Operating Cost Modelling

In the Airline Response Model, operating costs are calculated, per flight and per passenger, for each airline as a function of average nominal flight segment travel times and average flight delays, as shown in Figure 4-1. Additional inputs include fuel prices, aircraft fuel efficiencies, and operating costs per hour and per passenger. The calculated airline- specific operating costs are then output to the Network Optimisation Models, which calculate segment flight frequencies and itinerary passenger demand subject to the maximisation of airline profit. The calculated operating costs are also output to an Average Operating Cost Calculator, which calculates system average operating costs by city-pair (O-D) for input to the Average Fare Model.

As described in Section 4.1, airline operating costs modelled include direct operating costs and indirect operating costs associated with aircraft, traffic and passenger servicing; reservation and sales; and other system overheads. Direct operating costs cover fuel and oil costs, crew costs, maintenance costs, aircraft rental, depreciation and amortization costs, and en-route airspace charges. Aircraft servicing costs cover the handling of aircraft on the ground and landing fees. Traffic servicing costs cover the processing of passengers, baggage and cargo at airports. Passenger servicing costs cover meals, flight attendants and in-flight services. Reservation and sales costs cover airline reservations and ticket offices, including travel agency commissions. Other indirect and system overhead costs cover advertising and publicity expenses and general and administrative expenses.

As recommended by Belobaba (2006), some airline operating costs are modelled per flight, while others are modelled per passenger. Costs per flight include all direct operating costs and aircraft servicing costs, with the exception of the proportion of fuel burn that can be attributed directly to passengers. This extra fuel burn, along with traffic servicing costs, passenger servicing costs, reservations and sales costs, and other indirect and system overhead costs are modelled per passenger.

Direct operating costs and aircraft servicing costs are estimated for the three aircraft size classes described in Section 5.3, averaging over all aircraft types in each size class. With the exception of fuel costs and landing fees, all direct operating costs and aircraft servicing costs are derived from the U.S. DOT Form41 data Schedule P52 (DOT, 2005-1). Aircraft

Chapter 5

servicing costs, traffic servicing costs, passenger servicing costs, reservation and sales costs, and other indirect and system overhead costs are input directly from the U.S. DOT Form41 data Schedule P12 (DOT, 2005-2). It is assumed that all operating costs, with the exception of fuel costs, will not change significantly over time beyond inflation.

Landing fees are input directly for each airport from the International Air Transport Association’s (IATA) Airport and Air Navigation Charges Manual (IATA, 2008). For those airports for which landing fees are not available (7 of the 22 airports simulated in Chapters 6 and 7), they are assumed to be equal to the average of those airports for which landing fees are available. Landing fees are not assumed to change over time, except in specific scenarios described in Section 7.2, in which an increase in regional costs is modelled through an increase in landing fees at a subset of airports.

Fuel costs are calculated independently as a function of fuel price and aircraft fuel burn in each of the aircraft flight phases, i.e., ground idle, taxi, take-off, climb-out, cruise, airborne holding, descent, approach and landing. The fuel price is taken from Air Transport Association (ATA) data (ATA, 2008), and is assumed to change over time according to changes in oil price forecast by the MIT Integrated Systems Model (IGSM), run for the U.S. Climate Change Science Program (CCSP) (CCSP, 2007). The IGSM is an integrated energy- economy-environment model with internally consistent population, income, and oil price scenarios. For the United States it represents a relatively high growth scenario when compared to the other energy-economy-environment models run for the CCSP.

Fuel burn rates are estimated using the EUROCONTROL Base of Aircraft Data (BADA) (EUROCONTROL, 2004) and the ICAO Aircraft Engine Emissions Databank (ICAO, 2008) for the representative aircraft in each of the aircraft size and age categories described in Section 5.3. Fuel burn rates are assumed to decrease over time because of the gradual development of more fuel efficient technology and its introduction into the fleet. Fleet fuel burn is assumed to decrease by 0.7% per year through the introduction of more advanced technology, which would replace retired aircraft and satisfy growing demand. This reduction is the average rate forecast by the Energy Information Administration to 2030 (Energy Information Administration, 2009). It does not account for reductions in fuel burn that would be achieved through the introduction of radically new technology such as an

Detailed Modelling

advanced open rotor engine or blended wing body aircraft, which are introduced in specific scenarios in Section 7.3.

The duration of each flight phase is input from the Travel Time Calculator, described in Section 5.3 above, with the exception of ground idle, taxi and airborne holding, which are input from the Delay Calculator, described in Section 5.3.

Most airline operating costs vary by aircraft type. However, some operating costs vary by airport, such as some aircraft, traffic and passenger servicing costs. These are lower at airports with particularly high traffic because of economies of scale. O’Kelly and Bryan (1998) estimate that economies of scale at hub airports in the United States result in aircraft, traffic and passenger servicing costs being between 40% and 95% (averaging 73%) of those at non-hub airports. These reduced costs make it attractive for airlines to operate hub-and- spoke networks. These economies of scale are modelled in this dissertation by exogenously decreasing aircraft, traffic and passenger servicing costs at hubs airports, and increasing them at non-hub airports. This is done in such a way that the aircraft, traffic and passenger servicing costs at hub airports are 73% of those at non-hub airports, and that average aircraft, traffic and passenger servicing costs across all airports equal those calculated from the DOT Form41 data Schedule P12 (DOT, 2005-2).

As described in Chapter 1, there are also other cost advantages of hub-and-spoke networks over point-to-point networks. Hub-and-spoke networks require fewer flights to connect several airports than do point-to-point networks, because they consolidate passengers along major traffic routes. The higher traffic volumes on hub-and-spoke flights allow airlines to operate larger aircraft on these routes, which typically have lower aircraft operating costs per RPM than smaller types. However, the reduction in operating costs is partly offset by the greater total distances flown by passengers and the associated increase in fuel burn. Thus, in some cases, point-to-point networks have lower cost. An example is when a hub is located far from the shortest path between two airports. The lowest cost network is therefore typically a combination of both hub-and-spoke and point-to-point networks.

After the calculation of airline operating costs per flight and per passenger, average operating costs are calculated over all airlines by the Average Operating Cost Calculator. These are output by city pair, per passenger, to the Average Fare Model, described in Section

Chapter 5

5.5, for estimation of average fares by city-pair market. Airline costs per flight are allocated to O-D passengers according to the itinerary passenger demand for each city pair that is on board each flight, while airline costs per passenger are applied directly to O-D passengers.