Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
42 C H A P T E R 6 This chapter provides detailed guidance on preparing a DDFS. Key steps include defining param- eters and estimating markets, fleet mix, flight times, gate/RON assignments, and passengers per flight. Guidance is also provided for non-passenger aircraft operations and quality assurance and control. Preparing a DDFS is a complex effort and the associated guidance is necessarily detailed. As this chapter is primarily directed toward preparers, users interested in a high-level overview may wish to review the summary in Chapter 1 instead. This chapter and the associated exhibits describe the process involved in preparing a DDFS for base year and future conditions. It begins with a general overview of DDFS preparation and fol- lows with step-by-step processes for estimating markets, fleet mix, flight times, gate assignments, passenger loads, and non-passenger aircraft operations. The chapter concludes with guidance on applying constraints, updating DDFSs, and quality control checks. Each of the main sections in this chapter discusses a specific element of DDFS preparation and includes background, detailed step-by-step instructions, and observations and cautions. Preparation steps are accompanied by flow charts in most instances. Steps shown in purple indicate that a decision or assumption is required. Steps shown in blue indicate a data input, typically from annual forecasts or airport/industry sources. Steps in green indicate an intermediate preparation step, usually an internal calculation. Interme- diate or final outputs are shown in orange. In addition, steps shown on the exhibits that are in bold type and ALL CAPS are needed only for DDFSs prepared for future conditions. If the steps are not in bold and all caps, they apply to DDFSs prepared both for base year and future conditions. As noted in Chapters 3 and 4, not all planning/operational issues require a DDFS, and not all potential elements of a DDFS are necessary to address each planning/operational issue. Many of approaches in this chapter involve a significant amount of work that, in many instances, may not be necessary. Please refer to Chapter 4 to identify the DDFS elements that should be included. The following DDFS preparation steps are discussed in this chapter. Click to directly access the section. ⢠Section 6.1 General Steps for Preparing a DDFS ⢠Section 6.2 Setting the Stage ⢠Section 6.3 Forecasting Future Passenger Markets and Fleet Mix ⢠Section 6.4 Forecasting DDFS Flight Times ⢠Section 6.5 Assigning Gates ⢠Section 6.6 Forecasting Passengers by Flight ⢠Section 6.7 Nonscheduled Aircraft Operations How to Prepare a DDFS for Base Year and Future Conditions
How to Prepare a DDFS for Base Year and Future Conditions 43 ⢠Section 6.8 Application of Constraints ⢠Section 6.9 DDFS Updates ⢠Section 6.10 Quality Assurance and Control 6.1 General Steps for Preparing a DDFS This section describes the initial steps required prior to preparing a DDFS and the general processes required to prepare a DDFS. Key initial steps include obtaining stakeholder input on assumptions and setting the stage (see Section 6.2) by determining parameters and collecting pertinent data. The main steps involved in preparing a DDFS are outlined on Exhibit 6.1. First, once the annual forecasts are obtained, market and fleet mix forecasts are then assembled to generate a design day estimate of arrivals and departures by aircraft type, as discussed in more detail in Section 6.3. Next, flight times are estimated for each arrival and departure and arriv- als and departures are paired (see Section 6.4 for more detail). Once flight times are estimated, passenger loads, gate assignments, and flight tracks can be assigned if necessary. These steps are discussed in more detail in Sections 6.5 and 6.6. Stakeholder Input To help ensure the accuracy and credibility of DDFS assumptions, input and buy-in should be obtained early from DDFS users and other key stakeholders, especially the airlines. Key players to engage in this process are corporate airline real estate staff who, in turn, can approach the airlineâs route planners to obtain realistic, applicable input. Included should be perspectives on: ⢠Future fleet mix ⢠Trade-off between cost and passenger service when assigning new flight times ⢠Target aircraft turnaround times by general aircraft category ⢠Target tow-on and tow-off times by general aircraft category ⢠Average gate utilization targets ⢠Airline perspective on the service/cost trade-off between contact gates and hardstands ⢠Gate buffer times by aircraft and flight category (short-haul, long-haul, domestic, international) ⢠Policy on spare gates, if any ⢠Willingness to use common-use gates at peak times ⢠Contingency plans/priorities during irregular operations When making assumptions regarding facility use (i.e., gate or runway assignments), it is par- ticularly important to obtain input from the stakeholders that manage those facilities. Under some circumstances, airlines may be able and willing to provide a future schedule. Note that airline schedules are subject to frequent change depending on changes in business philosophies, aircraft orders, aircraft retirement plans, and competitive factors. In addition, their input may reflect strategies for minimizing the cost and maximizing the control of airport facilities. Discussions with the user of the DDFS (typically the airport operator) should address whether or not the purpose of the DDFS justifies the additional effort involved in assigning specific flights to specific gates. The answer may be yes if the purpose is detailed terminal simulation or con- cessions planning at an airport with multiple terminals or concourses. At smaller airports with centrally located terminal processing facilities, the effort may not be justified. To help ensure the accuracy and credibility of DDFS assumptions, input and buy-in should be obtained early from DDFS users and other key stakeholders.
44 Guidebook for Preparing and Using Airport Design Day Flight Schedules oc es s g a d Q ua y ss u a ce St ak eh ol de rC oo rd in a on /C om m un ic a on Match Aircra Arrivals and Departures (See Exhibit 6.4) Iden fy Parameters Design Day Defini on Treatment of Adjacent Days Ga ng/RON Rules DDFS Assign Gates/ Parking Posi ons (See Exhibit 6.5) Collect Data Exis ng Schedule Profile for Nonscheduled Opera ons Airport Facili es ANNUAL FORECAST Assign Flight Tracks Es mate Passenger Loads (See Exhibit 6.7) DETERMINE POLICY/ PHYSICAL ASSUMPTIONS/ CONSTRAINTS CAPACITY CONSTRAINTS POLICY FACTORS FUTURE GATE/AIRFIELD LAYOUT COLLECT/PREPARE MARKET AND FLEET MIX FORECASTS (SEE EXHIBITS 6.2 AND 6.3) FORECAST FLIGHT TIMES (SEE EXHIBIT 6.4) Prepare Design Day Es mate of Arrivals and Departures by Aircra Type Assumption/Input Intermediate Preparation Step Output Legend Note: Items in BOLD CAPS are required for future DDFSs, but not for base year DDFSs. Pr in n lit A r n Data Input Exhibit 6.1. Preparing a DDFS.
How to Prepare a DDFS for Base Year and Future Conditions 45 6.2 Setting the Stage Once stakeholder input has been obtained, staging steps, including the formulation of policy and facility assumptions, parameters, and data collection can be finalized. Determine Assumptions on Policies, Physical Constraints, and Future Airport Layouts If the DDFS is being prepared for future conditions, assumptions on future operating policies and physical constraints should be determined at this point. These can include nighttime operat- ing restrictions, such as noise curfews; demand management policies, such as slot restrictions; and physical gate or airfield capacity constraints that cannot or are not expected to be mitigated. These factors can all affect the estimates of future flight times. In addition, if the DDFS is intended for use in modeling a planned future airport layout, pertinent information such as gate and concourse locations and capabilities should be deter- mined. Gating, hardstand, and RON use rules should also be established at an early stage as they can require airline input. Gating, hardstand, and RON rules are discussed in more detail in Section 6.5. Parameters The parameters that will govern DDFS preparation help determine the required data inputs and should be determined by the DDFS pre- parer in coordination with the DDFS user. Among the most impor- tant of these is the design day definition. For most facility planning, the design day is a typical busy day that best represents the trade-off between achieving acceptable service levels most of the time and avoid- ing the cost of overbuilding. In many instances, this is defined as an average day or an average weekday during the busiest month. Design days can also be defined as percentiles, for example, the 10th percentile would represent the 36th busiest day of the year, so that 10 percent of the days of the year would be busier and 90 percent would be less busy. For some environmental analyses, the design day is defined as an average annual day. ACRP Report 82: Preparing Peak Period and Operational Profiles- Guidebook http://onlinepubs.trb.org/onlinepubs/acrp/acrp_rpt_082.pdf and ACRP Report 25: Airport Passenger Terminal Planning and Design, Volumes 1 and 2 http://onlinepubs.trb.org/ onlinepubs/acrp/acrp_rpt_025v1.pdf provide guidance to help select representative design days. In some instances, it may be desirable to select more than one design day. At some airports, such as Minneapolis-St. Paul International Airport, numbers of originating passengers peak in a different season than enplaned passengers and aircraft operations. Therefore, a DDFS prepared to examine landside requirements may entail a different design day than a DDFS prepared to examine airfield capacity. An alternative to preparing two different DDFSs is to adjust the DDFS output to account for seasonality. Chapter 7 provides guidance on making these adjustments. Click here to access Chapter 7. At this stage, the preparer and user should determine whether a single day or multiple adja- cent days should be modeled. Although preparing a design week flight schedule requires signifi- cantly more effort, it helps avoid some of the problems involved in selecting a single design day and provides better fidelity in the following circumstances. Day of the week variations: In the past, airlines tended to operate the same schedule during each weekday, despite demand being higher on Mondays and Fridays and lower on Tuesdays For most facility planning, the design day is a typical busy day that best represents the trade-off between achieving accept- able service levels most of the time and avoiding the cost of overbuilding.
46 Guidebook for Preparing and Using Airport Design Day Flight Schedules and Wednesdays. Now, airlines are increasingly matching service to demand, resulting in a less homogeneous weekday schedule. Asynchronous peaks: At some airports, international demand peaks on Saturdays, when domes- tic demand is low and airlines use the available aircraft capacity to offer international service to leisure markets. In these circumstances, using a traditional design day definition could result in the understatement of international demand and international facility requirements. Nighttime Operations: Some airports, especially those with significant international or all- cargo aircraft operations, stay busy through the midnight hour. Demand on some facilities, such as RON parking, may peak at that time. A midnight cutoff for the design day may signifi- cantly reduce the effectiveness of a DDFS in those instances. Under some limited circumstances, a partial DDFS that encompasses the anticipated peak period and the times immediately preceding and following can be used instead of a DDFS for a full design day. Partial DDFSs may be appropriate at small uncongested airports where the peak activity period is clearly defined and carryover delay and recovery times are not an issue. Partial DDFSs are best used to examine requirements for facilities closely tied to peak passenger or aircraft operation flows. The activity peaks at some facilities, such as terminal curbsides, are offset significantly from the at-gate peaks, and may be influenced by off-peak activity flows in ways that are not readily apparent. Requirements for other facilities, such as gates, may be great- est at off-peak times, such as late at night. In these instances, partial DDFSs are not appropriate. Data Inputs Data used to prepare a DDFS can include the following, depending on which DDFS elements are incorporated: ⢠An existing passenger airline flight schedule representing the design day. ⢠An existing all-cargo airline flight schedule if available. U.S.-flag airlines typically no longer publish their schedules, but may make them available to airport operators for ramp management. ⢠Daily activity profiles for nonscheduled operations. Estimates of nonscheduled operations can be obtained from the FAAâs Distributed Operations Network (OPSNET) database or ven- dors that process data from the ASDI system. In addition, some airport traffic control towers (ATCTs) collect data on operations by hour. ⢠Airport noise monitoring data, if available. These data can also be used to identify flight times for nonscheduled operations. ⢠U.S. DOT T-100 data on passengers and load factors by market and airline. ⢠U.S. DOT O&D Survey data for O&D and connecting passenger segmentations by market and airline. ⢠Airport information on existing and future gate and aircraft parking layouts. ⢠Archived FIDS and airfield surface tracking data, if available. A more extensive description of potential data sources is provided in Appendix F. Airline schedules can be filed up to 12 months in advance, but their accuracy and reliability tend to be less accurate the further in the future they go. In some cases, airlines overschedule in their advance schedules and then trim flights to match booking demand. In other instances, it is uncertain which aircraft would be most appropriate to serve each route, and some airlines use place-holder aircraft in their advance schedules to be later replaced with right-sized aircraft when advance bookings provide a better gauge of demand. Therefore, when an advance schedule is used as the basis for a DDFS (e.g., a July peak month schedule is used as the basis for a DDFS being prepared in March), some adjustments for these
How to Prepare a DDFS for Base Year and Future Conditions 47 factors may be required. These adjustments include scaling back future seat capacity in the advance schedule to match current year-to-year growth, or adjusting the fleet mix in the advance schedule to match the airlinesâ current fleet mix. Some DDFS preparers use airline schedules without adjustment and others make adjust- ments to match daily aircraft operations. Because of cancellations, not all scheduled operations actually occur. Therefore, many DDFSs contain slightly more aircraft operations than are actu- ally flown. These differences are subtle, but may significantly alter results at highly constrained airports where a slight change in operational levels can result in a significant change in average delay. Therefore, it is essential to document whether a DDFS represents scheduled operations or completed operations. Perhaps the most important input to a DDFS is the annual activity forecast. These forecasts may be master plan forecasts, forecasts prepared for other purposes, such as determining finan- cial feasibility, or FAA Terminal Area Forecasts (TAF). The TAFs lack the necessary fleet mix detail to be used directly for DDFS preparation without significant additional work on the part of the preparer. Note, however, that a DDFS that is based on an annual forecast that is not consistent with the TAF (defined as within 10 percent over the first five years or 15 percent over the first 10 years), additional discussion with the FAA will be required to obtain FAA agreement necessary for FAA funding and environmental approval (FAA 2008). The assumptions underlying the annual and DDFS forecasts should be consistent. For exam- ple, if the annual forecasts include assumptions regarding new nonstop markets or a future fleet mix, those forecast assumptions are typically incorporated in the DDFS to save time and effort in addition to maintaining consistency. Appendix A provides an example of some of the activities involved in setting the stage. Click here to access Section A.2. Setting the Stage. 6.3 Forecasting Future Passenger Markets and Fleet Mix O&D markets are a major determinant of aircraft type and passenger characteristics. How- ever, there is no single industry standard for forecasting new markets and the fleet mix associ- ated with each market. Two options for forecasting markets and fleet mix are presented here. The first option, described in Exhibit 6.2, is a detailed bottom-up passenger-focused approach, whereas the second option, shown in Exhibit 6.3, is a less-detailed top-down operations-focused approach. In some instances, markets may not need to be identified (see Chapter 4). Background/Considerations Regardless of the option selected, the following should be considered when forecasting future markets and fleet mix. Airline market share is very difficult to predict because of mergers, bankruptcies, changes in alliances, and changes in business plans. For airport planning that does not involve a DDFS, air- craft operations and passengers are usually not differentiated by airline and the issue is avoided. In a DDFS, aircraft need to be assigned to gates and markets. As gates and markets are often airline-specific (especially if the gates are not common use), the airline designation becomes more important. At most connecting hub airports, one airline serves most of the connecting traffic and has very different ratios of O&D to connecting passengers than the other airlines. The assumptions under lying annual and DDFS forecasts should be consistent.
48 Guidebook for Preparing and Using Airport Design Day Flight Schedules The approach involves the following steps. MARKET GROWTH ASSUMPTIONS ANNUAL FORECASTS ALLOCATE DEPARTING SEATS BY AIRCRAFT TYPE (AND AIRLINE) FUTURE DESIGN DAY OPERATIONS BY MARKET AND AIRCRAFT TYPE (PROCEED TO EXHIBIT 6.4) Design Day Defini on ESTIMATE FUTURE DESIGN DAY PASSENGERS AIRCRAFT USE ASSUMPTIONS DISTANCE SIZE TYPE FREQUENCY ALLOCATE PASSENGERS BY MARKET AIRLINE STRATEGY ASSUMPTIONS MARKET SHARE FUTURE FLEET FUTURE LOAD FACTOR ASSUMPTIONS MARKET AIRLINE ESTIMATE DEPARTING SEATS BY MARKET Assumption/Input Data Input Intermediate Preparation Step Output Legend Note: Items in BOLD CAPS are required for future DDFSs, but not for base year DDFSs. Exhibit 6.2. Forecasting future markets and fleet mix for DDFS (Option 1).
How to Prepare a DDFS for Base Year and Future Conditions 49 MARKET SHARE ASSUMPTIONS ANNUAL AIRCRAFT OPERATIONS FUTURE DESIGN DAY OPERATIONS BY MARKET AND AIRCRAFT TYPE (PROCEED TO EXHIBIT 6.4) Design Day Definion AIRCRAFT USE ASSUMPTIONS DISTANCE FREQUENCY AIRLINE STRATEGY ASSUMPTIONS MARKET SHARE FUTURE FLEET MIX Assumption/Input Data Input Intermediate Preparation Step Output Legend Note: Items in BOLD CAPS are required for future DDFSs, but not for base year DDFSs. ANNUAL FLEET MIX FORECAST DESIGN DAY FLEET MIX SEASONALITY FACTORS Exhibit 6.3. Forecasting future markets and fleet mix for DDFS (Option 2).
50 Guidebook for Preparing and Using Airport Design Day Flight Schedules Identifying the airline then becomes important in determining O&D flows. Some nonstop markets may be viable for some airlines but not others. For example, a hubbing airline would be able to supplement O&D passengers with connecting passengers, which often determines whether there is a sufficient amount of traffic to sustain service. Approaches to estimating the growth of existing nonstop mar- kets can range from assuming a constant market share to tying mar- ket growth to economic factors such as income, or even to developing separate forecast equations for each market. Sometimes, growth factors are applied directly to aircraft operations; sometimes, they are applied to scheduled departing seats that are then converted to aircraft operations; and other times they are applied to passenger traffic that is then converted to departing seats and aircraft operations. The growth may be accommodated with larger aircraft, higher load factors, more flights, or a combination of those factors. Approaches to identifying new nonstop markets are equally varied. They include: Defining certain flights as operating to and from new markets without specifying the market: This approach has the virtue of simplicity, but some flight characteristics, such as aircraft type and flight time, are dependent on the market served. Segmenting new markets by category (i.e., short- haul domestic, long-haul domestic, international) can mitigate some of the shortcomings of this approach. Judgment-based analysis: In this approach, the preparer uses his or her knowledge of airline behavior and existing nonstop markets to anticipate the markets in which airlines are likely to add new nonstop service. This approach is difficult to document, which may be a shortcoming in highly controversial projects. Using historical airline service as a guide to future airline service: In this approach, it is assumed that airlines are most likely to reintroduce nonstop service in markets that have been served nonstop in the past. An inherent assumption is that the factors leading to the future intro- duction of new nonstop service will be similar to the underlying factors in the past. However, many markets have lost nonstop service because of the evolution of the aviation industry. For example, many small short-haul markets have lost service because airlines no longer operate the 19-seat aircraft that are optimal for those markets and/or travel by automobile is a viable alternative. This trend is unlikely to be reversed. Using O&D passenger or airline revenue thresholds: In this approach, the current minimum threshold (measured in numbers of O&D passengers or airline revenue) needed to justify nonstop service is calculated within each distance band from the airport under analysis (500 miles, 1,000 miles, etc.). As demand increases, consistent with the annual forecasts, more and more unserved markets exceed the minimum threshold and are, therefore, assumed to gain nonstop service. This approach is more resource-intensive than the others. This approach has theoretical appeal, but has not been empirically validated. Also, the thresholds will likely differ for hubbing airlines that can rely on additional connecting traffic compared to nonhubbing airlines. Air service analysis: In this approach, a detailed air service analysis, addressing potential airline revenue, demographic characteristics, potential feed traffic, and other factors that airlines consider in their route analyses, is conducted for each potential nonstop market. This is a very costly under- taking, but, in some cases, the airport operator may have already undertaken air service analyses as part of its airline route development and marketing efforts, which can be applied to the DDFS. It is also important to identify which existing nonstop markets may be lost. As noted ear- lier, many short-haul markets are losing service because the optimal aircraft for serving those At most connecting hub airports, one air- line serves most of the connecting traffic and has very different ratios of O&D and connecting passengers than the other airlines.
How to Prepare a DDFS for Base Year and Future Conditions 51 markets are no longer in operation. This suggests that the thresholds for nonstop service may be shifting upwards. Fleet mix will vary by market. Generally, short-haul markets are served with small aircraft at high frequency, and long-haul markets are served with large aircraft at low frequency. Competitive markets (those served by more than one airline) tend to be served by smaller aircraft with greater frequency than noncompetitive markets of similar size and segment distance. Business markets tend to be served with smaller aircraft at greater frequency than leisure markets because business travelers select flights largely on the basis of schedule. Oper- ating a greater number of smaller aircraft costs the airlines more on a seat-mile basis, but they are able to recoup those costs because of the premium fares paid by time-sensitive business travelers. Steps for Forecasting Passenger MarketsâOption 1 (Passenger Based) Exhibit 6.2 shows a top-down passenger-based approach for forecasting and distributing future design day activity among markets (Option 1). The result is a fleet mix forecast showing airline, aircraft type, and daily service frequency in each market. The approach involves the following steps: Steps for Estimating Passenger Markets and Fleet MixâOption 1 1. Estimate future design day passengers if not available from the annual forecasts. ACRP Report 82: Preparing Peak Period and Operational ProfilesâGuidebook http://onlinepubs.trb. org/onlinepubs/acrp/acrp_rpt_082.pdf provides options for estimating design day passengers based on the userâs criteria. 2. Allocate passengers among markets. Some detailed annual forecasts include distribution of passengers by market. If annual market forecasts are not available, potential allocation methodologies are listed below, ranked in order of least complex to most complex: a. Allocate passengers according to existing shares. b. Grow passengers in each market according to recent trends and then normalize (propor- tionally adjust) results to sum to original design day total. c. Grow passengers in each market according to the anticipated growth in a market-demand proxy, such as income in the destination market, and then normalize results to sum to original design day total. d. Grow passengers in each existing market in accordance with 2.c. above, identify new nonstop markets using one of the approaches discussed earlier in this section, and then normalize results to sum to original design day total. e. Prepare a separate forecast for each market, including new nonstop markets, and then normalize results to sum to original design day total. 3. Estimate future load factor for each market and then divide the passenger forecasts for each market prepared in Step 2 above by the estimated load factors to generate a departing seat forecast for each market. 4. Estimate the fleet mix most likely to account for daily departing seats to each market. This will involve some judgment and should include the following considerations: a. Existing service patterns to the market. b. Current airline route strategies. c. Degree of competition in the market. Markets in which airlines compete tend to have more service frequencies using smaller aircraft than monopoly markets of similar size and distance. d. Known planned aircraft orders and retirements for each airline.
52 Guidebook for Preparing and Using Airport Design Day Flight Schedules e. Relationship among market size, average aircraft size, and flight frequency. This relation- ship tends to change with increased distance; long-haul markets tend to be served by larger aircraft with fewer frequencies compared to short-haul markets of similar size (measured in departing seats). f. Adjust as necessary so that the sum of aircraft types serving each market is consistent with the annual fleet mix forecast. Some preparers use a variation of Option 1, in which, after passengers are allocated among markets (Step 2), it is assumed that the airline(s) will attempt to maximize aver- age load factors in the market before adding service. The maximum load factor is defined as the maximum load factor in other similar markets. Once load factor is maximized, the airline(s) are assumed to increase equipment gauge (aircraft size), to the extent possible. Maximum realistic aircraft size is again defined by experience in other, similar markets. Flight frequencies are added only after load factor and aircraft gauge are increased to the maximum realistic levels. Steps for Forecasting Passenger MarketsâOption 2 (Operations Based) Exhibit 6.3 shows a simpler, top-down operations-based approach for estimating and dis- tributing future design day activity among markets (Option 2). This approach involves the following steps. Steps for Forecasting Passenger Markets and Fleet MixâOption 2 1. Assemble a design day fleet mix from the annual fleet mix. At some airports, the fleet mix during the busy season differs from the annual fleet mix. However, the design day fleet mix trends should be consistent with annual trends. For example, if the annual forecasts show a phase-out of 50-seat regional jets, this phase-out should be reflected in the design day fleet mix. 2. Estimate the percentage of flights to existing nonstop markets and new nonstop markets. Methods for forecasting new markets are discussed earlier in this section. 3. Allocate the existing market share of flights among existing markets according to existing distributions. 4. Adjust the fleet mix in each market in a manner consistent with the assumed fleet mix changes for the primary airlines serving the market and the range of characteristics of the aircraft types. Adjust as necessary so that the sum of aircraft types serving each market matches the annual fleet mix. Estimating Future Passenger Markets and Fleet Mixâ Observations and Cautions As a general caution, it is not a given that new nonstop markets will be served. Some airports, especially smaller airports, have been losing rather than gaining nonstop markets. As each aircraft that lands must take-off again, the distribution of aircraft types and airlines is symmetrical between arrivals and departures in the long term. There are some exceptions during transition days, such as Fridays, Saturdays, Sundays, and Mondays, as airlines ramp their opera- tions up or down to adjust for lower or higher weekend demand. When the selected design day is in the middle of the week, this is typically not an issue, but it must be considered when preparing multiple DDFSs representing several adjacent days.
How to Prepare a DDFS for Base Year and Future Conditions 53 Typically there is a rough symmetry in the distribution of aircraft types and airlines between arrivals and departures within a given market pair, but the symmetry is not exact. Slight differ- ences in aircraft types between arrivals and departures are not unusual and, in some cases, the total number of aircraft arrivals and departures does not match in a given market pair. For an actual example of how some of these approaches to forecasting markets and fleet mixes were applied, see Appendix A. Click here to access Section A.3. Future Markets and Fleet Mix. 6.4 Forecasting DDFS Flight Times Forecasting future flight times is a key part of preparing a DDFS. This section provides general background, a detailed step-by-step process for forecasting flight times, and some observations and cautions. Background/Considerations There is no standard approach to forecasting future flight times, but based on the research conducted as part of the development of this guidebook, several general observations and principles should be considered. ⢠In general, the hourly pattern of arrivals and departures tended to be consistent during the past 10 years at most airports. That is, the peaks and valleys in the daily profile of activity at the airport tend to occur around the same time over the years as a result of geographic location and airline route network strategies. This is not to say that the profiles are absolutely rigid. Variations of 5 to 10 percent in each hourâs share of design day scheduled operations or sched- uled seats are not unusual. Airport size is positively correlated with the stability of schedules (e.g., large hubs have more stable schedules than medium hubs and so on). ⢠Total operating schedules are more stable than separate arrival and departure schedules. ⢠The stability of domestic operations schedules is similar to that of total operations, while international operations schedules are more variable. ⢠Individual airline schedules are less stable than total airport schedules. ⢠There is no discernible trend in airline-specific schedule stability; the scale of individual air- line operations at an airport is a more important driver than the specific airline. For example, there is no evidence that, given an equal number of operations, a low-cost carrierâs schedule is more or less stable than a legacy carrierâs schedule. ⢠Approaches to determining flight times for new service will vary by airline. Network airlines focus on providing flights at peak times to and from their hubs. Other airlines focus on the peaks as well, but also on maximizing gate utilization to provide operating cost efficiencies. ⢠International schedule profiles tend to be less stable than domestic schedule profiles. Flights to specific international regions (e.g., Europe, northeast Asia, and southern Latin America) tend to operate within specific schedule windows, but if the mix of flights to each region changes, the overall profile of international operations may change significantly. See Appendix N in ACRP WOD 14 (Technical Report accompanying ACRP Report 82) http://onlinepubs.trb.org/ onlinepubs/acrp/acrp_w014.pdf for additional detail on international profiles. ⢠Major structural changes to an airline schedule, such as an increase or decrease in the number of connecting banks, a transition to a rolling hub, or de-hubbing, will cause more significant changes to schedule profiles. ⢠Although airports exhibit certain tendencies regarding the timing of the peak hour within the day, shifts in the timing of the peak of 2 or 3 hours can still occur. ⢠Over time, the peak hour share of daily operations can range from three percent to almost 20 percent above or below the long-term trend, depending on the size of the airport.
54 Guidebook for Preparing and Using Airport Design Day Flight Schedules ⢠Even when daily flight frequencies remain constant, flight times to individual markets change often. These changes most often are within 10 to 15 minutes of the prior scheduled time, but can be longer. ⢠Existing flight times are much more likely to change if an airline adds or deletes a flight to a market. If there is a change in frequency, airlines will reschedule remaining flights to maximize service coverage during the day. Under these circumstances, flight times can vary by several hours, especially if the flight is not the first or last flight of the day. Appendix Q in ACRP WOD 14 (Technical Report accompanying ACRP Report 82) http://onlinepubs.trb.org/onlinepubs/ acrp/acrp_w014.pdf provides more background information. These observations are discussed in more detail in Appendix B. Click here to access Appendix B: Stability and Predictability of Critical DDFS Factors. Steps for Forecasting DDFS Flight Times Exhibit 6.4 shows a general approach for forecasting flight times for passenger aircraft operations. This approach involves the following steps. Steps for Forecasting DDFS Flight Times 1. Begin with the market forecast of aircraft operations by airline, aircraft type, and flight frequency and an existing schedule of passenger aircraft arrivals and departures (see Section 6.2). 2. For each destination (departure) market, update each existing flight to reflect changes in equipment, if any, and add new service frequencies. When forecasting scheduled times for new flights to existing markets, consider the following factors: a. Avoid scheduling two flights to the same market by the same airline at the same time. Airlines try to avoid wingtip-to-wingtip flights when possible. b. If the airport being analyzed serves as a connecting hub for an airline, add new flight fre- quencies to those connecting banks that currently have no service to the destination market. Check to see that scheduled arrivals at the destination market would occur at a reasonable time (see 2.d. below). c. If the airline operates a connecting hub at the destination airport, schedule flights so that the arrival times at the destination airport occur in connecting banks that currently have no service from the airport being analyzed. d. Schedule flights to avoid arriving during nighttime hours (2300â0600) at U.S. destination airports. Note, however, that some U.S. and non-U.S. airports with major international service can be very busy during nighttime hours. e. Determine whether or not to adjust existing flight times in markets where new flights are added. For example, an airline may currently schedule two departures to a market, one in the morning and one in the late afternoon. If a flight is added, the new schedule may include one flight in the morning, one in the early afternoon, and one in the evening. Thus, the original late afternoon flight time would be eliminated and replaced by two new flight times. f. A graph of existing departing flight times by market time zone or distance can serve as a quick reference of realistic times for new flights. 3. Estimate times for service to new markets, taking into consideration the factors described in Step 2. Use similar markets (in terms of size, distance, and time zone) with existing service as a guide to likely service times. 4. Repeat Steps 2 and 3 for origin (arrival) markets. 5. As each aircraft that lands must take-off, arriving flights must be paired or matched with departing flights, unless the flight is the first departure or last arrival of the day. (At busy airports, pairings may change as a result of operational considerations. For example, if an Existing Flight times are much more likely to change if an airline adds or deletes a flight to a market.
How to Prepare a DDFS for Base Year and Future Conditions 55 PEAK SPREADING FACTORS Exisng Flight Schedule FUTURE DESIGN DAY OPERATIONS BY MARKET AND AIRCRAFT TYPE (EXHIBITS 6.2 AND 6.3) FLIGHT TIME ASSUMPTIONS UPDATE EXISTING FLIGHTS TO REFLECT FUTURE FLEET MIX Pair Arrivals and Departures Ungated DDFS (Proceed to Exhibit 6.5) Turnaround Assumpons RECONCILE ARRIVAL AND DEPARTURE TIMES ASSIGN TIMES AND AIRCRAFT TYPES FOR NEW FLIGHTS Assumption/Input Data Input Intermediate Preparation Step Output Legend Note: Items in BOLD CAPS are required for future DDFSs, but not for base year DDFSs. PEAKING CONTROLS CAPACITY CONSTRAINTS THROUGHPUT SLOTS GATES REVIEW EXISTING FLIGHT TIMES TO ENSURE EVEN SCHEDULE COVERAGE Exhibit 6.4. Forecasting DDFS flight times for scheduled aircraft operations.
56 Guidebook for Preparing and Using Airport Design Day Flight Schedules arriving flight is delayed, a different aircraft than originally planned may be substituted for a scheduled departure. As a practical matter, this is difficult to model in a planning DDFS.) The time difference between the arrival and subsequent departure must be long enough to allow the offloading and loading of passengers, cargo, and fuel and short enough to allow the aircraft to be fully utilized during the day. In general, turnaround times are determined by the structure of the connecting banks and aircraft size although brake and tire cooldown times are sometimes factors. Small regional aircraft can often turn around in 20 to 30 minutes. Mainline aircraft generally take at least 45 minutes or more, unless they are operated by South- west Airlines, in which case they can turn around in as little as 30 to 35 minutes. Widebody aircraft in domestic service, including Alaska and Hawaii, usually require at least an hour and widebody aircraft in overseas international service often require a 2-hour turnaround time. 6. Many preparers use the first-in/first-out approach to pair flights. This involves sorting arriv- als and departures by airline and aircraft type, and then pairing each arrival with the first departure that meets the turnaround criteria in Step 5. 7. In some instances, airlines will hold a few aircraft departures back to provide some contin- gency in the schedule in case of delayed arriving aircraft or mechanical breakdowns or to serve a market at a more competitive time. The existing schedule should provide a good guide as to how often airlines plan for these contingencies. For example, if a midday departure cannot be paired using the approach in Step 6, it may be an aircraft that has been held over since the previous night. 8. Generally, after all apparent pairings of arriving and departing aircraft have been completed in a future DDFS, there will be a few remaining flights for which there are no obvious pairs. From a demand standpoint, the best times to add new arriving flights do not always corre- spond to the best times to add new departing flights. For example, at East Coast airports, there may be a late morning schedule gap for departures to West Coast airports. Adding long-haul aircraft departures at that time may not be feasible, however, as the availability of long-haul aircraft arriving during that period may be insufficient because those arriving aircraft would have departed their West Coast airport of origin sometime between midnight and 0600. If the number of unmatched pairs is too high to be reasonably explained by airline contingency planning, it will be necessary to iteratively adjust flight times, while adhering to the consider- ations in Steps 2 and 3, until the remaining arriving and departing aircraft can be paired. 9. Additional adjustments may be required to ensure that the schedule complies with policy constraints, such as slot restrictions, or physical constraints, such as gate or airfield through- put capacity. See Section 6.8 for additional discussion of constraints. Forecasting DDFS Flight TimesâObservations and Cautions When airport activity increases, peak activity also increases, but usually at a lower rate. There- fore, the peak percentage tends to decrease as an airport becomes busier. ACRP Report 82: Pre- paring Peak Period and Operational ProfilesâGuidebook http://onlinepubs.trb.org/onlinepubs/ acrp/acrp_rpt_082.pdf provides additional discussion of peak spreading and some approaches to forecasting the effects. There are differing strategies for dealing with peak spreading in DDFSs. One top-down strat- egy is to pre-define the peak activity periods and use these predefined peak periods as controls for determining the distribution of flight times in the DDFS. The second, bottom-up strategy is to build the DDFS without peak controls and let the DDFS results define the peak period. The advantage of the top-down strategy is that it helps mitigate conscious or unconscious biases that a DDFS preparer may have for or against certain flight times. In some instances, the reconciliation of arrival and departure links may unintentionally cause the migration of operations or aircraft equipment to times not representative of real world operations. When air- port activity increases, peak activity also increases, but usually at a lower rate.
How to Prepare a DDFS for Base Year and Future Conditions 57 The bottom-up strategy is advantageous when the airlines serving an airport are expected to change their strategies. For example, a change in the number of connecting banks will create a discrete change in the peak hour percentage. The markets and number of operations in each forecast bank will depend on individual market characteristics and are difficult to determine using a top-down strategy. Another example would be a gateway airport currently serving Euro- pean markets, with a characteristic late afternoon and evening international peak. If service at the airport is added to South American or Asian markets, existing peaks will be a poor guide for determining flight time distributions for the additional flights. Based on the discussion above, imposing top-down controls on the distribution of DDFS flight times may be too rigid, and may generate results that fail to incorporate more subtle trends that can emerge from bottom-up DDFS preparation. A potential compromise may be to establish soft controls (see Section 6.10) to establish boundaries from which DDFS-generated profiles could not deviate without good cause. Regardless of the strategy used, the future profile of activity should be cross-checked with the existing profile to ensure that there are no deviations that cannot be readily explained by changes in service patterns, such as the mix of domestic and international flights. Appendix A provides an example of how new flight times were forecast. Click here to access Section A.4 Flight Times. 6.5 Assigning Gates For many airfield simulation analyses and for terminal planning, the aircraft arrival and departure pairs will need to be assigned to gates. Background/Considerations At most airports, airlines determine which flights use specific gates. Airlines begin the process of converting airline schedules to gate requirements with their airline schedule planning group generating a proposed flight schedule. Then, either the airlineâs schedule planning, operations planning, or corporate real estate (CRE) group converts the schedule to preliminary require- ments, in some cases relying on commercial gate plot software and in other cases relying on internally developed software programs. The next airline step depends on the complexity of the gating issues. Determining the gate requirements for three to four flights spread throughout the day may be a matter of the CRE group simply confirming that a single gate is available for lease or use from the airport operator and on what terms. On the other hand, optimizing the number of gates required for a large num- ber of originating flights followed by a much smaller number of flights throughout the day usu- ally requires further analysis and discussion between the CRE group, station managers, schedule planning, and, in some cases, finance. The question is often whether or not the originating flights can be towed to and from the gate to avoid the need to lease gates that will not be needed after the peak. There is a trade-off between customer service (aircraft at gate when needed) and cost control (aircraft towed to or from the gate until needed). Other areas of internal discussion in determining gate requirements are the effect of weather, the need to add more buffers between flights if the schedule is tight, and the availability of common-use or other gates on a per-use basis to accommodate peaks. In balancing customer schedule preferences with airport costs, the relative prioritization depends on the business model of each airline and may change over time. Gate assignment decisions are usually made by airline station managers. Sometimes, there is additional coordination with airport operations, especially for common-use gates, and U.S. CBP when international services are required. In allocating flights to particular gates, the critical
58 Guidebook for Preparing and Using Airport Design Day Flight Schedules factor is the ability to accommodate particular aircraft in terms of length and wingspan. Beyond that, airlines consider multiple factors when assigning flights to particular gates, with different airlines taking different approaches. The following is a composite list of the airline consider- ations involved in assigning flights to specific gates: ⢠Park aircraft reasonably close to their usual departure and arrival runways. ⢠If a high proportion of deplaning passengers are connecting to a single destination, use a nearby gate for the connecting flight. ⢠Evenly distribute workload among available station personnel. ⢠When possible, assign the same flights to the same gates, especially for business markets. ⢠Separate the boarding areas for business and leisure flights. ⢠Avoid overcrowding boarding areas with high-demand flights. Assign flights with high passenger loads to gates with larger holdrooms and avoid assigning two such flights to adjacent gates. ⢠Increase buffer times between flights as much as possible. ⢠Minimize congestion in taxiways adjacent to concourses. ⢠Locate the last arriving flights in a bank near the center of the hubbing airlineâs gates rather than on the periphery to avoid passengers having to run from the end gate. ⢠Assign flights from the same bad weather/delay destinations to the same gates (as the delays tend to affect multiple flights in a similar manner). ⢠Coordinate with airport operators to assign flights to gates near concessions that best serve particular passenger characteristics, especially for business travelers. Another factor to consider is that some large airports have stand-alone terminal buildings. Moving connecting passengers between terminal buildings is inconvenient for passengers, air- lines, and airport staff. Therefore, to the extent possible, airlines and airline alliances should be assigned gates within the same terminal building. Steps for Assigning Gates To the extent possible, the above factors should be considered when assigning gates in a DDFS. In many instances, a DDFS preparer will not have access to airline gating policies or the resources to model each aspect. Exhibit 6.5 provides a more general approach for assigning gates in a DDFS. Many firms have proprietary models that automate the process. Whether the process is manual or automated, the following factors should be considered. Steps for Assigning Gates 1. If the requirements of the analysis dictate that the DDFS be gated, a gate layout repre- senting future terminal area conditions is required when existing gate capacity is not sufficient to fully accommodate forecast activity. If a future layout is not available, un- accommodated flights can be assigned to virtual gates. However, general locations must be assigned to those virtual gates if the DDFS is used as input to airfield simulation analysis or if the terminal and roadway analysis is subdivided by terminal building, concourse, or other segmentation. 2. Not all gates are configured to accommodate all aircraft types. Aircraft should be assigned only to gates that can, or are planned to, accommodate those aircraft categories. 3. Sufficient buffer time between a departing flight and the next arriving flight at a gate should be included. Current gate scheduling practices at the airport under analysis should be examined to determine the appropriate buffer times. At preferential- or exclusive-use gates, the buffer time is typically no less than 15 minutes for a domestic flight and no less than 30 minutes for an overseas international flight. Many airlines use buffer times of 30 minutes or more even for domestic flights. If common-use gates are contemplated, buffer times should be increased
How to Prepare a DDFS for Base Year and Future Conditions 59 Balance Gate Use Un-Gated DDFS (see Exhibit 6.4) Gated DDFS (without Passengers) (Proceed to Exhibit 6.7) Assign Flights to Gates GaÂng Model Manual Assignment Gate Use AssumpÂons Airline Assignments Aircra Capacity DomesÂc/InternaÂonal Buffer Times Tow-on/Tow-off Times Gate UtilizaÂon Rates Assumption/Input Intermediate Preparation Step Output Legend Note: Items in BOLD CAPS are required for future DDFSs, but not for base year DDFSs. POLICY/PHYSICAL ASSUMPTIONS/CONSTRAINTS CAPACITY CONSTRAINTS POLICY FACTORS Current Gate Layout FUTURE GATE LAYOUT (SEE EXHIBIT 6.1) Exhibit 6.5. Assign gates for DDFS. because individual airlines have less internal flexibility to optimize the distribution of their aircraft among gates to accommodate disrupted schedules. Long-haul flights, because of head- winds and other contingencies, tend to have more unpredictable arrival times than short-haul flights and may, therefore, warrant a longer buffer time. In addition, disrupted schedules are more likely at highly congested airports, and increased buffer times will be more appropriate in those instances. 4. Some airlines, especially those that operate connecting hubs, lease spare gates to accommo- date disrupted schedules. Spare gates are not always obvious, and may change from hour
60 Guidebook for Preparing and Using Airport Design Day Flight Schedules to hour, but at any given time, a certain percentage of an airlineâs gates will have no flights scheduled to provide for unexpected aircraft. There are no general rules regarding the need for spare gates. Historical estimates range from two percent to seven percent of gates at large- hub airports that are considered as spares. However, instead of designating certain gates as unscheduled spare gates, airlines are now more likely to schedule all gates, but with some additional buffer time to better accommodate irregular operations. 5. Buffer times and spare gates are intended to address the same issue: to provide additional gate capacity in case flight schedules are disrupted and off-schedule flights result in a higher demand for gates than anticipated under the original schedule. Therefore, it is not realistic to be too generous or too conservative with both buffer times and spare gates. If an airline has long buffer times, it can operate with fewer spare gates. If it has short buffer times, more spare gates will be required. 6. Some airlines have preferred runways for destinations in a given direction, and they assign gates to minimize taxiing time to those runways. Existing gate assignment patterns should be examined for these practices. If a gate assignment chart is not available from the airport operator or airline, gate assignments for individual flights can be determined in real-time from FIDS, which are often accessible using the Internet. 7. At large airports, gating models can theoretically schedule 15 to 20 daily flights at one gate (usually the first to be gated) and only one peak hour flight at another gate (usually the last to be gated). This scheduling does not occur in the real world. Airlines and airport operators will attempt to balance gate use to avoid overly stressing a given facility. Utilization across gates in the design day schedule should be balanced to match current use patterns. In general, airlines rarely exceed 8 to 10 daily turns per gate. 8. At many airports, RON demand exceeds gate demand. In those instances, arriving RON air- craft must be towed from a gate to a remote parking area to free the gate for the next arriving aircraft. The following morning, the aircraft is towed back to a gate to depart once the gate has been vacated by a previous aircraft departure. Aircraft dwell times before tow-off and after tow-on must be assigned to these aircraft to allow passengers sufficient time to deplane or enplane. These times can vary from 20 minutes or 30 minutes for small aircraft to 45 minutes or more for large aircraft. Assigning GatesâObservations and Cautions Users should be aware that gating models or approaches may yield different results depending on the algorithms used. A model that gates aircraft in order of size may gener- ate fewer large aircraft gate requirements but more overall gate requirements. A model that gates aircraft in order of arrival/departure time (from first in the morning to last in the evening) may generate fewer overall gate requirements but a greater number of large gate requirements. Exhibit 6.6 provides a simple example. Case 1 and Case 2 show alternative gating approaches to an identical four-flight schedule. In Case 1, widebody aircraft are gated first. Once they are gated, the two narrowbody aircraft cannot share a gate and, therefore, two additional gates are required. In Case 2, all aircraft are gated in order of arrival flight time. In this approach, only two gates are required, but they must both accommodate widebody aircraft. Once all flights are gated, the schedule is ready for airfield planning or simulation modeling. Additional steps, outlined in the next section, will be necessary to use the schedule for terminal or landside planning. Appendix A provides a description of the gating process, including a Gantt chart. Click here to access Section A.5 Gate Assignments.
How to Prepare a DDFS for Base Year and Future Conditions 61 6.6 Forecasting Passengers by Flight When DDFSs are used for terminal and landside planning, they must be translated into pas- senger flows. This requires forecasting passenger loads from enplaning load factors and O&D/ connecting passenger splits for each flight. Background/Considerations Enplaning Load Factor: The simplest approach for assigning load factors is to assume the same airport average load factor for all flights. This approach is used when analysis resources are lim- ited and the degree of precision is not critical. Several approaches can be used when more detail is required. One approach is to assume that enplaning load factors in each market will increase at the same rate as the overall forecast airport load factor. Another approach is to assume that mar- ket load factors will converge toward a common mean. The assumption underlying the second approach is that airlines will view high load factors as a signal to add more service, and low load factors as a signal to reduce service. Research conducted during the development of this guidebook suggests that load factors are, in fact, converging toward a common mean. Load factors are increas- ing in general, but mostly as a result of significant increases in load factors on flights to markets that previously had low load factors. O&D/Connecting Passenger Split: When DDFSs are used for terminal or landside planning at a connecting hub airport, an estimate of the split between O&D (local) and connecting passengers on each flight is recommended as such split affects facilities differently and their relative impor- tance may change over the course of the day. This estimate can be complicated for individual flights for two reasons. First, because of fares and schedules, not all passengers take the most direct route to their destinations. For example, not all passengers traveling from San Francisco International Airport Case 1 Case 2 Exhibit 6.6. Impact of gating algorithm on gating requirements.
62 Guidebook for Preparing and Using Airport Design Day Flight Schedules (SFO) to John F. Kennedy International Airport (JFK) in New York will board a nonstop New York flight. Some may connect through another airport to take advantage of a lower fare. Secondly, the definition of a local (O&D) passenger for airport planning differs from the defi- nition used by the U.S. DOT, which collects O&D statistics. Using the SFOâJFKâBerlin passen- ger example, that passenger would be an originating passenger for the San FranciscoâNew York flight for airport planning purposes, but the U.S. DOT would classify that passenger as a San FranciscoâBerlin O&D passenger. From the perspective of the SFOâJFK flight, the SFOâBerlin passenger counts as a beyond O&D passenger. When resources are available, detailed O&D data segmented by itinerary can be used to sepa- rately identify these different O&D categories (nonstop O&D, O&D with a connecting stop, and beyond O&D). In addition, some database vendors use the itinerary data to generate beyond O&D estimates for each market pair. A reduced level of effort may entail some simplifying assumptions, such as: ⢠All O&D passengers to short-haul destinations fly nonstop if nonstop service is available. ⢠The ratio of O&D passengers to total enplaned passengers to long-haul destinations is capped to ensure that it does not exceed 1.00. This could happen because the sum of O&D passengers (flying nonstop or connecting through another airport) can exceed the sum of enplaned pas- sengers flying nonstop to the same destination. ⢠All excess originating passengers (passengers traveling to markets without nonstop service and to long-haul markets in excess of the number of enplaned passengers) are assumed to be funneled through other airline hubs. The easiest method is to assume that each airlineâs originating passengers to enplaned pas- sengers ratio is the same for all markets. This method is appropriate for spoke airports with minimal connecting activity. Steps for Estimating Passengers by Flight Exhibit 6.7 shows a step-by-step approach for assigning passengers to a DDFS for a large con- necting hub airport. This approach involves the following steps and can be applied to both base year and future DDFSs. Steps for Estimating Passengers by Flight 1. Obtain load factors by airline for each market for the existing design day month. These data are available from the U.S. DOTâs T-100 database. 2. If the design day is intended to represent a specific day of the week, adjust the load factors collected in Step 1 to represent the design day of the week, if daily airport or airline data are available. 3. If airline or airport data are available, adjust the load factors in Step 2 for the time of day for both arrivals and departures. 4. Apply the load factors calculated in Steps 1, 2, and 3 to the available seats on each flight in the DDFS. 5. Normalize the results to ensure that the average load factor across the day (total daily enplaned passengers divided by total daily departing seats in the market) matches the daily average cal- culated in Step 2. Steps 3, 4, and 5 can be skipped if an airline operates only one daily flight to a market, which is often the case for international markets. 6. Estimate the existing ratio of originating passengers to enplaned passengers (ratio of ter- minating passengers to deplaned passengers should be similar) for each market and airline. These data are available from the U.S. DOTâs O&D Survey and T-100 data on a quarterly
How to Prepare a DDFS for Base Year and Future Conditions 63 and annual basis for U.S.-flag airlines. Some considerations are necessary when using these ratios because originating passengers to enplaned passengers ratios for a given flight will not always match market originating passengers to enplaned passengers ratios: a. Airlines flying to other hubs will often be carrying O&D passengers to beyond markets and a market originating passengers to enplaned passengers ratio will understate the on-board ratio. For example, an American Airlines flight leaving from Hartsfield-Jackson Atlanta International Airport (ATL) to Dallas Fort Worth International Airport (DFW) will be carrying O&D passengers from ATL to Phoenix Sky Harbor International Airport, ATL to Connecng Passenger Distribuon Data/ Assumpons Load Factor Data/ Forecasts Market Airline Time of Day Seats by Aircra Type Calculate Originang, Terminang, and Connecting Passengers by Flight DDFS (without Passengers) (see Exhibit 6.5) O&D Percentage Data/ Assumpons Market Airline Time of Day DDFS (with Passengers) Calculate Enplaned/ Deplaned Passengers by Flight Assumption/Input Data Input Intermediate Preparation Step Output Legend Exhibit 6.7. Estimating passengers by flight for DDFS.
64 Guidebook for Preparing and Using Airport Design Day Flight Schedules Tucson International Airport, and ATL to Albuquerque International Sunport, and so on, not just O&D passengers from ATL to DFW. For this reason, beyond O&D passengers need to be considered. Beyond O&D passenger data are available from some vendors, but for airlines that do not operate a hub at the airport under analysis, it is often simpler to apply an airport-wide originating passengers to enplaned passengers ratio rather than an individual market originating passengers to enplaned passengers ratio. b. Even for a hub airline, some flight itineraries include multiple stops. In these instances, the number of originating passengers for the one-stop market would have to be added to the non- stop market to estimate true on-board originating passengers to enplaned passengers ratios. c. In many long-haul markets, the market originating passengers to enplaned passengers ratio exceeds 1.00, which is mathematically impossible for a given flight. This occurs for various reasons, usually associated with price or schedule. Passengers will take an alternative con- necting flight rather than the nonstop flight to reach their destination. In these instances, it will be necessary to adjust the on-board originating passengers to enplaned passengers ratios to 1.00 or less. d. If resources permit, examine the full routing O&D passenger data to refine the on-board originating passengers to enplaned passengers ratios by market pair. If resources do not per- mit, it will be necessary to make an across-the-board adjustment to the individual on-board originating passengers to enplaned passengers ratios to ensure that the aggregate originating passengers to enplaned passengers ratio matches the overall airport originating passengers to enplaned passengers ratio. e. The O&D Survey database does not provide O&D passenger information for foreign-flag airlines. Those airlines would need to be surveyed to obtain information on their origi- nating passengers to enplaned passengers ratios. In general, some connecting passenger activity is associated with all international overseas flights. The connecting percentage is much higher for foreign-flag airlines that code-share or are in an alliance with a domestic airline, if any, that uses the airport under analysis as a connecting hub. 7. Apply any forecast changes in the originating passengers to enplaned passengers ratio from the annual forecasts to existing ratios calculated in Step 6 to estimate future originating pas- sengers to enplaned passengers and terminating passengers to deplaned passengers ratios for each market and airline combination. 8. Some judgment will be required to adjust the originating passengers to enplaned passengers and terminating passengers to deplaned passengers ratios by time of day and the factors fol- lowing should be considered: a. Unless red-eye flights are operated from South America, or from the West Coast, to East Coast airports, flights that depart prior to the first arrival bank will carry virtually no connecting passengers. Likewise, flights that arrive after the last departing bank will have virtually no connecting passengers. b. There should be a rough correlation between deplaning connecting passengers in a given arrival bank and enplaning connecting passengers in the succeeding departure bank. At no time should the number of cumulative daily enplaning connecting passengers exceed the cumulative number of deplaning connecting passengers. c. At international gateway airports, the connecting passenger percentage of total passengers typically peaks during the overseas international arrival and departure peaks as that is when the connecting opportunities peak. Estimating Passengers by FlightâObservations and Cautions The steps and guidance provided earlier apply mainly to U.S. airports and access to data such as the U.S. DOTâs T-100 database and O&D Survey data. At a few non-U.S. airports, more detailed data are collected from the airlines, which allows load factors and originating passengers
How to Prepare a DDFS for Base Year and Future Conditions 65 to enplaned passengers ratios to be more precisely defined by day of week and time of day. How- ever, at many airports, especially those in the developing world, this type of data is not available. In those instances, the preparer must rely on airline and passenger surveys, field observations, and professional judgment to estimate the number and segmentation of passengers on each flight. The U.S. DOT O&D Survey data provide a segmentation of originating passengers by point of origin for round trips. This information can be used to segment O&D information by resident and nonresident passengers. This segmentation has little effect on terminal requirements, but can have a major effect on landside requirements. As an example, almost all automobile parking demand is generated by resident passengers and almost all rental car demand is generated by nonresident passengers. If landside analysis is one of the purposes of the DDFS, an additional flight-by-flight segmentation of O&D passengers into resident and nonresident categories can be included. Note that this information will vary by market (which is available from the O&D Survey) and by time of day (which is only available from a passenger survey). Appendix A provides an example of how originating passengers to enplaned passengers ratios were estimated for an actual DDFS. Click here to access Section A.6 Passengers by Flight. 6.7 Nonscheduled Aircraft Operations DDFSs prepared for large airports tend to focus on scheduled passenger airline aircraft operations because they often represent the largest category of operations and they affect critical facilities, such as the terminal building, gates, and curbsides. However, if intended to address airfield issues, the DDFS is incomplete if it does not address nonscheduled opera- tions, including charter airline passenger, all-cargo airline, air taxi, GA, and military air- craft operations. Except for some all-cargo airline flights, these categories are nonscheduled by definition. In a DDFS, these categories are typically represented by a sample of daily activity that represents a typical distribution of operations. Irregular operations are scheduled operations that have been disrupted because of bad weather or other reasons. Step 4 in Section 6.5 discusses the use of buffer times or spare gates to account for these operations. Another option is to prepare a DDFS based on a past example of a significantly disrupted day at an airport, and use it to evaluate impacts on facilities and contingency plans. Charter Airline Passenger Aircraft Operations Data on charter airline passenger aircraft operations are typically available from airport sources or the U.S. DOTâs T-100 database, and can include airline, aircraft type, and market. If not available from airport sources, information on flight times can be obtained by interviewing the operators or accessing ASDI data. Without information provided by the operators, most preparers of DDFSs assume that the future distribution of charter operations will be similar to the current distribution. Load factors can be obtained from the U.S. DOTâs T-100 database, and passengers are almost always 100 percent O&D. All-Cargo Airline Aircraft Operations As discussed earlier regarding passenger data, approaches to developing the cargo elements of a DDFS have not been standardized. Exhibit 6.8 provides one approach for preparing DDFS cargo elements. The steps include the following.
66 Guidebook for Preparing and Using Airport Design Day Flight Schedules FLEET MIX AND MARKET ASSUMPTIONS FUTURE CARGO MIX INTEGRATED CARRIERS DOMESTIC NONINTEGRATED CARRIERS INTERNATIONAL NONINTEGRATED CARRIERS Cargo Schedules (if available) ESTIMATE FUTURE FLIGHT TIMES AND PAIR FLIGHTS Cargo DDFS (with Parking Assignments) Assign Parking Posions Assumption/Input Data Input Intermediate Preparation Step Output Legend Note: Items in BOLD CAPS are required for future DDFSs, but not for base year DDFSs. ANNUAL FORECASTS Exisng Distribuon of Cargo Flights by Time POLICY/PHYSICAL ASSUMPTIONS/CONSTRAINTS CAPACITY CONSTRAINTS POLICY FACTORS Current Parking Layout FUTURE PARKING LAYOUT Parking Assumpons Airline Assignments Aircra Capacity Domesc/Internaonal Buffer Times Exhibit 6.8. Forecasting all-cargo airline aircraft flights for DDFS.
How to Prepare a DDFS for Base Year and Future Conditions 67 Steps for Adding All-Cargo Airline Aircraft Operations to a DDFS 1. Collect relevant data. Some scheduled all-cargo aircraft operations data are still available on published schedule databases (but should be verified because wet-lease cargo operations, where one airlineâs aircraft and crew are leased to another airline, are sometimes missed). These data can be supplemented with data from the U.S. DOTâs T-100 database and ASDI data for existing flight times. For a DDFS, interviews with the all-cargo airlineâs local or head- quarters planning staff can be used to identify the airlineâs plans for the specific airport, related strategies for moving cargo across the airlineâs network, and the airlineâs âwindowâ of operations at its hubs. 2. Note that seasonal peaking characteristics for cargo are different than for passengers. Cargo demand tends to peak in the fall, often in October or December, in response to holiday retail sales demand. If the focus of the DDFS analysis is cargo, the design day should be adjusted accordingly. 3. Segment cargo operations by category, if not available from the annual forecasts. This is impor- tant because integrated (express) carriers, other U.S.-flag airlines, and other foreign-flag airlines have differing characteristics that affect flight times, markets served, and fleet mix. 4. Determine new markets and growth in existing markets. Unless the annual forecasts contain a detailed market forecast or the airlines provide information, this will largely be a matter of judgment. 5. New flight times for express carrier operations should be selected with care. These carriers need to schedule to their sort hub operations at a centralized facility, which means there is little opportunity for peak spreading or other adjustments. Non-express all-cargo airlines have much more leeway in selecting flight times and, therefore, they tend to operate on a much less regular schedule. Overseas international airlines often experience nighttime curfews at destination airports, which can constrain their windows of operation. 6. At major sort hubs, turnaround times are determined by the sort time required. At out- stations, turnaround time minimums may be determined by brake cooling time, as widebody aircraft carry substantial loads and a cooldown period for tires and brakes of up to 45 minutes may be required. 7. Assigning parking positions to cargo airlines is similar to the passenger airline gating approaches. If cargo facilities are not the focus of the analysis, assigning cargo aircraft to general parking locations, rather than specific parking positions, is adequate. Other Nonscheduled Aircraft Activity Exhibit 6.9 provides a general approach to forecasting the air taxi, GA, and military elements of a DDFS. Air Taxi Aircraft Operations Air taxi aircraft operations are conducted by small for-hire commercial aircraft governed by 14 CFR Part 135. The fleet mix is similar to GA aircraft. Detailed information on air taxi opera- tions is scarce and typically available only through interview with air taxi operators or from ASDI databases. Similar to charter operations, without information provided by the operators, most preparers of DDFSs assume that the future distribution of air taxi operations will be similar to the current distribution. FAA ATCT counts include scheduled regional airline operations using aircraft with fewer than 60 seats and some air cargo feeder flights with air taxi counts, so the two categories need to be distinguished when preparing a DDFS.
68 Guidebook for Preparing and Using Airport Design Day Flight Schedules FLEET MIX AND MARKET ASSUMPTIONS ESTIMATE FUTURE FLIGHT TIMES Air Taxi, General Aviaon, and Military DDFS (with Parking Assignments) Assign Parking Area Assumption/Input Data Input Intermediate Preparation Step Output Legend Note: Items in BOLD CAPS are required for future DDFSs, but not for base year DDFSs. ANNUAL OPERATIONS FORECASTS AIR TAXI GENERAL AVIATION MILITARY Exisng Distribuon of Operaons by Time Air Taxi General Aviaon Military POLICY/PHYSICAL ASSUMPTIONS/ CONSTRAINTS CAPACITY CONSTRAINTS POLICY FACTORS Current Parking Facilies FUTURE PARKING FACILITIES PEAK SPREADING ASSUMPTIONS Exhibit 6.9. Forecasting DDFS air taxi, GA, and military aircraft activity.
How to Prepare a DDFS for Base Year and Future Conditions 69 GA Aircraft Operations The available data for GA activity differ depending on whether the operators are flying under instrument flight rules (IFR) or visual flight rules (VFR). Most IFR operations appear in the ASDI database, but VFR operations do not. The FAAâs Distributed OPSNET database provides hourly estimates of all GA aircraft operations, but it is assumed in the estimates that the hourly distribution of VFR operations is the same as for the distribution of IFR operations. Informa- tion on the true hourly distribution of VFR operations is typically available only through ATCT counts or through independent counts or surveys. As a practical matter, DDFSs are very seldom needed or prepared for airports that primar- ily serve GA. Most DDFSs are prepared for large commercial-service airports, where GA accounts for a small percentage of activity. When DDFSs are prepared for airfield analy- sis, it is important to account for the GA component. However, most GA aircraft at large commercial-service airports operate under IFR flight plans, and are, therefore, captured in the ASDI databases. Without information provided by GA aircraft operators, most preparers of DDFSs assume that the future distribution of GA aircraft operations will be similar to the current distribution. Military Aircraft Operations Secondary information on military aircraft operations is probably less reliable than such infor- mation on any other category of aircraft operations because vital information is often deleted from the ASDI-sourced databases. The FAAâs Distributed OPSNET database provides hourly counts of military aircraft operations for airports with an ATCT, but because flight times are removed for many flights, they default to 0 and appear as a spike in the hour between midnight and 0100. Before using the FAAâs Distributed OPSNET data, preparers should make an adjustment for this spike to avoid over forecasting nighttime operations. Most preparers of DDFSs assume that the future distribution of military operations will be similar to the current distribution. However, if military activity accounts for a significant portion of an airportâs operations, the military commander should be consulted to determine if there are any upcoming changes in mission, as such changes often significantly alter the number of operations, fleet mix, and main hours of operation. 6.8 Application of Constraints One of the purposes of DDFS analysis is to determine future facility requirements. Therefore, in many cases, a DDFS is based on an unconstrained forecast to ensure that demand is properly modeled. In other cases, however, constraints cannot be realistically eliminated because of limits in physical space, lack of funding, policy restrictions, or political opposition. If that is the case, the DDFS must incorporate the constraint. Airport constraints can take many forms, including: ⢠Airfield constraints, such as the lack of runways, taxiways, or queuing space ⢠Terminal building constraints, such as the lack of gates or other passenger processing facilities ⢠Landside constraints, such as the lack of curb space, access roads, or parking ⢠Policy constraints, such as slot restrictions or nighttime noise restrictions Typically, the constraint on overall activity is addressed in the annual forecasts. However, as the effect of constraints is most evident during peak periods, constraints must be specifically addressed in the DDFS. The approach used to incorporate constraints into DDFSs depends on whether the constraint is physical, financial, policy-related, or other. As the effect of constraints is most evident during peak periods, con- straints must be specifically addressed in the DDFS.
70 Guidebook for Preparing and Using Airport Design Day Flight Schedules Airspace/Airfield Constraints The effects of airspace/airfield constraints are often measured in terms of hourly throughput capacity. ACRP Report 79: Evaluating Airfield Capacity http://onlinepubs.trb.org/onlinepubs/ acrp/acrp_rpt_079.pdf provides guidance on ways to measure throughput capacity. Chen and Gulding identified a relationship between scheduled demand and throughput capacity in which it was recognized that scheduled demand can exceed capacity over short periods of time, but not over longer periods of time. This relationship is summarized in Table 6.1 and discussed in more detail in the Chen and Gulding paper, as well as in ACRP Report 82: Prepar- ing Peak Period and Operational Profiles â Guidebook http://onlinepubs.trb.org/onlinepubs/ acrp/acrp_rpt_082.pdf. These relationships can be used as controls when preparing a DDFS. If the estimated num- ber of flights exceeds capacity during the first DDFS pass, preparers should move or elimi- nate flights consistent with airline strategies. Generally, airlines prioritize their high-revenue operations, typically international or long-haul domestic flights, at the expense of low-revenue operations, typically short-haul commuter flights. Terminal Building Constraints The effect of a terminal building constraint on a DDFS will differ depending on if the constraint affects aircraft (gates) or passengers (passenger processing facilities). The number of aircraft departures per gate often increases with airport size. Large airports serving large markets can support more flight frequencies that generate higher gate utilization. Very large airports such as ATL and Chicago OâHare International Airport accommodate up to eight departures per gate per day. Most other busy airports accommodate six departures per gate per day, and smaller commercial airports accommodate three to four turns per gate per day. Maximum turns per gate will also depend on the airline. Gate utilization by network airlines will often be constrained by their connecting bank structure at either the origin or destination airport. Low-fare airlines have more flexibility and can sometimes average up to 10 departures per gate per day or more. If the DDFS needs to be adjusted to accommodate gate constraints, the adjustment should reflect airline priorities (i.e., low-revenue flights should be moved or eliminated first). Non-gate terminal constraints affect passengers directly and airlines indirectly. Passengers will usually adjust to potential terminal bottlenecks by increasing lead times, and airlines gen- erally are not compelled to modify their schedules. If non-gate terminal constraints become extremely onerous, passengers will either fly less or use another airport, but this factor is ideally addressed in the annual forecasts rather than the DDFS. Peak Period Definion Maximum Demand/Capacity Rao (Aircra Operations) 15 Minutes 1.41 1 Hour 1.21 2 Hours 1.14 3 Hours 1.06 Source: ACRP Report 82, Preparing Peak Period and Opera onal Profiles - Guidebook Exhibit 9.1 âMaximum demand/capacity raÂo by peak period definiÂonâ (2013). hÂp://onlinepubs.trb.org/onlinepubs/acrp/acrp_rpt_082.pdf Table 6.1. Maximum scheduled demand by peak period definition.
How to Prepare a DDFS for Base Year and Future Conditions 71 Landside Constraints The effect of landside constraints on DDFS preparation is similar to the effect of non-gate terminal building constraints. These constraints directly affect passengers and passengers adjust, to a point, by increasing lead times or changing to a ground transportation mode. If the landside constraints become extremely onerous, passengers will either fly less or use another airport, but again, this factor is ideally addressed in the annual forecasts rather than the DDFS. Policy Constraints As practiced in the United States, policy constrains, such as slot restrictions, perimeter rules, or nighttime noise restrictions, essentially dictate the number, type, and timing of aircraft opera- tions at a given airport. Some of these policies, such as the slot controls at the three main New York airports and Washington Reagan National Airport are imposed by the FAA, whereas others, such as the Perimeter Rule at New York LaGuardia Airport, are locally imposed. DDFS preparers must estimate the composition (markets and fleet mix) of these aircraft operations. Examination of constrained airports indicates that airlines will take the following actions, in order or priority, when faced with a constraint (Kennon et al. 2013): 1. Increase fares to take advantage of reduced competition and to cover increased operating costs. 2. Reschedule some flights to off-peak hours, subject to market constraints. 3. Increase the size of the aircraft serving a market, provided the right-sized aircraft is in the airlineâs fleet. 4. Eliminate or reduce service to some markets. In general, load factors are not higher at constrained airports than at unconstrained airports. When slots are subject to a use or lose it provision, some airlines will operate unprofitable flights to retain a slot in anticipation of preserving it for a profitable flight in the future. 6.9 DDFS Updates Updating a DDFS and retaining the original level of detail and fidelity are not simple. Even when the change just involves the entry of a new airline or addition of a new market, competitive responses and adjustments by existing airlines mean that the changes may reverberate throughout a significant portion of the schedule. Often, a new DDFS must be prepared to represent the changed conditions. As an alternative to preparing a new DDFS, tools are built into a number of simulation models that provide user input for rule-based, randomized cloning of activity. These models are used by several airport operators and consultants analyzing near-term issues or conducting sensitiv- ity analyses. Cloning also can be used to provide a sensitivity analysis of DDFS-based modeling results. However, these approaches lack fidelity and granularity and are limited in certain areas, such as determining gate assignments. Appendix A includes discussion of how some elements of a DDFS were updated as project requirements changed. Click here to access Section A.9 DDFS Updates. 6.10 Quality Assurance and Control DDFSs are inherently detailed and their preparation involves substantial individual judgment in the selection of new markets, flight times, arrival/departure links, and other elements, which are all inter-related. As a result, there is a potential for both error and bias and quality assurance is an important consideration.
72 Guidebook for Preparing and Using Airport Design Day Flight Schedules Some consultants use proprietary spreadsheet programs to help reduce and analyze the data, check DDFS control totals, ensure that numbers of arriving and departing connecting passen- gers are balanced, and help the gate assignment process. Some consultants also have internal broad guidelines for preparing DDFSs, such as assumptions regarding load factor triggers for up-gauging aircraft or adding flights. For these consultants, it is critical that the rules used to prepare future schedules are checked at every stage, including physical and other constraints. Exhibit 6.4 in ACRP Report 82: Preparing Peak Period and Operational ProfilesâGuidebook http://onlinepubs.trb.org/onlinepubs/acrp/acrp_rpt_082.pdf provides a detailed checklist of recommended quality control checks for DDFSs, which is reproduced and supplemented as Appendix E in this guidebook. This micro-checklist should be coupled with a macro-checklist to ensure that the passenger and aircraft operation distributions resulting from the DDFS are reasonable. At the same time, care must be taken to ensure that insights resulting from the bottom-up DDFS approach are not suppressed by the imposition of strict control totals. One way to achieve this balance would be to adopt an approach similar to the FAAâs approach for determining consistency between an airport operatorâs forecast and the TAF. With the TAF, a deviation of more than 10 or 15 percent does not result in automatic rejection; instead, it trig- gers a more in-depth investigation into the differences. With this strategy, if a DDFS bottom-up approach results in hourly distributions that differ significantly from existing distributions, it would trigger a detailed review, but not an automatic rejection. Table 6.2 lists suggested toler- ance bounds for DDFS peak hour operation results based on historical deviations of peak hour percentages at a sample of airports of differing sizes. The percentages should be applied to the peak hour operations from the annual forecasts. For example, if the peak hour forecast was 120 operations for a large-hub airport, the lower bound would be 114 operations (95 percent à 120) and the upper bound would be 126 operations (105 percent à 120). Appendix A provides an example of the application of quality controls in the preparation of a DDFS. Click here to access Section A.10 Quality Assurance and Control. FAA Hub Category Lower Bound Peak Hour Target Upper Bound Large Hubs 95% of target 100% 105% of target Medium Hubs 92% of target 100% 108% of target Small Hubs 89% of target 100% 111% of target Non-hubs 87% of target 100% 113% of target Sources: Oliver Wyman and HNTB analyses of Official Airline Guide schedule data. Table 6.2. Tolerance bounds for DDFS peak hour operation results.
