target level with a desired level of condence. These models provide airline managers with exibility in determining their own overbooking policy, for example, by increas- ing denied boarding tolerance or reducing statistical condence. An extension of the statistical overbooking ap- proach is the cost-based overbooking model, which explicitly accounts for the actual costs associated with denied boardings and with empty seats (spoilage). The objective is to nd the optimal overbooking level that minimizes the total combined costs of denied boardings and spoilage by performing a search over a reasonable range of AU values. The cost-based over- booking model is the current state of the practice at many airlines. However, this approach represents a static formulation of the overbooking problem, in that the dynamics of passenger bookings, cancella- tions, and no-shows are not explicitly accounted for in determining an overbooking level. The OR literature contains many additional works on the airline overbooking problem, some of which propose dynamic programming (DP) formulations. Rothsteins (1968) Ph.D. thesis was the rst to describe such a DP approach, while Alstrup et al. (1986) extended the DP formulation to a two-class, joint overbooking and fare class mix problem. More recently, Chatwin (1996) as well as Feng and Xiao (2001) have proposed DP-based approaches that allow for the incorporation of time-dependent no-show and cancellation rates associated with multiple fare classes. In practice, few airlines have implemented such complex DP formulations because of the difcul- ties of providing adequate and accurate inputs in the form of booking and cancellation rates by the time- period before departure. The economic motivation for airline overbooking is substantial. In the United States, domestic airline no-show rates average 15%25% of nal predeparture bookings. Given that most airlines struggle to attain a 5% operating margin (revenues over costs), the loss of 15%25% of potential revenues on fully booked ights (which would occur without overbooking) can repre- sent a major negative impact on prots. As part of a revenue management system, effective overbooking has been shown to generate as much revenue gain as optimal fare class seat allocation, described below. 3.3. Fare Class Mix The second major technique of airline revenue management is the determination of the revenue- maximizing mix of seats available to each booking (fare) class on each future ight-leg departure. Vir- tually all airline RM systems were developed with the capability to optimize fare class mix as their pri- mary objective. As introduced earlier, RM systems forecast the expected demand for each fare class on each future ight-leg departure by applying statisti- cal models to historical booking data for the same fare class on previous departures of the same ight. The forecasting of expected booking demands for future ight departures has been addressed in many OR papers, including Littlewood (1972), LHeureux (1986), Lee (1990), and Curry (1994). These demand forecasts are then used as inputs to a seat allocation optimization model, which determines booking limits to be applied to each of the booking classes on the ight departure in question. The vast majority of airline reservations systems now have inventory structures based on serial nesting of booking classes, as shown in Figure 3. Seats are not allocated to partitioned classes, but are instead protected for higher fare classes and nested booking limits are applied to the lower fare classes. All available seats in the shared inventory are made available to the highest booking class, in the (unlikely) event that the entire capacity of the aircraft can be lled with demand for the highest-priced fare product. This ensures that the 378 Transportation Science/Vol. 37, No. 4, November 2003 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry TOTAL AVAILABLE SEATS BL1 BL2 BL3 } } Protected for Class 1 from2, 3, and 4 Protected for 1+2 from 3 and 4 } Protected for 1+2+3 from 4 BL4 Figure 3 Nested Booking Limits and Class Protection Levels airline would never turn down a high fare request, as long as there are any seats still available for the ight. OR work on seat allocation problems for two fare classes (full-fare and discount fare) can be traced to Littlewood (1972). Littlewoods (1972) rule for protect- ing full-fare seats was extended to multiple nested fare classes by Belobaba (1987, 1989), who wrote the rst doctoral dissertation on airline pricing and revenue management. The Expected Marginal Seat Revenue (EMSR) approach for setting RM booking limits was then rened to become the EMSRb model (Belob- aba 1992a). These models were the rst to recognize that the optimality conditions for traditional resource (seat) allocation problems did not result in the max- imum expected revenues in a multiple class, nested booking limit environment. Both models are based on heuristic decision rules for nested booking classes and have become the most commonly used seat inventory control models in airline RM systems. Optimal formu- lations of the multiple nested class problem have also been published by Curry (1990), Brumelle et al. (1990), and Wollmer (1992). The general premise of these fare class mix mod- els is as follows. Given the forecast demand for each booking class, expressed in terms of a mean and stan- dard deviation, along with its associated average fare, the expected marginal revenue of each incremental seat on a ight leg can be determined. It is equal to the average fare of the booking class under consider- ation multiplied by the probability that demand will materialize for that incremental seat. The optimal pro- tection level for a higher-class seat is equal to the num- ber of seats with an expected marginal seat revenue greater than or equal to the average fare in the next lower class. Because higher fare classes have access to unused lower class seats in a nested booking class inventory structure, the problem is to nd seat pro- tection levels for higher classes, and booking limits on lower classes. Simulations and actual airline experi- ence have indicated that use of these decision mod- els for establishing the fare class mix of seats on each ight departure can increase total airline revenues by 2%4% (Belobaba 1989). Recent research on the single-leg fare class mix problem has been rather limited, as both researchers and practitioners have focused on network opti- mization techniques for airline revenue management instead, as described below. The most recent pub- lished works involve extensions that include the application of DP methods and the joint optimization of fare class booking limits and overbooking levels (for example, Zhao and Zheng 2001), as well as the joint optimization of the airline pricing and fare class mix problems, as described in 3.5. 3.4. Network Revenue Management: Origin-Destination Control Network RM (or O-D Control) represents a major step beyond the fare class mix capabilities of most third- generation RM systems, and is currently being pur- sued by the largest and most advanced airlines in the world. As its name implies, O-D control gives the air- line the capability to manage its seat inventory by the revenue value of the passengers origin-destination itinerary on the airlines network, not simply accord- ing to the fare class requested on a single ight leg. Because a leg-based RM system cannot distinguish among itineraries in the same fare class, optimizing fare class mix on each ight leg individually will not ensure that total network revenues are being maxi- mized. This is especially true for the large connecting hub networks operated by many airlines, in which a substantial proportion of passenger itineraries involve multiple ight legs and a connection at the hub. As a rst step in the implementation of O-D control strategies, American Airlines in the 1980s developed Transportation Science/Vol. 37, No. 4, November 2003 379 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry what has come to be known as virtual nesting (Smith et al. 1992), an assignment process that maps each itinerary/fare type to a hidden or virtual value class within the airlines own reservations system. For an itinerary/fare assigned to a given virtual class, the seat availability on a given ight leg is estab- lished by the booking limit for that virtual class. Ini- tial implementations of virtual nesting were based on total itinerary fare values, giving preference to longer- haul connecting passengers with higher total fares. The shortcoming of this greedy approach was that it did not address the need under certain circum- stances to give preference to two local passengers instead of taking the connecting passenger. OR models have been instrumental in the sub- sequent development of O-D control strategies that allowed airlines to manage seat inventories on the basis of the network revenue value of each itinerary/fare combination. Network revenue value can be dened as the total itinerary fare (ticket value) minus the revenue displacement that might occur on connecting ight legs if the passengers request for a multiple-leg itinerary is accepted. For example, the network revenue value of an $800 total fare on the rst leg of a connecting itinerary must be reduced by $300 if acceptance of this passenger would result in displacement of $300 in total network revenue on the second ight leg. The displacement cost associated with the last (lowest valued) available seat on a given ight leg is its marginal value to the total network revenuethe amount by which the total network rev- enue would decrease if we were to remove (or sell) one seat on the ight leg. If the ight leg consistently has empty seats, the marginal network value of its last seat would be zero. In the more realistic situa- tion where the ight leg has a signicant probability of being full and it carries both local and connecting passengers, determining the impact on total network revenues of removing a seat can require substantially more complicated analysis, including network opti- mization models. Various network optimization and heuristic algo- rithms have been applied to the problem of deter- mining the network revenue value contribution of each O-D itinerary and fare product (or ODIF). As described in Belobaba (1998), several large airlines make use of leg-based EMSR values to approximate the network revenue value associated with a given connecting itinerary/fare combination. Because these approaches do not make use of any formal network optimization models, they are referred to as leg- based heuristic models for estimating network rev- enue displacement. The majority of published works devoted to the O-D seat inventory control problem have proposed mathematical programming approaches to determine optimal seat allocations for every ODIF over a net- work of connecting ight legs (see, for example, Glover et al. 1982, Dror et al. 1988, and Curry 1990). Williamsons (1992) doctoral thesis provides a compre- hensive review of the literature relevant to the applica- tion of mathematical programming and network ow models to the O-D seat inventory control problem. More recently, there have been new works propos- ing more sophisticated network models. For exam- ple, Bratu (1998) developed a probabilistic network convergence algorithm for estimating the marginal network value of the last available seat on each leg in an airline network. Talluri and van Ryzin (1999a) propose a randomized linear programming method, while stochastic dynamic programming is also being pursued as an alternative algorithm for determining the marginal network value of a range of remaining seats on each ight leg to account for changes to the marginal value as bookings are accepted. Given an estimate of down-line revenue displace- ment it is possible to map ODIFs to virtual classes based on their estimated network revenue value. Dis- placement adjusted virtual nesting (DAVN) increases availability to connecting passengers, while adjust- ment for down-line displacement of revenues ensures that two local passengers (with a higher total revenue) will receive preference when two connecting ights are expected to be heavily booked. The estimated marginal network revenue of the lowest-valued available seat on a ight leg can also be used in a much simpler inventory control mech- anism based on bid price controls. At the time of an ODIF request for seat availability, the ODIF fare is evaluated against the sum of the leg bid prices over the itinerary being requested. As was the case for esti- mates of revenue displacement associated with con- necting passengers, leg bid prices can similarly be 380 Transportation Science/Vol. 37, No. 4, November 2003 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry calculated either with network optimization tools or leg-based heuristics such as EMSR approaches. Talluri and van Ryzin (1999b) have done much work on network bid price control and identify the condi- tions under which this approach provides revenue optimality, while de Boer et al. (2002) compare the performance of deterministic and stochastic network formulations for O-D control. Until recently, relatively few airlines had imple- mented network optimization models for dynamic calculation of displacement costs and/or bid prices for O-D control. Because most reservations systems and, in turn, third-generation RM systems were developed on the basis of leg/fare class data, most airlines did not have access to the detailed historical ODIF booking data required by network optimiza- tion models. Use of large-scale network optimization models also raised technical and computational issues related to the solution times and frequency of reop- timization. However, with the development of airline databases designed to capture detailed ODIF histori- cal data, along with advances in both solution algo- rithms and computational speeds, network revenue management has been implemented by over a dozen airlines in different parts of the world. The benets of leg-based revenue management and incremental benet of O-D controls over leg- based fare class controls have been estimated by several researchers through simulation. For example, Williamson (1992) developed a network revenue man- agement simulation approach that allowed different schemes for optimization and control of seat inven- tories to be tested. An even more realistic approach to simulating the impacts of different RM schemes in a full-scale, competitive airline network environment is that of the passenger origin-destination simulator (PODS). Developed originally by researchers at Boe- ing (Hopperstad 1997), PODS has been enhanced to realistically simulate large networks in which compet- ing airlines generate RM forecasts and set seat inven- tory controls based on historical (i.e., previously simulated) data. At the same time, the simulated pas- sengers in PODS choose among alternative airlines, fares, restrictions, schedules, and seat inventory avail- ability as established by each airlines own RM sys- tem. PODS has been used to simulate the competitive impacts of RM (Belobaba and Wilson 1997), as well as the benets of improved forecasting models and the impacts of RM on airline alliances. The simulations cited here along with others per- formed by academics and airlines have provided con- sistent estimates of the potential for revenue gains of 1%2% from advanced network revenue management methods, above and beyond the 4%6% gains real- ized from conventional leg-based fare class control. The potential to realize even 1% in additional revenue through network RM is substantial enough that many of the worlds largest airlines have implemented or are in the process of developing their O-D control capabilities. For a large airline with annual revenues of $5 to $10 billion or more, successful implementa- tion of a network RM system can lead to total revenue increases of $50 to $100 million per year. 3.5. Future Challenges Development of OR models for the next genera- tion of airline revenue management is currently an extremely popular topic among academics and prac- titioners alike. The most obvious next steps in the further enhancement of airline revenue management systems is to integrate the pricing and seat inventory control decisions currently being made with different decision support tools and, at many airlines, in dif- ferent parts of the organization. Clearly, the ability to relax the traditional RM assumption that fare struc- tures are given and xed has the potential to further increase the revenue gains of RM. Joint pricing and inventory optimization requires the incorporation of passenger choice and demand elasticity models, and promising OR work in this direction has been pub- lished by Weatherford (1997), Gallego and van Ryzin (1997), and Cote et al. (2003), among others. In a recent Ph.D. dissertation de Boer (2003) examines this problem in a network context and presents a variety of other modeling advances and insights. Looking ahead, it is apparent that information about the utilization of seat inventories and the response of passenger demand to different pricing strategies can and should provide useful feedback to eet assignment and even scheduling of airline ight departure times. The integration of airline pric- ing and seat inventory decisions with those of the Transportation Science/Vol. 37, No. 4, November 2003 381 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry scheduling and eet assignment functions is therefore considered to be the ultimate challenge for airline operations research. Development of more respon- sive and even dynamic decision support systems that take into account expected passenger choice behav- ior, as well as the expected response of competitors in terms of price, schedule, and capacity in determin- ing a prot-maximizing strategy for the airline, will only become more important with growing competi- tive pressures in the airline industry. 4. Applications to Aviation Infrastructure The infrastructure of the global aviation system con- sists of two principal elements, airports and air trafc management (ATM) systems. Airports can be further subdivided into airside facilities (runways, taxiways, aprons, aircraft stands) and landside facilities (pas- senger and cargo buildings, curbside), while ATM systems are now viewed as being comprised of a tacti- cal subsystemair trafc control (ATC)and a strate- gic oneair trafc ow management (ATFM). The design, development, and operation of all these facil- ities and systems has attracted extensive interest on the part of operations researchers, usually in response to ongoing developments in the eld. For example, much of the fundamental work on airside capacity was performed during the 1960s and early 1970s, the time when it was rst realized that runways con- stituted an important bottleneck of the air transport system. Overall, the body of work on aviation infras- tructure has led to insights and models that have proved of critical importance in practice and have, in some cases, been adopted by airport and ATM ser- vice providers on a global scale. Because of space limitations, this section briey reviews OR applica- tions in airport airside operations and air trafc ow managementonly two of the four major areas iden- tied above. Surveys of OR models for the analy- sis of passenger terminal operations can be found in Tosic (1992) and de Neufville and Odoni (2003). Of the many OR-related topics addressed by research on air trafc control, the widely investigated subject of detecting and resolving potential conicts between airborne aircraft is reviewed well in Kuchar and Yang (2000). Various other analytical and simulation mod- els on several different aspects of ATC are covered in Odoni et al. (1997). 4.1. Airside Operations The runway complexes of major airports are among the scarcest resources of todays international air transport system and, barring a drastic change in the landing and takeoff requirements of commercial air- craft, will continue to be so in the foreseeable future. New runways are very expensive to build, require great expanses of land, and most importantly have environmental and other impacts that necessitate long and complicated approval processes with uncertain outcomes. It is not surprising therefore that one of the most mature areas of transportation science deals with the modeling of runway operations and, more generally, airside operations. The products of this work include both analytical (mathematical) mod- els and simulation tools. 4.1.1. Analytical Capacity and Delay Models. Analytical models preceded viable simulation tools by about 20 years. In a landmark paper, Blumstein (1959) dened the capacity of a runway as the expected number of movements (landings and takeoffs) that can be performed per unit of timetypically one hourin the presence of continuous demand and without violating air trafc control separation require- ments. He also presented a model for computing the capacity of single runways used for arrivals only, for departures only, and for strings of arrivals fol- lowed by strings of departures. Subsequent general- izations included the possibility of inserting depar- tures between successive arrivals, possibly by increas- ing (stretching) the separation between arrivals (Hockaday and Kanafani 1972) and the treatment of some of the parameters of Blumsteins (1959) mod- els as random variables, instead of constants (Odoni 1972). Extensions to cases involving two or more simulta- neously operating runways were also developed at an early stagesee, e.g., Swedish (1981). The complex- ity of multirunway models depends greatly on the extent to which operations on different runways inter- 382 Transportation Science/Vol. 37, No. 4, November 2003 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry act because of air trafc control separation require- ments. For example, in the case of a pair of intersect- ing runways, the location of the intersection relative to the points where takeoffs are initiated or where landing aircraft touch down greatly affects the com- bined capacity of the two runways. Similarly, in the case of two parallel runways, the capacity depends on the distance between the centerlines of the runways. Approximate capacity analyses for airports with two parallel or intersecting runways are quite straightfor- ward. Multirunway analytical capacity models also provide good approximate estimates of true capacity in cases involving three or more active runways, as long as the runway congurations can be decom- posed into semi-independent parts, each consisting of one or two runways. Such models have proved extremely valuable in airport planning, as well as in assessing the impacts of proposed procedural or tech- nological changes on airport capacity. Another topic of intensive study has been the esti- mation, through the use of queueing models, of the delays caused by the lack of sufcient runway capac- ity. This is a problem that poses a serious challenge to operations researchers: The closed-form results devel- oped in the voluminous literature of classical steady- state queueing theory are largely nonapplicableat least when it comes to the really interesting cases. The reason is that airport queues are, in general, strongly nonstationary. The demand rates and, in changing weather conditions, the service rates at most major airports vary strongly over the course of a typical day. Moreover, the demand rates may exceed capacity (j >1), possibly for extended periods of time, most often when weather conditions are less than optimal. This has motivated the development of numer- ical approaches to the problem of computing air- port delays analytically. In another landmark paper, Koopman (1972) arguedand showed through exam- ples drawn from New Yorks Kennedy and LaGuardia Airports, at the time among the worlds busiestthat the queueing behavior of an airport with | runway equivalents (i.e., | nearly independent servers) can be bounded by the characteristics of the M(t)/M(t)/k and the M(t)/D(t)/k queueing models, each providing worst-case and best-case estimates, respectively. Note that this allows for dynamic changes in the ser- vice rates, as well as in the demand rates. Extending the work of Koopman (1972), the A(|),| | (|),| system was proposed by Kivestu (1976) as a model that could be used to directly com- pute approximate queueing statistics for airports rather than separately solving the M(t)/M(t)/k and M(t)/D(t)/k models and then somehow interpolating their results. (Note that negative exponential service times (A) and constant service times (D) are sim- ply special cases of the Erlang (| | ) family, with | =1 and | = , respectively.) Kivestu (1976) noted that | should be determined from the relationship ||S],u S =
|, where ||S] and u
S denote the expected value and the standard deviation of the service times and can be estimated from eld data. He also devel- oped a powerful numerical approximation scheme that computes the (time varying) state probabilities for the A(|),| | (|),| system efciently. Malone (1995) has demonstrated the accuracy and practicality of Kivestus (1976) approach and developed additional efcient approximation methods, well suited to the analysis of dynamic aireld queues. Fan and Odoni (2002) provide a description of the application of Kivestus (1976) model to a study of the gridlock con- ditions that prevailed at LaGuardia Airport in 2000 and early 2001. Additional (numerical) analytical models for com- puting airport delays have been developed over the last few years. Peterson et al. (1995) and Daniel (1995) describe two different models for computing delays at hub airports, which are characterized by sharp banks or waves of arrivals and departures. Hansen (2002) has used a deterministic model, based on the notion of cumulative diagrams, to compute delay externalities at Los Angeles International Air- port. Finally, Long et al. (1999) and Malone (1995) present two dynamic queueing network models and their application to the study of congestion in the National Airspace System. Ingolfsson et al. (2002) offer a comprehensive survey and comparison of sev- eral alternative approaches to the analysis of nonsta- tionary queueing systems. Many of the best features of some of the analyt- ical capacity and delay models just described have been integrated recently in a number of new software Transportation Science/Vol. 37, No. 4, November 2003 383 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry packages (Long et al. 1999, Stamatopoulos et al. 2003, EUROCONTROL 2001) that perform both capacity and delay analyses and, in the instance of the last two references, include models of aprons and aircraft stands. After some necessary enhancements and an adequate amount of testing in a variety of airport environments, packages of this type may provide air- port planners within a few years with an easy-to-use and very fast set of tools for the study of a host of airside issues. 4.1.2. Airside Simulations. General-purpose sim- ulation models of airside operations rst became viable in the early 1980s and have been vested with increasingly sophisticated features since then. Three models currently dominate this eld internationally: SIMMOD, The Airport Machine, and the Total Air- port and Airspace Modeler (TAAM). A report by Odoni et al. (1997) contains detailed reviews (some- what out-of-date by now) of these and several other airport and airspace simulation models and assesses the strengths and weaknesses of each. At their cur- rent state of development (and in the hands of expert users), they can be powerful tools in studying detailed airside design issues, such as guring out the best way to remove an airside bottleneck or estimating the amount by which the capacity of an airport is reduced due to the crossing of active runways by taxiing aircraft. Unfortunately, these models are frequently misused in practice, at great cost to the client organization. This happens when they are applied to the study of macroscopic issues that can have only approxi- mate answers because of the uncertainty inherent in the input data. An example is a question that often confronts airport operators: When will airside delays reach a level that will require a major expansion of an airports capacity (e.g., through the construction of a new runway)? Questions of this type, often requiring a look far into the future, are best answered through the approximate analytical models surveyed earlier, which permit easy exploration of a large number of alternative scenarios and hypotheses. Detailed simu- lation models, by contrast, cannot cope well with the massive uncertainty involved because they require inputs that are difcult to produce (e.g., a detailed schedule of aircraft movements at the airport for a typical day 10 or 15 years hence) and lack credibility under the circumstances. 4.1.3. Optimizing Airside Operations. The air- side models discussed so far are descriptive in nature. Their objective is to help users understand and pre- dict the operational characteristics of the various airside facilities under different operating scenarios. A considerable amount of OR work with an optimiza- tion focus also exists, much of it concerned with the effective use of runway systems. The capacity of a runway is largely determined by the separation requirements specied by the providers of ATM services (e.g., the FAA in the United States). For any pair of consecutive runway opera- tions these requirements depend on the type of air- craft involved. For example, in the United States, when an arriving heavy (H) aircraftdened as one with a maximum takeoff weight (MTOW) greater than 255,000 lbsis immediately followed by an ar- riving small (S) aircraft (MTOW - 41,000 lbs), the required separation between them, at the instant when H is about to touch down on the runway, is 6 nautical miles (10.9 km). This is because heavy aircraft (wide-body jets) may generate severe wake turbulence, which may be hazardous to other aircraft behind it. By contrast, when an aircraft of type S is followed by one of type H, the required separation is 2.5 nautical miles (4.5 km). Note that given a num- ber n of aircraft, all waiting to land on a runway, the problem of determining the sequence of landings, such that the time when the last aircraft lands is min- imized, is a Hamiltonian path problem with n points. However, this is only a static version of a prob- lem which in truth is a dynamic one: Over time the pool of aircraft available to land changes, as some aircraft reach the runway while new aircraft join the arrivals queue. Moreover, minimizing the latest landing time (or maximizing throughput) should not necessarily be the objective of optimal sequencing. Many alternative objective functions, such as mini- mizing the average waiting time per passenger, are just as reasonable. A further complication is that the very idea of sequencing runs counter to the tra- ditional adherence of ATM systems to a rst-come, rst-served (FCFS) discipline. Deviations from FCFS raise concerns among some airside users about the 384 Transportation Science/Vol. 37, No. 4, November 2003 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry possibility of systematic discrimination against cer- tain classes of aircraft operators (e.g., general avia- tion) when it comes to runway access. In a dynamic environment, this may even result in a compromise of safety, if some aircraft are indenitely relegated to the end of the queue as new aircraft show up to land. These observations have led many investigators to study the runway-sequencing problem with the objec- tive of increasing operating efciency while ensur- ing that all airport users are treated equitably. Dear (1976) and Dear and Sherif (1991) developed the con- cept of constrained position shifting (CPS), i.e., of a limit in the number of positions by which an aircraft can deviate from its FCFS position in a queue. For instance, an aircraft in the 16th position in a FCFS queue would have to land in one of the positions 1418 if the specied maximum position shift (MPS) is 2. Through many numerical examples and for several reasonable objective functions, Dear (1976) showed that by setting MPS to a small number, such as two or three, one can obtain most of the benets of an unconstrained optimized system (e.g., 60%80% of the potential improvements). This nding motivated several researchers (e.g., Psaraftis 1980, Venkatakr- ishnan et al. 1992, Bianco et al. 2001) to investigate a number of increasingly complex and realistic ver- sions of the sequencing problem. Two advanced ter- minal airspace automation systems, CTAS and COM- PAS, that have been implemented in the United States and in Germany, respectively, incorporate sequenc- ing algorithms based on CPS (Erzberger 1995). How- ever, this feature of CTAS and of COMPAS has not been activated, primarily because of concerns about a potential increase in controller workload. Gilbo (1993) and Hall (1999) have gone beyond the sequencing of arrivals only by considering how avail- able capacity can best be allocated in a dynamic way between landings and takeoffs to account for the dis- tinct peaking patterns in the arrival and departure streams at airports over the course of a day. Pujet et al. (1999) have further examined the issue of opti- mizing the number of aircraft taxiing out during peri- ods of congestion, based on the empirical observation that departure rates at major airports seem to decrease when the number of active aircraft on the taxiway system exceeds a certain airport-specic threshold. Although still at the theoretical stage, some of these promising ideas will eventually nd their way into practice. 4.2. Air Trafc Flow Management The most advanced OR work on aviation infrastruc- ture to date is undoubtedly associated with air traf- c ow management (ATFM). ATFM took on major importance in the United States and Europe during the 1980s, when rapid trafc growth made it neces- sary to adopt a more strategic perspective on ATM. Rather than addressing congestion through local mea- sures (e.g., by holding arriving aircraft in the airspace near delay-prone airports) the goal of ATFM is to pre- vent local system overloading by dynamically adjust- ing the ows of aircraft on a national or regional basis. It develops ow plans that attempt to dynam- ically match trafc demand with available capacity over longer time horizons, typically extending from 312 hours in the future. The prototypical applica- tion of ATFM is in ground holding, i.e., in intentionally delaying an aircrafts takeoff for a specied amount of time to avoid airborne delays and excessive con- troller workload later on. Other ATFM tactics include rerouting of aircraft and metering (controlling the rate) of trafc ows through specied spatial boundaries in airspace. An important difference in the nature of the ATFM problem in the United States and in Europe should also be noted. In the United States, ATFM is primar- ily driven by airport capacity constraints, whereas in Europe en route airspace acts as the principal bot- tleneck. Europes Central Flow Management Unit, located in Brussels, currently determines (heuristi- cally) ground delays to ensure that no en route sec- tor capacity constraints are violated. This difference may, however, become moot in the near future due to continuing progress in increasing en route airspace capacity in Europe. OR model development related to ATFM can be viewed as going through two distinct stages. The rst stage involved problem denition and develop- ment of large-scale mathematical optimization mod- els of an aggregate scope. Attwool (1977) was the rst to cast ATFM issues in mathematical terms, while Transportation Science/Vol. 37, No. 4, November 2003 385 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry Odonis (1987) detailed description of the single- airport ground holding problem (GHP) as a dynamic and stochastic optimization problem stimulated much of the subsequent work. Important advances in mod- eling and solving the GHP are marked by the stochas- tic programming models of Richetta and Odoni (1993), the extension to a multiairport setting by Vranas et al. (1994), and the inclusion of en route constraints and rerouting options by Bertsimas and Stock (1998). Many other interesting papers on vari- ous aspects of optimizing ATFM and GHP appeared in the 1990s. Good reviews of the literature and of computational results can be found in Andreatta et al. (1993) and Hoffman and Ball (2000). The one common characteristic of the models devel- oped in this rst stage is the implicit assumption of a single decision-making authority attempting to opti- mize a global objective function: The providers of ATFM services (e.g., the FAA in the United States, Eurocontrol in Europe) are responsible for the allo- cation of ground holding delays among individual ights and/or for the rerouting, if necessary, of ights. The objective is to optimize in the aggregate, e.g., by minimizing the overall direct operating costs associ- ated with ground holding and rerouting decisions, summed over all airlines and aircraft. This, however, is an operating philosophy that airlines strongly dis- agree with. They correctly argue that only individ- ual airlines have the information necessary to make decisions on what is best for their own ights. As an obvious example, airline A, faced with a period of delays at a given congested airport, may assign very high priority to the timely arrival of one of its ights, X, because that ight may be carrying many business-class passengers who will be connecting to other ights or because it carries crews for subsequent ights departing from that airport. The assignment of priority to ight X has, in fact, little to do with direct operating costs of aircraft and is based on the busi- ness model of airline A and on information that only A possesses. In response to such airline concerns, as well as to various complaints about the limitations of the ATFM system during the 1980s and 1990s, the FAA has been engaged for the past 10 years in developing the Col- laborative Decision-Making (CDM) Program. After an initial planning period of about ve years, CDM was elded for the rst time in 1998 in connection with the FAAs Ground Delay Programs (GDPs), which go into effect whenever long air trafc delays are anticipated at an airport due to poor weather or other reasons, thus often necessitating ground holding. CDM marks a truly fundamental innovation in the ATM system, possibly the most important one in at least 30 years. The three main elements on which it is based are: (a) a dedicated data communications network (CDM- net), which facilitates the continuous exchange of information between the FAA and the airlines (plus any other CDM participants) about the current and near-future states of the ATM system; (b) the use of a common database and a common set of software tools by all CDM participants; and (c) the partial decentral- ization of decision making. With respect to (c), it is the FAAs responsibility to forecast the capacity that will be available at each part of the ATM system dur- ing the relevant time horizon, as well as to allocate this capacity among the individual airlines and the other ATM system users. And it is the responsibility of each individual airline to decide how it will use its allocated share of capacity at each part of the system. This is a somewhat simplistic description of what, in practice, is a complicated process that employs sev- eral types of distributed decision-making techniques, such as rationing by schedule (RBS) and schedule compressionsee Wambsganss (1996) and Vossen et al. (2003) for details. The driver for the adoption and implementation of the CDM concept was a small, OR-minded team in the FAA, the U.S. Department of Transportation, and especially the Metron Corporation (Chang et al. 2001). The CDM Program has already led to major reduc- tions in delays and missed connections for air travel- ers and to documented savings of hundreds of mil- lions of dollars in airline operating costs. ATFM- related OR research has concurrently shifted away from large-scale, aggregate optimization models and toward real-time decision support tools that assist air trafc managers in the FAA and Airline Opera- tions Centers in taking maximum advantage of the massive, up-to-date information base that CDM has made available. It is important to note, however, that many of the ideas and formulations developed in the 386 Transportation Science/Vol. 37, No. 4, November 2003 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry pre-CDM models can still be adapted to the CDM environment, often with little modication. For exam- ple, one of the most critical problems in the plan- ning of GDPs continues to be the determination of airport acceptance rates (AAR) for several hours into the future and in the presence of uncertainty about airport capacity and air trafc demand. The efcient stochastic integer program developed for this purpose by Ball et al. (2003) can be viewed as a direct descen- dant of the pre-CDM model proposed by Richetta and Odoni (1993). ATFM in the CDM era also provides fertile ground for much future research because the scope of poten- tial OR analysis and modeling has expanded greatly. Examples of some topics, along with occasional recent references, include: identifying (as an airline) ights that should be cancelled or delayed (and by how much) in connection with GDPs, recovering (as an air- line) from irregular operations (cf. 2) resulting from GDPs or other ATFM interventions, ensuring equity of access to airports and ATM resources (Vossen et al. 2003), collaborative routing of aircraft through con- gested airspace (Ball et al. 2002), introducing bartering and possibly market-based mechanisms in the alloca- tion of airport slots (Vossen and Ball 2001, Hall 1999), and developing efcient simulation environments for the testing of alternative ATFM strategies. 5. Conclusion Operations research has been one of the principal con- tributors to the enormous growth that the air trans- port sector has experienced during the past 50 years. In the best tradition of OR, the development of mod- els and of solutions has been motivated by issues and problems encountered in practice and has led, in sev- eral instances, to insights of a general nature and to important methodological advances in the OR eld at large. At this point, OR models and algorithms are diffused throughout the sector and constitute an inte- gral part of the standard practices of airlines, airports, and ATM service providers. In view of the numerous challenges that it cur- rently faces, it is safe to expect a continuing cen- tral role for OR in the air transport sectors future. As indicated in this paper, there are many promis- ing topics for future research in each of the areas examined. At the most fundamental level, and in general terms, the frontiers can be summarized as follows: Relaxing the boundaries between the successive stages of aircraft and crew schedule planning, so that schedule design, eet assignment, aircraft main- tenance routing, and crew scheduling might eventu- ally be performed in an integrated way, rather than solved sequentially as interrelated, but distinct sub- problems. Including pricing decisions in revenue manage- ment, instead of treating fares and fare classes as xed, externally specied inputs. Developing fast decision support tools that increase the safety and efciency of air transport oper- ations by taking advantage of the massive, real-time data ows in an increasingly info-centric aviation infrastructure. References Abara, J. 1989. Applying integer linear programming to the eet assignment problem. Interfaces 19 2028. Ageeva, Y. 2000. Approaches to Incorporating Robustness into Airline Scheduling. Thesis, MIT, Cambridge, MA. Alstrup, J., S. Boas, O. B. G. Madsen, R. Vidal, V. Victor. 1986. Book- ing policy for ights with two types of passengers. Eur. J. Oper. Res. 27 274288. Anbil, R., E. Gelman, B. Patty, R. Tanga. 1991. Recent advances in crew-pairing optimization at American Airlines. Interfaces 21 6274. Andreatta, G., A. R. Odoni, O. Richetta. 1993. Models for the ground-holding problem. L. Bianco, A. R. Odoni, eds. Large- Scale Computation and Information Processing in Air Trafc Con- trol. Springer-Verlag, Berlin, Germany. Arabeyre, J. P., J. Fearnley, F. C. Steiger, W. Teather. 1969. The air- line crew scheduling problem: A survey. Transportation Sci. 3 140163. Armacost, A., C. Barnhart, K. Ware. 2002. Composite variable for- mulations for express shipment service network design. Trans- portation Sci. 36 120. Attwool, V. W. 1977. Some mathematical aspects of air trafc sys- tems. J. Instit. Navigation 30 394411. Ball, M. O., A. Futer, R. Hoffman, J. Sherry. 2002. Rationing schemes for en route air trafc management. CDM paper, http://www.metronaviation.com/cdm/cr/long_term.html. , R. Hoffman, A. Odoni, R. Rifkin. 2003. A stochastic integer program with dual network structure and its application to the ground-holding problem. Oper. Res. 51 167171. Transportation Science/Vol. 37, No. 4, November 2003 387 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry Barnhart, C., A. Farahat, M. Lohatepanont. 2002a. Airline eet assignment with enhanced revenue modeling: An alterna- tive model and solution approach. Working paper, Operations Research Center, MIT, Cambridge, MA. , T. Kniker, M. Lohatepanont. 2002b. Itinerary-based airline eet assignment. Transportation Sci. 36 199217. , E. Johnson, R. Anbil, L. Hatay. 1994. A column genera- tion technique for the long-haul crew assignment problem. T. Ciriano, R. Leachman, eds. Optimization in Industry: Vol- ume II. John Wiley and Sons, U.K., 722. , N. Boland, L. Clarke, E. Johnson, G. Nemhauser, R. Shenoi. 1998a. Flight string models for aircraft eeting and routing. Transportation Sci. 32 208220. , F. Lu, G. Nemhauser, M. Savelsbergh, P. Vance. 1998b. Branch- and-price: Column generation for solving huge integer pro- grams. Oper. Res. 46 316329. , , R. Shenoi. 1998c. Integrated airline scheduling: Opera- tions research in the air industry. G. Yu, ed. International Series in Operations Research and Management Science, Vol. 9. Kluwer Academic Publishers, Norwell, MA, 384403. , A. M. Cohn, E. L. Johnson, D. Klabjan, G. L. Nemhauser, P. H. Vance. 2003. Airline crew scheduling. Randolph W. Hall, ed. Handbook of Transportation Science, 2nd ed. Kluwer Academic Publishers, Norwell, MA. Belobaba, P. P. 1987. Air travel demand and airline seat inven- tory management. Ph.D. dissertation, MIT Flight Transporta- tion Laboratory Report R87-7, Cambridge, MA. . 1989. Application of a probabilistic decision model to airline seat inventory control. Oper. Res. 37 183197. . 1992a. The revenue enhancement potential of airline rev- enue management systems. ASTAIR Proc. Adv. Software Tech. Air Transport, London, U.K. . 1992b. Optimal vs. heuristic methods for nested seat alloca- tion. AGIFORS Reservations Control Study Group Meeting, Brus- sels, Belgium. . 1998. The evolution of airline yield management: Fare class to origin-destination seat inventory control. Handbook Airline Marketing. McGraw-Hill, 285302. , J. L. Wilson. 1997. Impacts of yield management in competi- tive airline markets. J. Air Transport Management 3 310. Berge, M. 1994. Timetable optimization: Formulation, solution approaches, and computational issues. AGIFORS Proc., 341 357. Bertsimas, D., S. Stock. 1998. The air trafc ow management prob- lem with en route capacities. Oper. Res. 46 406422. Bianco L., P. DellOlmo, S. Giordani. 2001. Coordination of trafc ows in the TMA. L. Bianco, P. DellOlmo, A. Odoni, eds. New Concepts and Methods in Air Trafc Management. Springer-Verlag, Berlin, Germany. Blumstein, A. 1959. The landing capacity of a runway. Oper. Res. 7 752763. Bratu, S. 1998. Network value concept in airline revenue manage- ment. Masters thesis, MIT, Cambridge, MA. . 2003. Real time optimization models to recover aircraft schedules and minimize passenger disruptions. Working Paper SB0301, Center for Transportation Logistics, MIT, Cambridge, MA. Brumelle, S. I., J. I. McGill, T. H. Oum, K. Sawaki, M. W. Tretheway. 1990. Allocation of airline seats between stochastically depen- dent demands. Transportation Sci. 24 183192. Butchers, E. R., P. R. Day, A. P. Goldie, S. Miller, J. A. Meyer, D. M. Ryan, A. C. Scott, C. A. Wallace. 2001. Optimized crew schedul- ing at Air New Zealand. Interfaces 31 3056. Cappanera, P., G. Gallo. 2001. On the airline crew rostering prob- lem. Technical Report TR-01-08, Department of Computer Sci- ence, University of Pisa, Italy. Caprara, A., P. Toth, M. Fischetti, D. Vigo. 1998. Modeling and solv- ing the crew rostering problem. Oper. Res. 46 820830. Chan, Y. 1972. Route network improvement in air transportation schedule planning. Technical Report R72-3, MIT Flight Trans- portation Laboratory, Cambridge, MA. Chang, K., K. Howard, R. Oiesen, L. Shisler, M. Tanino, M. C. Wambsganss. 2001. Enhancements to the FAA ground-delay program under collaborative decision making. Interfaces 31 5776. Chatwin, R. E. 1996. Multi-period airline overbooking with multi- ple fare classes. Naval Res. Logist. 43 603612. Chebalov, S., D. Klabjan. 2001. Robust crew scheduling: Move-up crews. INFORMS Annual Conf., Miami, FL. Christou, I. T., A. Zakarian, J. Liu, H. Carter. 1999. A two-phase genetic algorithm for large-scale bidline-generation problems at Delta Air Lines. Interfaces 29 5165. Clarke, L., C. Hane, E. Johnson, G. Nemhauser. 1996a. Maintenance and crew considerations in eet assignment. Transportation Sci. 30 249260. , E. Johnson, G. Nemhauser, Z. Zhu. 1996b. The aircraft rota- tion problem. Ann. Oper. Res. 69 3346. Clarke, M. 1997. Development of heuristic procedures for ight rescheduling in the aftermath of irregular airline operations. Sc.D. dissertation, MIT, Cambridge, MA. , B. Smith. 2000. The impact of operations research on the evo- lution of the airline industry: A review of the airline planning process. Research paper, Sabre Inc., Dallas, TX. Cohn, A., C. Barnhart. 2003. Improving crew scheduling by incor- porating key maintenance routing decisions. Oper. Res. 51 387396. Collaborative Forum of Air Transport Stakeholders. 2003. Fast Facts. Flyer obtainable from Forum members (e.g., International Air Transportation Association, www.iata.org). Cook, T. 2000. Creating competitive advantage using model-driven support systems. MIT Global Airline Indust. Study Distinguished Speaker Sem. Ser., Cambridge, MA. Cordeau, J., G. Stojkociv, F. Soumis, J. Desrosiers. 2000. Benders decomposition for simultaneous aircraft routing and crew scheduling. Technical Report G-2000-37, GERAD, cole Poly- technique de Montral, Quebec, Canada. 388 Transportation Science/Vol. 37, No. 4, November 2003 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry Cote, J. P., P. Marcotte, G. Savard. 2003. A bi-level modelling approach to pricing and fare optimisation in the airline indus- try. J. Revenue Pricing Management 2 2336. Curry, R. E. 1990. Optimum seat allocation with fare classes nested by origins and destinations. Transportation Sci. 24 193204. . 1994. Forecasting for revenue management. Technical brief, Scorecard, Aeronomics, Inc., Atlanta, GA. Daniel, J. I. 1995. Congestion pricing and capacity of large hub air- ports: A bottleneck model with stochastic queues. Econometrica 63 327370. Dawid, H., J. Konig, C. Strauss. 2001. An enhanced rostering model for airline crews. Comput. Oper. Res. 28 671688. Day, P. R., D. M. Ryan. 1997. Flight attendant rostering for short- haul airline operations. Oper. Res. 45 649661. Dear, R. 1976. The dynamic scheduling of aircraft in the near- terminal area. Ph.D. dissertation, Technical Report R76-9, Flight Transportation Laboratory, MIT, Cambridge, MA. , Y. S. Sherif. 1991. An algorithm for computer assisted sequencing and scheduling of terminal area operations. Trans- portation Res. 25A 129139. de Boer, S. V. 2003. Advances in airline revenue management and pricing. Ph.D. dissertation, MIT, Cambridge, MA. , R. Freling, N. Piersma. 2002. Mathematical programming for network revenue management revisited. Eur. J. Oper. Res. 137 7292. de Neufville, R., A. Odoni. 2003. Airport Systems: Planning, Design and Management. McGraw-Hill, New York. Desaulniers, G., J. Desrosiers, M. Gamache, F. Soumis. 1998. Crew scheduling in air transportation. T. Crainic, G. Laporte, eds. Fleet Management and Logistics. Kluwer Academic Publishers, Norwell, MA. , , M. M. Solomon, F. Soumis. 1997. Daily aircraft routing and scheduling. Management Sci. 43 841855. Desrochers, M., F. Soumis. 1988. A generalized permanent labeling algorithm for the shortest path problem with time windows. INFOR 26 191212. Desrosiers, J., Y. Dumas, M. Desrochers, F. Soumis, B. Sanso, P. Trudeau. 1991. A breakthrough in airline crew schedul- ing. Report G-91-11, GERAD, cole Polytechique de Montral, Quebec, Canada. Dror, M., P. Trudeau, S. P. Ladany. 1988. Network models for seat allocation on ights. Transportation Res. 22B 239250. Erdmann, A., A. Nolte, A. Noltemeier, R. Schrader. 1999. Modeling and solving the airline schedule generation problem. Technical Report zpr99-351, ZAIK, University of Cologne, Germany. Erzberger, H. 1995. Design principles and algorithms for automated air trafc management. AGARD Lecture Series 200, Brussels, Belgium, http://www.ctas.arc.nasa.gov/. Etschmaier, M. M., D. F. X. Mathaisel. 1984. Airline scheduling: The state of the art. AGIFORS Sympos., Strasbourg, France. EUROCONTROL. 2001. CAMACA: The commonly agreed method- ology for airside capacity assessment. Brussels, Belgium, http://www.eurocontrol.int/camaca/. Fan, T. P., A. R. Odoni. 2002. A practical perspective on airport demand management. Air Trafc Control Quart. 10 285306. Feng, Y., B. Xiao. 2001. A dynamic seat inventory control model and its optimal policy. Oper. Res. 49 938949. Feo, T. A., J. F. Bard. 1989. Flight scheduling and maintenance base planning. Management Sci. 35 14151432. Ferguson, A. R., G. B. Dantzig. 1956a. The problem of routing air- craft. Aeronautical Engrg. Rev. 14. , . 1956b. The allocation of aircraft to routesAn example of linear programming under uncertain demand. Management Sci. 3 4573. Gallego, G., G. van Ryzin. 1997. A multi-product dynamic pricing problem and its applications to network yield management. Oper. Res. 45 2441. Gamache, M., F. Soumis. 1993. A method for optimally solving the rostering problem. Les Cahier du GERAD, G-90-40, cole des Hautes tudes Commerciales, Montral, Canada. , R. Soumis, D. Villeneuve, J. Desrosiers, E. Gelinas. 1998. The preferential bidding system at Air Canada. Transportation Sci. 32 246255. Gershkoff, I. 1989. Optimizing ight crew schedules. Interfaces 19 2943. Gilbo, E. P. 1993. Airport capacity: Representation, estimation, opti- mization. IEEE Trans. Control Systems Tech. 1 144154. Glover, F., R. Glover, J. Lorenzo, C. McMillan. 1982. The passenger mix problem in the scheduled airlines. Interfaces 12 7379. Gopalan, R., K. Talluri. 1998. The aircraft maintenance routing prob- lem. Oper. Res. 46 260271. Hall, W. 1999. Information ows and dynamic collaborative decision-making architecture: Increasing the efciency of ter- minal area operations. Ph.D. dissertation, Operations Research Center, MIT, Cambridge, MA. Hane, C. A., C. Barnhart, E. L. Johnson, R. E. Marsten, G. L. Nemhauser, G. Sigismondi. 1995. The eet assignment prob- lem: Solving a large-scale integer program. Math. Programming 70 211232. Hansen, M. 2002. Micro-level analysis of airport delay externali- ties using deterministic queuing models: A case study. J. Air Transport Management 8 7387. Hockaday, S. L. M., A. Kanafani. 1972. A methodology for airport capacity analysis. Transportation Res. 8 171180. Hoffman, K. L., M. Padberg. 1993. Solving airline crew-scheduling problems by branch-and-cut. Management Sci. 39 657682. Hoffman, R., M. O. Ball. 2000. A comparison of formulations for the single-airport ground holding problem with banking con- straints. Oper. Res. 48 578590. Hopperstad, C. A. 1997. PODS: Modeling Update. AGIFORS Yield Management Study Group, Montral, Canada (May 1416). Ingolfsson, A., E. Akhmetshina, S. Budge, Y. Li, X. Wu. 2002. A survey and experimental comparison of service level approxi- mation methods for non-stationary A,A,s queueing systems. Working paper, Department of Finance and Management Sci- ence, University of Alberta, Edmonton, Alberta, Canada. Jacobs, T. L., E. L. Johnson, B. C. Smith. 1999. O&D FAM: Incor- porating passenger ows into the eeting process. AGIFORS Sympos., New Orleans, LA. Transportation Science/Vol. 37, No. 4, November 2003 389 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry Jarrah, A., G. Yu, N. Krishnamurthy, A. Rakshit. 1993. A decision support framework for airline ight cancellations and delays. Transportation Sci. 27 266280. Kivestu, P. 1976. Alternative methods of investigating the time- dependent A,G,K queue. Thesis, Department of Aeronautics and Astronautics, MIT, Cambridge, MA. Klabjan, D., E. L. Johnson, G. L. Nemhauser, E. Gelman, S. Ramaswamy. 2001. Solving large airline crew scheduling prob- lems: Random pairing generation and strong branching. Com- putational Optimization and Algorithms 20 7391. , , , , . 2002. Airline crew scheduling with time windows and plane count constraints. Transportation Sci. 36 337348. Kohl, N., S. E. Karisch. 2003. Airline crew rostering: Problem types, modeling, and optimization. Carmen Research and Technology Report CRTR-2001-1, Goteborg, Sweden. Koopman, B. 1972. Air terminal queues under time-dependent con- ditions. Oper. Res. 20 10891114. Kuchar, J. K., L. C. Yang. 2000. A review of conict detection and resolution modeling methods. IEEE Trans. Intelligent Transporta- tion Systems 1 179189. Lan, S. 2003. Planning for robust airline operations: Optimizing aircraft routings and ight departure times to achieve min- imum passenger disruptions. Ph.D. dissertation, MIT, Cam- bridge, MA. Lavoie, S., M. Minoux, E. Odier. 1988. A new approach for crew pairing problems by column generation with an application to air transportation. Eur. J. Oper. Res. 35 4558. Lee, A. O. 1990. Airline reservations forecasting: Probabilistic and statistical models of the booking process. MIT Flight Trans- portation Laboratory Report R90-5, Cambridge, MA. LHeureux, E. 1986. A new twist in forecasting short-term passen- ger pickup. AGIFORS Sympos. Proc., 248261. Littlewood, K. 1972. Forecasting and control of passenger bookings. AGIFORS Sympos. Proc., 95117. Lohatepanont, M., C. Barnhart. 2001. Airline schedule planning: Integrated models and algorithms for schedule design and eet assignment. Transportation Sci. Forthcoming. Long, D., D. Lee, J. Johnson, E. Gaier, P. Kostiuk. 1999. Modeling air trafc management technologies with a queuing network model of the National Airspace System. Report NASA/CR- 1999-208988, NASA Langley Research Center, Hampton, VA. Malone, K. 1995. Dynamic queuing systems: Behavior and approx- imations for individual queues and networks. Ph.D. disserta- tion, Operations Research Center, MIT, Cambridge, MA. Marsten, R. 1994. Crew planning at Delta Airlines. Math. Program- ming Sympos. XV Presentation, Ann Arbor, MI. , R. Subramanian, L. Gibbons. 1996. Junior Analyst Extraordi- naire (JANE): Route development at Delta Airlines. AGIFORS Sympos., Atlanta, GA. McGill, J. I., G. J. van Ryzin. 1999. Revenue management: Research overview and prospects. Transportation Sci. 33 233256. Odoni, A. R. 1972. Efcient operation of runways. A. W. Drake, R. L. Keeney, P. M. Morse, eds. Analysis of Public Systems. MIT Press, Cambridge, MA. . 1987. The ow management problem in air trafc control. A. R. Odoni, G. Szego, eds. Flow Control of Congested Networks. Springer-Verlag, Berlin, Germany. , J. Deyst, E. Feron, R. J. Hansman, K. Khan, J. K. Kuchar, R. Simpson. 1997. Existing and required modeling capa- bilities for evaluating ATM systems and concepts. Interna- tional Center for Air Transportation, MIT, Cambridge, MA, http://web.mit.edu/aeroastro/ www/labs/AATT/aatt.html. Patty, B., E. Gelman, R. Tanga, R. Anbil. 1991. Recent advances in crew-pairing optimization at American Airlines. Interfaces 21 6274. Peterson, M. D., D. Bertsimas, A. Odoni. 1995. Models and algo- rithms for transient queuing congestion at hub airports. Man- agement Sci. 41 12791295. Psaraftis, H. N. 1980. A dynamic programming approach for sequencing groups of identical jobs. Oper. Res. 28 13471359. Pujet, N., B. Delcaire, E. Feron. 1999. Input-output modeling and control of the departure process of congested airports. AIAA Guidance, Navigation Control Conf., Portland, OR. Rexing, B., C. Barnhart, T. Kniker, A. Jarrah, N. Krishnamurthy. 2000. Airline eet assignment with time windows. Transporta- tion Sci. 34 120. Richetta, O., A. R. Odoni. 1993. Solving optimally the static ground- holding policy problem in air trafc control. Transportation Sci. 27 228238. Rosenberger, J., E. Johnson, G. Nemhauser. 2001a. A robust eet assignment model with hub isolation and short cycles. Work- ing paper, Georgia Institute of Technology, Atlanta, GA. , , . 2001b. Rerouting aircraft for airline recovery. Working paper, Georgia Institute of Technology, Atlanta, GA. , A. J. Schaefer, D. Goldsman, E. L. Johnson, A. J. Kleywegt, G. L. Nemhauser. 2002. A stochastic model of airline opera- tions. Transportation Sci. 36 357377. Rothstein, M. 1968. Stochastic models for airline booking poli- cies. Ph.D. thesis, Graduate School of Engineering and Science, New York University, New York. . 1971. An airline overbooking model. Transportation Sci. 5 180192. . 1985. O.R. and the airline overbooking problem. Oper. Res. 33 237248. Rushmeier, R., S. Kontogiorgis. 1997. Advances in the opti- mization of airline eet assignment. Transportation Sci. 31 159169. Ryan, D. M. 1992. The solution of massive generalized set par- titioning problems in air crew rostering. J. Oper. Res. Soc. 43 459467. , B. A. Foster. 1981. An integer programming approach to scheduling. A. Wren, ed. Computer Scheduling of Public Transport Urban Passenger Vehicle and Crew Scheduling. North-Holland, Amsterdam, The Netherlands. Schaefer, A., E. Johnson, A. Kleywegt, G. Nemhauser. 2001. Airline crew scheduling under uncertainty. Working paper, Georgia Institute of Technology, Atlanta, GA. Simon, J. 1968. An almost practical solution to airline overbooking. J. Transport Econom. Policy 2 201202. 390 Transportation Science/Vol. 37, No. 4, November 2003 BARNHART, BELOBABA, AND ODONI Operations Research in the Air Transport Industry Simpson, R. W. 1966. Computerized schedule construction for an airline transportation system. MIT Flight Transportation Labo- ratory Report FT-66-3, Cambridge, MA. Smith, B. C., J. F. Leimkuhler, R. M. Darrow. 1992a. Revenue man- agement at American Airlines. Interfaces 22 831. , , . 1992b. Yield management at American Airlines. Interfaces 22 831. Soumis, F., J. A. Ferland, J.-M. Rousseau. 1980. A model for large scale aircraft routing and scheduling problems. Transportation Res. 14B 191201. Stamatopoulos, M., K. Zografos, A. Odoni. 2003. A decision sup- port system for airport strategic planning. Transportation Res. C. Forthcoming. Stojkovic, G., F. Soumis, J. Desrosiers, M. Solomon. 2002. An opti- mization model for a real-time ight scheduling problem. Transportation Res. 36A 779788. Swedish, W. 1981. Upgraded FAA Aireld Capacity Model Supplemen- tal Users Guide. Reports MTR-81W16 and FAA-EM-81-1, The MITRE Corporation, McLean, VA. Talluri, K. 1998. The four-day aircraft maintenance routing problem. Transportation Sci. 32 4353. , G. van Ryzin. 1999a. A randomized linear programming method for computing network bid prices. Transportation Sci. 33 207216. , . 1999b. An analysis of bid price controls for network revenue management. Management Sci. 44 15771593. Thengvall, B., G. Yu, J. Bard. 2000. Balancing user preferences for aircraft schedule recovery during irregular airline operations. IIE Trans. 32 181193. Tosic, V. 1992. A review of airport passenger terminal operations analysis and modeling. Transportation Res. 26A 326. Vance, P., C. Barnhart, E. Johnson, G. Nemhauser. 1997. Airline crew scheduling: A new formulation and decomposition algorithm. Oper. Res. 45 188200. Vickrey, W. 1972. Airline overbooking: Some further solutions. J. Transport Econom. Policy 6 257270. Vossen, T., M. O. Ball. 2001. Optimization and mediated bar- tering models for ground delay programs. Working paper, University of Maryland, College Park, MD, http://bmgt1- notes.umd.edu/faculty/km/papers.nsf. , , R. Hoffman, M. Wambsganss. 2003. A general approach to equity in air trafc management and its appli- cation to mitigating exemption bias in ground delay pro- grams. Proc. 5th USA/Europe Air Trafc Management R&D Sem., http://atm2003.eurocontrol.fr. Vranas, P., D. Bertsimas, A. R. Odoni. 1994. The multi-airport ground-holding problem in air trafc control. Oper. Res. 42 249261. Wambsganss, M. 1996. Collaborative decision making through dynamic information transfer. Air Trafc Control Quart. 4 107123. Weatherford, L. R. 1997. Optimization of joint pricing and allocation in perishable asset revenue management problems with cross- elasticity. J. Combinatorial Optim. 1 277304. , S. E. Bodily. 1992. A taxonomy and research overview of perishable-asset revenue management: Yield management, overbooking and pricing. Oper. Res. 30 831844. Williamson, E. L. 1992. Airline network seat inventory control: Methodologies and revenue impacts. Ph.D. dissertation, MIT Flight Transportation Laboratory Report R92-3, Cambridge, MA. Wiper, D. S., J. D. Quillinan, R. Subramanian, R. P. Scheff, Jr., R. E. Marsten. 1994. Coldstart: Fleet assignment at Delta Air Lines. Interfaces 24 104120. Wollmer, R. D. 1992. An airline seat management model for a single-leg route when lower fare classes book rst. Oper. Res. 40 2637. Yu, G., M. Arguello, G. Song, S. M. McCowan, A. White. 2003. A new era for crew recovery at Continental Airlines. Interfaces 33 522. Zhao, W., Y. S. Zheng. 2001. A dynamic model for airline seat allo- cation with passenger diversion and no-shows. Transportation Sci. 35 8098. Received: June 2003; revision received: July 2003; accepted: August 2003. Transportation Science/Vol. 37, No. 4, November 2003 391
Journal of Air Transport Management Volume 8 Issue 4 2002 (Doi 10.1016 - s0969-6997 (02) 00003-0) Graham Francis Ian Humphreys Jackie Fry - The Benchmarking of Airport Performance