Preparing Peak Period and Operational Profiles—Guidebook (2013)

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.

5 This chapter describes the relationship between annual forecasts, peak period estimates, and operational profiles. The chapter also provides some key definitions. When evaluating current conditions, direct measures of passenger boarding activity are generally not available in increments of less than one month. In addition, airport forecasts are often only prepared on an annual basis. Therefore, peak period estimates and operational profiles are usually necessarily derived from annual activity. This chapter provides a brief overview of the annual forecasting process and the relationship between annual activity, the design day, the three types of operational profiles, and peak period estimates. Additional detail on operational profiles and peak period estimates is also provided. C h a p t e r 2 Background and Key Definitions Key components of airport activity forecasts include passenger enplanements, passenger originations, air cargo tonnage, and aircraft operations. Each time a person boards an aircraft, he or she is counted as a passenger enplane- ment. Each time they disembark, they are counted as a deplanement. Passengers who begin the air portion of their trip at an airport are counted as an originating passenger at that airport. If they end the air portion of their trip at an airport, they are counted as a terminating passenger. Combined originating and terminating passengers are often referred to as O&D passengers. A passenger who transfers from one aircraft to another is counted as a connecting passenger. Air cargo includes air freight and air mail. As a practical matter, the distinction between the two is becoming increasingly blurred and many carriers are ceasing to distinguish between the two. Each aircraft takeoff is counted as an aircraft operation and each aircraft landing is counted as an aircraft operation. 2.1 Annual Forecasts There are four broad areas of airport activity: passenger, cargo, general aviation/for-hire air taxi, and military. The for-hire air taxi category is discussed in more detail in Section 2.1.3. The forecast drivers and the available data differ from category to category, and therefore the typical forecast approaches also differ.

6 preparing peak period and Operational profiles—Guidebook 2.1.1 Passenger Forecasts Most passengers fly on scheduled carrier flights, flights that operate at times and on routes that are determined and published well in advance. There are three main ways of projecting scheduled passenger carrier activity; in order of increasing complexity, they are trend analysis, share analysis, and regression analysis. Trend analysis consists of calculating the growth rate over a historical period of time and projecting that growth rate to continue into the future. This implicitly assumes that the factors that drove passenger growth in the past, such as income growth or changes in fares, will be the same in the future. Also, the resulting forecasts can be very sensitive to the period of time selected to calculate the historical growth rates. Share analysis involves calculating passengers as a constant, increasing, or decreasing share of passengers in a regional or national forecast. This is an effective way of capturing anticipated changes in industry trends, assuming they have been captured in the national forecast, but it is not ideal for incorporating any local factors that diverge from national trends. Regression analysis is a statistical method of generating an equation (or model) which best explains the historical relationship among selected variables, such as passenger enplanements and income. This approach allows the user to assess the impact of changes in forecast drivers, such as the local economy or airfares, upon future passenger enplanements. For this approach to work effectively, accurate projections of the forecast drivers are necessary. These methods work best for local (origin and destination) passengers and airports where most of the passengers are local. Connecting passengers are driven more by the hub carrier’s routing decisions than by the local economy. When information from the airline(s) is unavailable, the typical approach is to estimate connections as a constant, or an increasing or declining percentage of local passengers. Passenger aircraft operations are usually estimated as a function of the passenger forecast instead of being projected independently. Typically, forecasts of passengers per operation are estimated based on assumptions for average seats per aircraft and load factor. In this instance, load factor represents the percentage of aircraft seats that are filled by passengers. These projec- tions are then divided into the passenger projections to generate a forecast of passenger aircraft operations. 2.1.2 Air Cargo Forecasts Air cargo includes both air freight and air mail. The approaches used to prepare annual cargo forecasts—trend, share, and regression analysis—are similar to those used to prepare passenger forecasts but are more difficult to apply for several reasons. First, although air cargo is time- sensitive, it is not as time-sensitive as passenger travel. Consequently, air cargo is more subject to competition from substitutes, either other modes of delivery, such as trucking, or other airports. Secondly, some types of data, like pricing, are much scarcer for air cargo than for passengers. Finally, the air cargo sector has evolved significantly over the past 30 years with the advent of door-to-door service by integrated carriers such as FedEx and UPS. This means that much of historical air cargo growth has been driven by service improvements, in addition to traditional drivers like economic growth and price. Compared to passenger forecasts, the number of variables that need to be considered to forecast air cargo is greater, and the information available about those variables is less. Many forecasters therefore use share analysis, which relies on industry air cargo forecasts by the FAA or aircraft manufacturers that have the resources to analyze the myriad of issues.

Background and Key Definitions 7 Forecasts of air cargo aircraft operations are also more complex than the forecasts of passenger aircraft operations. First, cargo tonnage must be segmented between belly and all-cargo activity. Belly cargo, named because it is carried in the bellies of passenger aircraft, does not generate additional operations. All-cargo aircraft operations are usually estimated by projecting factors for average aircraft tonnage capacity and load factor. These factors are then used to estimate future tons per aircraft operation. Finally the forecast of total non-belly cargo is divided by the forecast of tons per operation to generate a forecast of all-cargo aircraft operations. 2.1.3 General Aviation and For-Hire Air Taxi Forecasts General aviation operations include all operations other than commercial or military operations. For forecasting purposes, for-hire air taxi operations are often included with general aviation since the two categories share similar aircraft and data availability characteristics. The FAA defines an air taxi as “an aircraft designed to have a maximum seating capacity of 60 seats or less or a maximum payload capacity of 18,000 pounds or less carrying passengers or cargo for hire or compensation.”i Many of these air taxi operations are commercially scheduled passenger and cargo operations. The remaining air taxi operations include small aircraft hired for a specific purpose, as distinguished from regularly scheduled flights. Data for these for-hire operations are much scarcer than for scheduled commercial operations, so they are typically included with general aviation or forecast separately. At many smaller airports, the general aviation/for-hire air taxi category is the only relevant category. The methods used to forecast general aviation and for-hire air taxi operations are similar to those used for commercial operations: trend, share, and regression analysis. Most general aviation activity is discretionary so the link with economic growth is weaker than is the case with commercial aviation and therefore, regression analysis is less effective. Because of this, many forecasters choose share analysis which relies on industry forecasts by the FAA or general aviation aircraft manufacturers who have the resources to analyze the business, social, and technological trends driving general aviation. 2.1.4 Military Operations Forecasts Military aviation activity is determined by external events and policy factors, which are very difficult to forecast in the long term. Standard practice is to assume future military operations will remain at base year levels, unless information on a change in mission is available. 2.1.5 Application of Annual Forecasts to Operational Profiles and Peak Period Forecasts Planners are often asked to prepare an annual forecast prior to or along with the peak period estimate or operational profile, or are directed to use a specific annual forecast as a starting point. In some instances, the choice of which annual forecast to use is left to the planner. If more than one forecast is available, the key selection factors are anticipated accuracy and level of detail. Some relevant factors are the level of effort devoted to the forecast, the amount of scrutiny and review to which it was exposed, how recently it was prepared, and how well it tracks current activity and recent trends. The amount of detail available in the annual forecast is also important. Specifically, a fleet mix forecast is required to prepare day/night splits or design day flight schedules. Some of the types of annual forecasts typically available include master plans, system plans, and the FAA’s TAF.

8 preparing peak period and Operational profiles—Guidebook Master Plans Airport master plans provide a roadmap for long-term airport development, typically over a period of 20 years. A master plan forecast generally provides more detail than the other sources and, if current, is usually the best choice for an annual forecast. Most master plan forecasts include a fleet mix forecast, which, when combined with base year day/night distributions and schedules, form a solid foundation from which to prepare future day/night splits and design day schedules. System Plans System plans involve the planning and prioritization of a regional system of airports, typically on a statewide basis. System plan forecasts are typically carried out to much less detail than master plans, and aside from critical aircraft information, do not provide much data on fleet mix. In addition, they tend to apply uniform forecast methodologies that may be optimal for the system but not necessarily for the specific airport in question. FAA’s Terminal Area Forecast The FAA’s TAF is revised yearly and is used to help determine FAA staffing levels, prioritize FAA capital spending, and to validate independent airport sponsored forecasts. Since the TAF is revised yearly for each airport, it is usually the most current alternative. It provides no fleet mix information, but does provide forecasts of both passengers and operations for air carriers and commuter/air taxis. This information, along with an estimate or assumption for load factor, can be used to estimate average seats per operation, which in turn can serve as a control on a future fleet mix estimate. Detailed direction on the preparation of annual forecasts is beyond the scope of this Guidebook, but useful guidance on forecast approaches can be found in the following documents: • FAA, Airport Master Plans, Advisory Circular No: 150/5070-6B. • De Neufville, R. and A. Odoni, Airport Systems: Planning, Design and Management • GRA, Inc. for FAA, Forecasting Aviation Activity by Airport • Transportation Research Circular E-C040, Aviation Demand Forecasting: A Survey of Method- ologies, TRB, National Research Council, Washington, DC, 2002 • William Spitz and Richard Golaszewski, ACRP Synthesis 2: Airport Aviation Activity Forecasting, TRB, National Research Council, Washington, DC, 2007 • ICAO, Airport Planning Manual, Doc 9184-AN/902 Part 1, International Civil Aviation Organization Note that design day, operational profiles, and peak period estimates can be calculated for existing conditions as well as future conditions. In many instances, existing measures of design day or peak period activity are needed to estimate current facility requirements, calibrate planning factors, or identify current environmental impacts. Measures of base year or forecast annual passengers and operations provide the foundation for estimating design day activity, which in turn provides the basis for estimating operational profiles and peak period activity levels. These elements are described in more detail. 2.2 Design Day The design day activity level is the level that airport planners use in sizing facilities and typically represents the level of activity that can be accommodated with an acceptable level of service. The intent is to strike a balance between under-designing, in which case the facility in question would perform at substandard levels of service too often in the view of airport stakeholders, and over-

Background and Key Definitions 9 designing, in which case the cost of the facility (again in the opinion of the airport stakeholders) would be too high to justify the percentage of time during which the facility performs at or above an acceptable level of service. Design day passengers are the total number of passengers during the design day, and design day operations are the total number of aircraft operations during the design day. The design day is derived from annual activity following the process in Exhibit 2.1. The following definitions apply to the exhibit: Monthly and weekly distributions represent the distribution of annual passengers and operations by month and the share of weekly passengers and operations by day-of-the-week. The user-defined threshold represents the percentage of days in the year in which passengers or operations will exceed those of the design day. For example, if the user chooses a 10 percent threshold, design day activity levels will be exceeded 10 percent of the time, or on 36 days during the year. Peak spreading factors are user-determined assumptions regarding the percentage that the peak month or design day activity (as a percentage of annual activity) will decline over the forecast period. For example, a peak spreading factor of -5.0 percent means that a peak month percentage that is currently 10 percent of annual activity would fall to 9.5 percent [10% × (100%-5%)]. The most common current practice in the United States is to define the design day as an ADPM or peak month average day (PMAD). This is calculated by identifying the month with the highest number of operations and passengers, and then dividing the operations or passengers in that month by the number of days in the month. There are several other design day definitions in use, however, especially outside of the United States. Examples include the following: • The average week day in the peak month (AWDPM) • The 15th busiest day of the year • The 30th busiest day of the year • The 90th percentile—corresponds to the 36th busiest day of the year Although the ADPM definition is less precise than most of the alternatives, it has found favor because it requires less data and effort to calculate, especially for passenger activity. The disadvantage of the ADPM method is that it can generate very different design day thresholds Annual Activity Forecast User-Defined Threshold Design Day Forecast Monthly and Weekly Distributions Peak Spreading Factor Data Input User Determined Assumptions Intermediate Output Final Output Exhibit 2.1. Relationship between annual activity forecasts and design day forecasts.

10 preparing peak period and Operational profiles—Guidebook from airport to airport. For example, at an airport with high seasonality (i.e., the peak month accounts for a relatively high percentage of annual activity), the ADPM design day will represent a high design day threshold corresponding to the 20th or 15th busiest day of the year. Conversely, at an airport with low seasonality, especially one with some day of the week variation in activity, the ADPM design day will represent a low design day threshold corresponding to the 100th or 150th busiest day of the year. Thus, use of the ADPM design day definition can result in facilities with very different service levels depending on the airport. The appropriate balance between over-design and under-design may differ depending on the type of facility. This Guidebook provides the tools to customize the design day definition for the specific facility in question (see Chapter 4). Although the FAA accepts the ADPM and AWDPM design day definitions for planning, other definitions are not precluded. For most environmental analysis, including noise analysis and air quality emissions inventories, the design day is defined as an average annual day (AAD), which is annual activity divided by the number of days in the year.ii In a minority of cases, such as State Implementation Plans (SIPS) prepared to show compliance with the Clean Air Act, the AWDPM is an accepted standard for airport air quality dispersion analysis.iii The user should consult FAA and Environmental Protection Agency (EPA) guidance concerning acceptable design day definitions, especially when preparing National Environmental Policy Act (NEPA) documents. 2.3 Design Day Profiles Design day profiles show arriving and departing passengers or aircraft operations by time of day, in increments of an hour or less. They are often calculated and presented graphically using rolling averages. Exhibit 2.2 shows examples of design day profiles for arriving and departing passengers at a connecting hub airport example. Design day profiles provide a measure of detail, useful for planning facilities, that is not available from peak period estimates. Many facility requirements (departure curb, ticketing, and security) are dependent on lead time, or the interval between the time an enplaning passenger arrives at a given facility and the time his or her flight departs the gate, while other facility requirements (baggage claim and customs) are dependent on lag time, or the interval between the time an aircraft arrives at a gate and the average time a deplaning passenger arrives at a given airport facility. Other facilities (restrooms, concessions) are dependent on a combination of the arriving and departing passenger flows. The peaks that emerge from these “upstream” and “downstream” passenger flows will not necessarily match the enplaning and deplaning peaks. It is much easier to estimate these derivative or second-order peaks from passenger profiles showing activity by time of day than from peak period estimates. In addition, the ability of some facilities to handle peak loads will depend on whether a queue already exists prior to the peak, which in turn depends on the level of activity prior to the peak. Design day profiles provide planners with the information needed to evaluate these issues. Exhibit 2.3 describes conceptually how design day profiles are derived from design day forecasts. The existing daily distribution is usually taken from FAA tower data or radar data for operations and from airline schedules for passenger distributions. Typically, load factor estimates are applied to seat arrivals and departures taken from the airline schedules to arrive at an estimate of passenger arrivals and departures by time of day. There are several ways of estimating future design day profiles. The simplest way is to assume that the base year distribution of daily activity will carry forward unchanged into the future. A second alternative is to assume a peak spreading component based on relationships between airport size and peak period percentage. This dampens the peaks and fills in the gaps in the daily

Background and Key Definitions 11 Exhibit 2.2. Design day passenger profiles—connecting hub airport example. Source: HNTB extraction of hourly passenger flow data from gated flight schedule prepared for typical connecting hub airport. Arriving Passengers—Deplanements and Terminations Departing Passengers—Enplanements and Originations 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 0 0 : 0 0 0 : 1 0 0 : 2 0 0 : 3 0 0 : 4 0 0 : 5 0 0 : 6 0 0 : 7 0 0 : 8 0 0 : 9 0 0 : 0 1 0 0 : 1 1 0 0 : 2 1 0 0 : 3 1 0 0 : 4 1 0 0 : 5 1 0 0 : 6 1 0 0 : 7 1 0 0 : 8 1 0 0 : 9 1 0 0 : 0 2 0 0 : 1 2 0 0 : 2 2 0 0 : 3 2 Depl. Term. 0 2000 4000 6000 8000 10000 12000 0 0 : 0 6 0 : 1 2 1 : 2 8 1 : 3 4 2 : 4 0 3 : 5 6 3 : 6 2 4 : 7 8 4 : 8 4 5 : 9 0 0 : 1 1 6 0 : 2 1 2 1 : 3 1 8 1 : 4 1 4 2 : 5 1 0 3 : 6 1 6 3 : 7 1 2 4 : 8 1 8 4 : 9 1 4 5 : 0 2 0 0 : 2 2 6 0 : 3 2 Enpl. Orig.

12 preparing peak period and Operational profiles—Guidebook schedule. A third alternative is to generate daily profiles by category of activity (i.e., domestic and international passengers), project each profile to grow at the annual rate of the correspond- ing activity category, and then aggregate the results to generate an estimated future daily profile. The fourth alternative is to aggregate a daily profile from a design day schedule (see Section 2.4 for more details). 2.4 Design Day Schedules The highest level of detail is provided in design day schedules. These schedules go by names such as event files, gated flight schedules, or hypothetical design day activity. They are intended to represent a snapshot of future activity at an airport or airport system on a flight-by-flight basis. The format of these schedules depends on their intended use. Design day flight schedules serve as input files for SIMMOD, TAAM, and other airfield simu- lation models. They include separate records for each flight, which detail airline, aircraft type, flight time, and origin or destination (O&D). When used for terminal analysis these schedules also include passenger loads, broken down by O&D and connecting passengers. The FAA uses modified versions of these schedules for national airspace planning. Design day schedules are also used for some types of environmental analysis. Air quality dis- persion analysis requires most of the information needed for airfield planning to model aircraft emissions, and typically needs measures of local (non-connecting) passenger activity to help model ground vehicle movements. One type of noise modeling, the Noise Integrated Routing System (NIRS) model, requires most of the airfield components of a design day schedule but does not require the passenger information. More detailed information on the appropriate tools and forecasts for planning and environmental analysis is provided in Chapter 3. The benefits of the design day schedule approach are (a) it provides the level of detail required to examine complex airspace and airfield operational issues, and (b), numerous terminal concepts involving alternative airline allocation scenarios can be quickly analyzed, since the forecasts are disaggregated down to the individual flight level. A disadvantage of the approach, in addition to the cost, is that it does not lend itself well to forecast-related sensitivity analysis due to the effort involved in preparing design day schedules for alternative forecast scenarios. Exhibit 2.4 shows conceptually how design day schedules are derived from design day forecasts. The design day forecast provides control totals for passengers and aircraft operations. Typically, a design day schedule is prepared by modifying an existing schedule to include assumptions on Design Day Forecast Design Day Profile Peak Spreading Factor Existing Daily Distribution Data Input User Determined Assumptions Intermediate Output Final Output Exhibit 2.3. Relationship between design day forecasts and design day profiles.

Background and Key Definitions 13 new markets, additional frequencies, and fleet mix changes. In some instances daily profiles are derived from design day schedules. In other instances, previously derived daily profiles are used to guide the addition of flights for future design day schedules. See Chapter 6 for additional detail. 2.5 INM Input Profiles The FAA requires noise analyses to be performed for an AAD with separate weightings for daytime (7 AM to 10 PM) and nighttime (10 PM to 7 AM) flights. The California State Department of Health Services (DOHS) requires three separate weightings for noise studies in that State, with evening (7 PM to 10 PM) also included. Exhibit 2.5 shows the relationship between the design day (defined as AAD for noise analysis), and the inputs used to generate day/night splits and stage length estimates. In addition to the distribution of daytime and nighttime aircraft arrivals and departures, the day/night forecast must provide AAD aircraft operations by individual aircraft type and aircraft departures by stage length. Aircraft type is a major determinant of the noise impact. Stage length Design Day Forecast Design Day Schedule New Flight Assumptions Design Day Profile Existing Schedule Data Input User Determined Assumptions Intermediate Output Final Output Exhibit 2.4. Relationship between design day forecasts and design day schedules. Average Annual Day Forecasts Forecast INM Input Profile Design Day Schedule Fleet Mix/Stage Length Forecast Changes in Day/Night Distributions Existing Day/Night Distributions by Category Data Input User Determined Assumptions Intermediate Output Final Output Exhibit 2.5. Relationship between average annual day forecasts and INM input forecasts.

14 preparing peak period and Operational profiles—Guidebook represents the distance to the destination market and determines how much fuel an aircraft must carry. The amount of fuel then determines aircraft weight, which determines the amount of power (and noise) that the aircraft must generate to take off as well as its rate of climb. Often, the current practice is to maintain the current day/night split in each major category (e.g., passenger, cargo, and general aviation). If the relative contribution of each category to overall airport activity shifts over time, the overall day/night split will change. If not, it will remain constant. The assumption of a constant day/night split within each category should be evaluated. Airline schedule changes, especially those that affect the organization of connecting banks, have a major impact on the day/night distribution. A connecting bank occurs when an airline schedules a large number of flights that arrive within a short period of time, discharge passengers that then enplane onto other aircraft, after which the same aircraft depart, again within a short period of time. In addition, nighttime flights tend to be less lucrative for the airlines, since passengers are less inclined to fly at those times. Consequently, the percentage of nighttime flights tends to increase when the economy is strong and decrease when the economy is weak. See Chapter 7 and Appendix I of the ACRP WOD 14 for more detail. The resources devoted to identifying future changes in day/night split should be based on an assessment of the potential changes to the noise impacts. For example, if there are a large num- ber of older heavy cargo aircraft on the cusp of day and night, further investigation, including discussions with the air carrier may be warranted. If, alternatively, the potential shift is likely to involve piston-powered general aviation (GA) aircraft, the effect on the noise impacts would be much smaller, and less analysis would be warranted. In some cases, the output of simulation models is used for noise analysis. In those instances, forecast day/night splits reflect the additional fidelity associated with the future schedule design effort. 2.6 Peak Period It is important to distinguish between the peak period definition and the peak period threshold. The peak period definition is the amount of time the peak period lasts, whether 15 minutes, 20 minutes, 30 minutes, one hour, or more. The peak period threshold represents the percentage of time during the year when the peak period activity is exceeded, whether five percent, 10 percent, or some other percentage. There is no single correct number for either the peak period definition or threshold. These may differ depending on the facility under analysis and the planner’s needs and judgment. The design day peak period is not, and should not be, the absolute highest peak period. In general, passenger activity during the absolute peak hour is about 20 percent higher than the design peak hour.iv In many instances the design day calculation described in Section 2.2 is just an intermediate step towards the calculation of the peak period, often defined as the peak hour. In most master plan forecasts, there is an assumption that the peak period occurs during the design day. As is the case with the appropriate design day definition, the definition of the peak period may depend on the type of facility being planned.v Facilities that are prone to breakdown or gridlock at high activity levels, as opposed to degradation of service, may necessitate a stricter peak period definition. Usually the peak period is derived from the rolling peak in the design day profile. If the forecast includes the construction of future design day schedules, peak period activity can be derived from those schedules. In those instances, peak spreading emerges as a result of filling in off-peak flights in the schedule construction process.

Background and Key Definitions 15 Exhibit 2.6 shows the relationship between the peak period and the design day (Section 2.2), and the design day profile (Section 2.3) or design day schedule (Section 2.4). The peak period percentage of the busy day tends to be lower at large airports than at small airports and should be expected to decline as an airport becomes busier, a peak spreading phenomenon similar to the monthly peak spreading described in Section 2.2. The context in which the peak period occurs is important. If the time immediately preceding is also very busy, inherited queues and other activity may further exacerbate stresses occurring in the peak period. Likewise, if the succeeding period is busy, the ability of the facility to recover from the stresses of the peak will be impeded. Design day profiles (see Section 2.2) are useful for identifying these issues. Design Day Forecast Design Day Profile Design Day Schedule User Defined Threshold User Definition of Peak Period Peak Period Data Input User Determined Assumptions Intermediate Output Final Output Exhibit 2.6. Relationship between design day profiles and schedules and peak period forecasts.