View
222
Download
0
Category
Preview:
Citation preview
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
1/126
2012 American Aviation Institute. Protected document, all rights reserved.
The State of U.S. Aviation
Comprehensive Analysis ofAirline Schedules & Airport Delays
Darryl Jenkins, Chairmandjenkins@aviationinstitute.org
Joshua Marks, Executive Directorjmarks@aviationinstitute.org
Michael Miller, Vice President, Strategymike@aviationinstitute.org
February 16, 2012
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
2/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
3/126
2012 American Aviation Institute. Protected document, all rights reserved.
ABSTRACT
As the U.S. economy rebounds from cyclical recession, inevitably there is a point where the media
asks if our national aviation system is at capacity, or is about to reach capacity. Consumer advocates and
regulators claim that airline over-scheduling (which we define as scheduling more flights at an airport
than reasonably sustainable over the long term) is responsible for excessive flight delays and cancellations.
Claims are often made about airport and airspace capacity, where airport and airspace choke points exist,
and how to fix them through government intervention.
This report intends to illustrate the dependencies and intricacies of airline, airport and air traffic
control constraints on the U.S. air transportation system. AAI has conducted an extensive analysis of
flight-level delays, airline capacity, operating metrics and flight schedule design. We believe it the most
extensive airline schedules and delay analysis undertaken to date, using more than 200 million data records
spanning 20 years, and investigating delays and aircraft performance at 400 United States airports. We
conclude that airline scheduling decisions are consistent with the expected capacity of given airports, asdefined by FAA Operational Metrics, historical and reasonable future weather conditions, and changes
made to airline scheduling practices including aircraft gate turn times, schedule de-peaking and en-route
scheduled block time. Significant variability of ramp, taxiway and local airspace capacity at certain
airports impacts real-world airline departure flow capacity and is the primary driver of unexpected lengthy
delays. We illustrate that direct government intervention in airline scheduling through demand
management (e.g. airport slot control) would be counter-productive. If the U.S. government undertook
demand managing of flights at major airports to match worst-case runway capacity, it would cause $18.7
billion in annual economic harm. However, long-term investment in airspace systems set in recent passage
of the FAA Reauthorization bill will have a significant benefit to both airspace efficiency and delays if
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
4/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
5/126
2012 American Aviation Institute. Protected document, all rights reserved.
CONTENTS
1. Executive Summary
2. Key Findings & Conclusions
3. Flight Delay Trends & Analysis
4. Airport Capacity Benchmarking
5. Schedule Peaks, Capacity & Delays
6. Rebutting The Overscheduling Argument
7. Conclusions and Recommendations
8. Exhibits & Appendices
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
6/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
7/126
2012 American Aviation Institute. Protected document, all rights reserved.
SECTION ONE: EXECUTIVE SUMMARY
This paper presents an exploratory analysis of the relationship among flight delays,
airline schedules and airport capacity. It is the most extensive data-driven analysis of airline
schedules and delays even undertaken. Data cited is for scheduled airlines. Airport data also
includes non-scheduled capacity, including cargo, military and general aviation operations. Flight
delays are gate departures or arrivals 15 minutes or greater after the scheduled times. Airline
scheduling is the assignment of specific aircraft and departure times to routes served. Airline
over-scheduling is the practice of scheduling more flights in a given time window than an airport
can accommodate without chronic delays given environmental conditions that influence ramp,
taxiway and runway capacity. To date, the connection between controllable schedule planning
and flight delays has been neither quantitatively defined nor proven.
We review the connection between airline flight schedules, airport conditions and
constraints, and observed flight delays and cancellations. We assess the relationship during
varying weather and operational conditions with focus on the following three questions:
1. How have airlines adapted their flight operations and schedule planning to mitigate
the impact of flight delays and cancellations?
2. Is the airline practice of scheduling arriving and departing banks of aircraft to
minimize connecting time for passengers (defined as peaked schedules) causing
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
8/126
Section One: Executive Summary
!
actionable policy recommendations regarding slot creation, slot trading and systemwide delay
management, either in the current ATC environment or under future NextGen infrastructure.
We find that intentional scheduling decisions by airlines play a limited role in causing (or
addressing) flight delays observed. Weather factors, regional choke points and intersecting
departure and arrival corridors between airports remain the most important factors connected to
observed delays. The overall level or composition of airline demand has minimal impact
(positive or negative) on observed delays except where the airport is operating at a very high level
of demand relative to capacity. In these limited situations, government-organized demand-
management programs (in the form of reservations and/or slots with rigid restrictions on trading
and re-allocation) limit airlines ability to optimize flight schedules in order to minimize delays.
As a result, we observe abnormal flight delays at airports with these government-managed
demand controls, and a significant follow-on impact of these controls systemwide.
We find no evidence that airport-wide airline scheduling decisions create excessive
exposure to external delay factors (weather and airspace). Balanced departure flows, buffers
in en-route scheduled time, and extended airport turn times provide a meaningful buffer against
weather and airspace congestion impact. There have been changes in each of these strategies
since the delay peak of the mid-2000s. We demonstrate that since 2005 most carriers have taken
significant pro-active steps insulate flights from the follow-on impact of delays.
We focus in particular on strategic delay management programs into their schedules,
i i h i li d d t fli ht ti t i t li b bl d l W
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
9/126
2012 American Aviation Institute. Protected document, all rights reserved.
Operational Benchmarks published for U.S. airport facilities, the actual capacity of the given
runway infrastructure (the ongoing sum of departure and arrival rates) and weather variability
observed during the past 10 years. We map airline schedule demand against capacity and observe
that operating in excess of the FAA Benchmarks does not necessarily result in flight delays. We
identify problems inherent in the FAA Benchmarking methodology, including exclusion of ramp
and taxiway capacity, weather variability by hour of the day, and other qualitative factors such as
controller experience that materially impact airport operation rates, as factors airlines take into
account today when setting schedules. We confirm that the FAA Benchmarks offer a usefulstarting point for airline schedule planning, but require adjustment to gauge absolute capacity of a
given airport under specific weather conditions.
Consumer advocates frequently discuss the theoretical concept of airline over-scheduling
in the media. Commentators casually define over-scheduling as the practice of consciously
operating more flights than an airport can reasonably accommodate. We find past evidence ofover-scheduling in only one market New York City and only during two specific windows, at
LaGuardia airport after the revocation of commuter slots in 2000, and at JFK airport during the
fall of 2006. Both proved expensive lessons for both airlines and the traveling public, and
resulted in a significant overhaul of airline scheduling practices. In todays environment, we find
no evidence connecting flight planning decisions made by a specific carrier at a specific
airport and resulting delays or cancellations that can be traced directly to that schedule.
We also find the claim that controllable factors such as airline schedules can
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
10/126
Section One: Executive Summary
!
periods present in banked models. We show that concentrated or banked hub models operating
below maximum capacity actually drive the lowest overall delay metrics relative to highly
utilized or rolling peers.
We conclude that delay reduction through strategic flight scheduling has become an
integral part of airline planning and decision-making. Rolling hubs, strategic delay management
through expanded block times (or schedule padding) and increased aircraft turn times
internalize the impact of weather and airspace variability.
We propose three policy or regulatory options that we believe could drive significant
furtherimprovements in delays and cancellations.
First, we observe a critical need for open-market slot-trading mechanisms at demand-
managed U.S. airports including New Yorks JFK and LaGuardia, and Washington Reagan
National. Political interference (driven by regional air service demands) restricts airlines ability
to re-allocate slots economically to match demand to capacity. In 2011 DOT approved an
elaborate slot exchange between Delta and US Airways at LaGuardia and Washington National,
but the conditions for divestment in this transaction indicate an increased ambition by
government to micro-manage flights and competition at slot-restricted airports. In addition, this
transaction took more than two years to complete, a further delay of open market competition.
Pooling commuter and mainline slots currently there are rules that force an airline to
operate aircraft that conform to one type of slot or the other would allow larger and more
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
11/126
2012 American Aviation Institute. Protected document, all rights reserved.
immune discussions give airlines unfair leverage over consumers, since they are incentivized to
reduce block time and reduce taxi times. Carriers would achieve lower labor and fuel costs.
Third, operational regulations including invasive limits on taxi times could be improved,
as these have had a significant impact on gate holds and flight cancellations since April 2010.
The flight cancellation rate during bad weather events has been consistent across operational
seasons. However, bad weather cancellation rates increased from 3.6% to 5.2% excluding the
severe weather patterns observed during the winter of 2010-2011. DOT consumer protection
regulations have shifted delay time from post-gate departure to before boarding, with the impact
of congesting gate and ramp facilities. We believe that clear enforcement guidance, reasonable
fines (versus multi-million dollar penalties in force today that exacerbate cancellation behavior)
and re-framing the regulation to measure gate return decisions by the pilot in command, versus
the aircraft at-gate, would significantly improve flight completion rates nationwide and help
hundreds of thousands of consumers to reach their destination faster.
We find that airline schedules are based on reasonable estimates of weather probability
and competitive runway demand, grounded in historical weather statistics and justified by FAA-
published operational metrics. Weather and regional airspace congestion outside airline control
are the primary drivers of excessive delays, just as they remain the key drivers behind flight
cancellations and extended on-board tarmac times. We conclude that changes in aircraft turntimes and incorporation of expected delays into published en-route times mitigates the impact of
delays, providing consumers with a reasonable expectation of arrival times. Fundamental change
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
12/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
13/126
Section Two: Key Findings and Prior Work
!
SECTION TWO: KEY FINDINGS AND PRIOR WORK
This paper presents an exploratory analysis of flight delays, airport capacity, airline
schedules and their corresponding relationships to departure and arrival delays observed at major
airports in the United States. Using weather, flight schedule, and delay information from aviation
(DOT and FAA), climate (NOAA) and commercial (OAG) sources, the paper explores the
relationships between schedules, capacity and flight-level operational data.
2.1 Key Findings
From our exploratory analysis, we observe the following:
1. Airline scheduling practices have adapted with an increased focus on on-time
performance. We focus on three broad categories of operational strategy changes:
en-route scheduled time (block time and schedule padding), hub-airport
connecting flight structures and banking, and scheduled/actual aircraft turn times at
key U.S. airports.
2. Over the past 10 years, airlines have progressively de-peaked their flight schedules
at major hubs. Outside of the New York metropolitan area, most hubs operate on a
rolling structure at 58% of published capacity.
3. The airports with the most significant peaks in operational scheduling are also the
airports with the lowest observed aggregate delay rates We recognize that this
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
14/126
Section Two: Key Findings and Prior Work
!
5. For short-term weather and airspace disruptions, peaked schedules insulate flights
from rolling delays originating at the airport. The valleys between peaks in a
schedule provide an opportunity to recover from short-term disruptions. In
contrast, airports that operate near capacity but without peaks offer little
opportunity or slack capacity to recover until the end of the day so delay and
cancellation events snowball. These airports can be demonstrated to originate
flight delays that continue throughout the system.
6. Peaked schedules as determined by individual airline scheduling decisions are a
red herring when considering flight delays at a given facility. Peaked schedule
decisions by individual carriers were relevant 15 years ago as delay drivers, but are
not as relevant today.
7. With the shift from peaked to rolling banked models at major hubs, airlines have
generally expanded the time allocated to turn aircraft at these facilities. Average
aircraft turn times have increased 4.2% systemwide between 2005 and 2010.
Longer turn times provide insulation against downstream impact from delayed
inbound arrivals. They also provide additional time to address carrier-caused
delays such as mechanical and crew staffing events.
8. There has been a significant shift in the variability of taxi-times at key U.S. airports
relative to changes in actual flight times. This is driven by the first-in, first-out
i t t i l t d Ai ft t d i t th t
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
15/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
16/126
Section Two: Key Findings and Prior Work
!
time rates set by the FAA. We point to airspace capacity improvements
particularly implementation of NextGen infrastructure as the key to improving
flight performance without economic harm.
15. We define potential follow-on research that could identify the specific marginal
conditions when schedules become unreliable, the recursive relationship of
weather, runway, taxiway and ramp congestion on operational performance, and
how airlines could exchange scheduling intentions to reduce the impact of
cumulative competitive decisions without violation of anti-trust regulations.
2.2 Prior Work
Our work continues analysis of delays, airport capacity and related trends by both
academic and industry practitioners. Table 1 below presents key research papers during the past
10 years that assess the relationships between flight delays and airport capacity.
Table 1: Prior Work on Airline Delays and Airport Capacity
For more information, see Xu, Laskey & SherryMethod for Deriving Multi-Factor Models for Predicting Airport Delays (2007)
Response variable Predictor variables Author Methods
Probability of on-timeperformance, delaysand flight cancellations
at New York LaGuardia
Discussion of generalflight delay and flightcancellations
Economic (e.g. revenue, loadfactors), Route Competition (e.g.Monopoly), Airport competition
(e.g. Concentration at origination,hub destination), Logistical (e.g. slotorigination, distance, hours untilnext flights). Weather (e.g. rain,minimum temperature frozen)
Rupp 2005 Nested Logit Model
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
17/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
18/126
Section Three: Flight Level Delay Trends
!
SECTION THREE: FLIGHT LEVEL DELAY TRENDS
The fundamental question when assessing delay causes is this: Do independent airline
scheduling decisions, made with rational internal justification in pursuit of each airlines
commercial objectives, cause delays in the national airspace system?1
The answer has both obvious and non-obvious components. To start with the simplest
case, if airlines scheduled no flights, there would be no flight delays. As system utilization builds,
there is a general relationship between demand and delays. However, previous work has
demonstrated that the correlation between demand and delays is weak until aggregate demand
levels approach the absolute capacity limits of a given airport.2
There are few cases where
airports actually operated at or above the published capacity limits, and in the most recent case
at New York LaGuardia after slot restrictions on regional jets were lifted in 2000 the impact on
delays was severe. Government-administered demand management programs such as slots or
allocations are imposed before airports reach the break point. It is worth noting that delay and
congestion issues are prevalent at airports with government-administered programs, for reasons
that we discuss later.
Weather, airspace congestion, airport capacity and aircraft dispatch reliability all have
stronger relationships with delays than airline schedule demand for airport resources. In this
paper, we investigate flight delay trends, airport capacity benchmarks and changes in airline hub
scheduling to explore the relationships among these factors and correlation to aggregate flight
d l d ll ti Th ti k t dd i h th h t i li
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
19/126
Section Three: Flight Level Delay Trends
To examine optimal scheduling levels, one must construct a definition for excessive
scheduling. As noted in prior works on a systemwide basis, there is an inherent trade-off between
capacity (number of flights) and delays.4
The higher the system load, the less flexibility airlines
have to recover their operations when weather, airspace, mechanical or security factors prevent
scheduled flight operations. We define airline over-scheduling as the practice of consciously
scheduling more flights to and from a given airport than the capacity level at that airport
reasonably expected based on a probability distribution of weather, airspace, and systemic
delay factors.
This definition parallels the independent scheduling decisions that occur at airlines as
they plan future schedules, block times and gate utilization. We will utilize this definition
throughout this paper, reviewing where capacity definitions used by the FAA and other regulatory
bodies differ from our standard.
3.1 Exploratory Analysis Objectives
Our exploratory analysis is intended to establish relationships among capacity, weather
and airline schedules in order to ultimately model flight delays. Our approach is granular, using
schedule, weather and on-time data on an airport-specific basis. Our approach builds up to
systemwide estimates, rather than focusing on systemwide performance metrics that mask
important differences in specific airport designs, schedules and weather conditions. Our analysis
uses publicly available data from government aviation sources (including FAA and DOT
hi t i l d t ) th i f ti (f NOAA d th N ti l Cli t D t C t ) ll
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
20/126
Section Three: Flight Level Delay Trends
!
d. Do airlines, individually or in aggregate, over-schedule their operations? If so, to what
extent do airline schedules correlate to flight delays observed?
e. Given the core conditions described above, what improvements in physical infrastructure
or airline practices could insulate schedules from repetitive delay drivers?
These are critical to consider in determining capacity planning decisions and delay
management strategies. Under the current radar-based traffic management infrastructure, capacity
flexibility is limited by absorption rates of airport-specific airspace. The transition to NextGen
GPS-based ATM infrastructure will offer new flexibility to airlines and airports. Understanding
the delay implications and relationships will be critical to justifying infrastructure investment by
both government and industry. Various academic, regulatory and industry reports have linked
airline schedules and flight delays, but no comprehensive model has yet demonstrated a direct
(causal) link between the two.
Much of the existing research on factors relevant to airline scheduling and delays is
quantitative and theoretical, sometimes ignoring real-world operational constraints. By starting
with systemwide data (as opposed to ground-up airport analysis) the body of research understates
key differences in runway configuration, weather patterns, airline competition and schedule
drivers for each airport. The optimal framework for schedule and delay analysis mirrors the
everyday scheduling decisions made by carriers. The optimal flight schedule tempers the
revenue-maximizing schedule with a given probability expectation of weather and other
operational disruptions. No airline plans to operate at 100% on-time performance levels, and
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
21/126
Section Three: Flight Level Delay Trends
In this section, we review the general delay trends and effects observable through the past
20 years. In particular, we focus on flight-level delay and cancellation data since October 2008,
when U.S. airlines reporting flight performance data to DOT began to report more detailed flight
ground delay, diversion and cancellation data. In Section Four, we will review airport capacity
benchmarking in order to compare delay trends against changes in airport capacity, and changes
in airline schedules, focusing on hub de-peaking and increased capacity utilization at key hub
airports between 2000 and 2010. Finally, in Section Five we will review the arguments made by
consumer advocates to support claims of airline overscheduling and demonstrate why such
claims are based on questionable assumptions and estimates.
3.2 General Delay Causes
Our objective for this section is to review the primary causes of airline delays, and
investigate the scheduling, equipment and network patterns that drive observable differences
among key U.S. airlines. There are three primary factors that cause airline delays and disruption
to the planned flight schedule:
1. Non-systemic, airline-specific factors, including mechanical failures, labor
resources, inability to load or deplane aircraft, inability to de-ice and unavailability
of required gate and ramp assets;
2. Competitive scheduling and operational decisions made by other carriers at the
airport, which may impact the availability of ramp, taxiway and runway assets
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
22/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
23/126
Section Three: Flight Level Delay Trends
To address this information gap, the FAA has published operational capacity benchmarks
for each major U.S. airport.7
The FAA's estimates provide three predicted capacity levels for
arrivals and departures: the rate during good weather, marginal weather and bad weather. Last
updated in 2004, the FAA estimates establish a baseline of data to use for analysis of airport
capacity. Because the Operational Benchmarks exclude information about a given airport
facilitys ramp and taxiway infrastructure, gate capacity, local airspace congestion and the very
relevant human factors around controller flexibility and recovery experience, the Operational
Benchmarks are a better estimate ofrunway capacity versus airportcapacity. In Section Four,
we examine how those benchmarks have evolved in practice as new runway capacity and
schedule optimization since 2004 have impacted airport and airline operations. We also show
why the Operational Benchmarks must be reviewed in conjunction with airline schedule, airport
movement area and terminal capacity issues.
In addition to core Operational Benchmarks, the FAA also sets arrival and departure
capacity for key U.S. airports throughout the day.8
These metrics are called Airport Arrival Rates
(AAR) and Airport Departure Rates (ADR). The available airport capacity is the sum of ADR
and AAR. These metrics represent the maximum flow of arrivals and departures that a given
airport can process. They incorporate more real-world factors than the theoretical Operational
Capacity Benchmarks, including the inherent trade-off between arrivals and departures (that is,
increasing arrival capacity by a given amount may result in a disproportionate decrease in
departure capacity because of runway configurations and arrival/departure corridors in the airport
vicinity).
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
24/126
Section Three: Flight Level Delay Trends
!
Table 2: Airport Utilization vs. Capacity
Source: FAA ASPM Full-Year 2009, 7am-10pm Time Window% DepDly = Percent of flights delayed by 15+ minutes at departure
Av Delay = average minutes of delay for delayed flights
Name of Airport Capacity Flights Utilization % DepDly Av Delay
New York LaGuardia 424,913 341,380 80.3% 21.3% 34.7
Newark Liberty 469,685 366,744 78.1% 26.6% 41.0
Atlanta Hartsfield-Jackson 1,205,933 926,551 76.8% 22.4% 41.0
New York John F. Kennedy 485,967 368,778 75.9% 22.5% 40.8
Philadelphia International 579,855 404,097 69.7% 24.2% 38.0
Chicago O'Hare 1,141,379 765,937 67.1% 21.0% 45.6Washington Reagan 394,189 256,876 65.2% 16.4% 27.6
San Francisco International 531,344 321,660 60.5% 23.1% 35.5
San Diego Lindbergh 278,107 162,753 58.5% 18.8% 20.4
Charlotte/Douglas Int'l 768,646 445,110 57.9% 18.6% 29.1
In Table 2, capacity refers to the rate of arrivals and departures that the aircraft reported it
could handle during the full year of 2009. This is based on the cumulative capacity sum of every
quarter-hour measurement reported. Flights reflect the departures and arrivals scheduled as the
annual sum of each quarter-hour, and utilization is the planned ratio of flights to capacity.
There is an observable relationship between airport utilization and delays, particularly as
the percentage utilization of given airports exceeds 70%. The correlation coefficient for major
airport facilities between utilization and departure delays in 2009 (during peak hours of 7am to
10pm) was 0.56, and the relationship strengthens as utilization approaches 100%. But there are
many highly utilized airports in the U.S. that operate during peak hours with above-average
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
25/126
Section Three: Flight Level Delay Trends
Airport capacity is fungible. It is a subjective metric given the quantitative (runway,
ramp and airspace) and qualitative (human factors) involved. Some airports (including Seattle,
Baltimore, Cleveland, Los Angeles and Tampa) routinely achieve operational throughput well in
excess of their respective FAA capacity benchmarks, while others (including Memphis, Portland,
Detroit, and Boston) operate comfortably below the FAA benchmark definition. For this reason,
the FAA benchmarks are not sufficient to assess whether airline schedules are appropriate, or
whether conscious over-scheduling of airport resources is occurring.
Qualitative factors include human factors (controller experience and flexibility, for
example) and independent decisions made by airlines to push or gate hold, which can be quite
relevant when assessing an airports throughput. Airlines must review all these factors to align
schedules with airport capacity. Airlines have multiple objectives in planning and executing flight
schedules. First, they flight schedules to match the times of day when passengers want to fly, and
when destination arrivals are available. This simple objective offers direct insight into why
airline schedules are bunched at origin and destination airports (versus connecting hubs) around
key departure and arrival windows. Business passengers value departures and arrivals that permit
full working days. International flights are usually timed around available slots and must account
for time zone differentials. As a result, there is an inevitable crowding of departure demand at
U.S. airports with strong local market demand during morning and evening hours, with lower
volume during the middle of the day. Because of demand patterns, a carriers flexibility to spread
flights more consistently through the day in key origin and destination markets is limited.
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
26/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
27/126
Section Three: Flight Level Delay Trends
Chart 1b: Correlation Between On-Time Performance and Flight Operations by MonthSource: DOT Part 234 On-Time Reporting Data (Annual -0.26)
There are also internal trade-offs at airlines between decisions to delay a given flight
versus cancel. As we have shown in prior papers, every flight cancellation decision in todays
hub and spoke networks usually results in at least one follow-on flight cancellation due to an
aircraft out of position. In contrast, particularly when sufficient aircraft turn time is built into
downstream flight operations; aircraft delays can sometimes be isolated to a specific flight
segment.
Chart 2a: Systemwide Cancellations & Diversions vs. Operations, by Hour
Source: DOT Part 234 On-Time Reporting Data
0.08
-0.43
-0.27-0.22
-0.35
-0.48
-0.58
-0.43
-0.21
-0.37
-0.21
-0.46
1 2 3 4 5 6 7 8 9 10 11 12
18,000!20,000!
900,000!1,000,000!
Total
BLUE)
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
28/126
Section Three: Flight Level Delay Trends
!
Chart 2a above shows the distribution of airline flight operations for domestic U.S. flights
by reporting carriers during 2010. The peak of flight operations during the early morning and late
afternoon can be clearly observed. In Chart 2a, the blue line tracks the total number of flight
operations scheduled by hour of the day (local time). The red line tracks the number of flight
cancellations and flight diversions, demonstrating the sum of irregular flight operations that
significantly disrupt downstream flights. There is a spike in cancellation and diversion decisions
during the late afternoon hours (from 4pm to 8pm local time). This is primarily due to the
combined impact of summer thunderstorm activity and cancellation decisions designed to contain
follow-on impact from upstream flight delays.
How do airline cancellation decisions by hour of the day compare to flight delay
decisions? Given the significant differences in numbers of flight delays, cancellations and
diversions relative to total flight operations, it is clearest to express these data on a percentage
basis. Chart 2b below demonstrates that (1) airline flight delays start at a relatively low level in
the morning and snowball through the day, and (2) cancellation decisions remain more constant
on a percentage basis.
Chart 2b: Systemwide Cancellations/Diversions versus Flight Delays
As Percentage of Total Scheduled Flight Operations (Full Year 2010)
Source: DOT Part 234 On-Time Reporting Data
20%!
25%!
Cancellations/Diversions Delay Rate
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
29/126
Section Three: Flight Level Delay Trends
carriers more susceptible to snowballing delays versus others? We examine these factors in the
next section.
3.5 General Delay Causes
Schedule-impacting factors can be broken into three categories: factors that impact flight
schedules prior to gate departure, factors that impact flights after push-back from the gate and
before takeoff, and factors that influence flights en-route and after landing. Before gate departure,
there are many events that can delay a departure, including mechanical issues, staffing, cateringand passenger boarding. Security and late inbound arrivals can also disrupt crew boarding and
aircraft availability. After gate departure, congestion on the ramp and taxiway can slow progress
towards the runway, and de-icing requirements can cause long delays as aircraft queue for
position on the de-icing pads. Airspace availability and capacity also impacts not only the pace
with which departures can occur, but also may cause ground-delay programs (or EDCT
programs, for Expect Departure Clearance Time) that result in long taxi-out times. Table 3 below
lists these general factors.
Table 3: Factors that Influence On-Time Performance
Before Gate Departure Awaiting Takeoff En-Route and Landing
Mechanical eventsCrew availability
Supplier deliveryPassenger & cargo boardingRamp personnel & equipmentWeather conditions
Ramp congestionTaxiway congestion
Visibility
Number of aircraft: Awaiting taxi Queued for runway
Convective activity, turbulenceand en-route winds
Weather at destination andalternate airport (IFR, etc.)
Departure and arrival queues at
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
30/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
31/126
Section Three: Flight Level Delay Trends
Chart 3 below shows that major network carriers such as Delta, American, United and US
Airways follow a similar profile of flight delays by hour of the day. Delays start in the morning
primarily due to carrier-related factors including late-inbound flights and mechanical events.
Delays build through the day, with between 15% and 25% of departures during the 6pm-8pm
window delayed. In contrast, point-to-point carriers such as JetBlue and particularly Southwest
Airlines have a more rapid accumulation of flight delays through the day. Southwest (WN) far
exceeds other airlines with flight delays during the evening, with more than 40% of departures
delayed by the 8pm hour.
Chart 3: Delayed Departures (as Percentage of Scheduled Departures) by Airline
By Hour of Departure (Full Year 2010)
15%!20%!25%!30%!35%!40%!45%!50%!
arturesbyH
ourofScheduledDeparture
Time
WN!B6!DL!AA!UA!
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
32/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
33/126
Section Three: Flight Level Delay Trends
Table 5: Minutes of Flight Delays
Source: DOT ASQP Part 234 Reports, Calendar Year2010
Flights Carrier Weather Airspace Security Late A/C Total Min.
6am-10am 1,748,868 4,601,508 570,538 3,265,070 27,701 1,391,586 9,856,403
10am-2pm 1,678,564 4,297,114 486,972 3,912,112 18,530 5,546,464 14,261,192
2pm-6pm 1,641,416 5,307,713 862,096 5,509,225 30,251 8,836,291 20,545,576
6pm-10pm 1,199,314 4,432,816 793,201 3,282,911 28,992 8,633,832 17,171,752
10pm-6am 181,956 578,354 72,405 260,062 1,936 510,113 1,422,870
Total 6,450,118 19,217,505 2,785,212 16,229,380 107,410 24,918,286 63,257,793
% of Total 30% 5% 26%
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
34/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
35/126
Section Three: Flight Level Delay Trends
Chart 5b: Delay Causes for Impacted Flights, by Departure Hour
Full Year 2010, U.S. Reporting Carriers on Domestic Flights
Chart 5b captures all flight delay causes that result in a late arrival (15 or more minutes
after scheduled arrival time). It groups Carrier, Weather, Airspace and Security causes into a
single category (Red) and isolates Late Arriving Aircraft (Blue). The snowball effect is clear.
3.6 Isolating Different Delay Factors
To investigate the causes of airline delays further, we explore factors that we hypothesize
Delays Caused by Late Arriving Inbound
Aircraft!
Other Delay Causes!(Weather, Airspace,Carrier & Security)!
0!20,000!40,000!60,000!80,000!
100,000!120,000!
!0001-0559
!
!0600-0659
!
!0700-0759
!
!0800-0859
!
!0900-0959
!
!1000-1059
!
!1100-1159
!
!1200-1259
!
!1300-1359
!
!1400-1459
!
!1500-1559
!
!1600-1659
!
!1700-1759
!
!1800-1859
!
!1900-1959
!
!2000-2059
!
!2100-2159
!
!2200-2259
!
!2300-2359
!NumberofFlights
Impacted(AllFlights2010)
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
36/126
Section Three: Flight Level Delay Trends
!
3.6.1 Length of Flight
Our first step is to identify any meaningful breaks in on-time performance results for
different types of airline routes. We isolated on-time arrival performance for 2010 by the
distance of flight, to the nearest 250 miles. As Chart 6 shows, on-time performance is stronger
for shorter-haul flights than for longer-haul. For consistency, we excluded all flights that were
delayed because of a late inbound aircraft. The resulting correlation coefficient between distance
and on-time arrival performance was -0.91, confirming a basic link between the factors.
Chart6: On-Time Domestic Performance by Flight Distance
Full Year 2010, U.S. Reporting Carriers (Excludes Late Arriving Aircraft Delays)
3.6.2 Major Hub Exposure
90.9%! 90.6%!
89.3%!89.5%!
89.1%! 88.8%!88.3%! 88.1%!
86.7%! 86.9%!84.8%!
0! 500! 1,000! 1,500! 2,000! 2,500! 3,000!Distance of Flight (miles)!
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
37/126
Section Three: Flight Level Delay Trends
Table 6a summarizes 2010 operating metrics for major airports in the U.S. It captures
domestic flight operations for reporting air carriers, using the DOT Part 234 ASQP data set.12
Table 6a: Key Delay and Cancellation Metrics, by Airport
Ranked by Airport Departures (Domestic Flights by Reporting Carriers)
Source: DOT ASQP Part 234 Reports, Calendar Year 2010On-Time Arrivals represents completion of flight from selected airport within 15 minutes of schedule
Delay minutes are based on impacted flights only
Rank AirportOn-Time
DeparturesOn-TimeArrivals
CancelledFlights
CarrierDelay (Avg)
WeatherDelay (Avg)
AirspaceDelay (Avg)
Inbound ACDelay (Avg)
1 ATL 81.0% 79.9% 2.0% 39.5 min 45.8 min 21.5 min 46.8 min
2 ORD 80.0% 79.0% 2.5% 35.2 31.7 24.1 45.3
3 DFW 80.5% 79.6% 1.7% 33.4 34.4 20.2 33.4
4 DEN 81.3% 81.2% 1.0% 27.5 35.6 20.5 38.3
5 LAX 83.5% 82.5% 1.1% 30.1 54.6 21.1 36.7
6 IAH 83.9% 81.1% 0.8% 31.3 26.6 20.3 42.0
7 PHX 84.1% 83.5% 0.8% 28.6 42.8 20.9 33.5
8 DTW 80.2% 78.1% 2.0% 41.0 53.8 23.2 43.4
9 LAS 79.9% 81.5% 0.7% 25.6 42.5 19.7 37.7
10 SFO 77.3% 78.0% 1.9% 33.8 39.2 16.2 56.5
11 MSP 81.4% 78.7% 1.7% 36.0 46.1 23.9 42.4
12 CLT 84.7% 81.6% 1.3% 34.4 44.6 28.2 38.6
13 SLC 85.7% 82.7% 0.8% 32.3 29.9 21.0 32.8
14 MCO 82.0% 82.4% 1.0% 30.4 34.8 28.3 41.3
15 EWR 79.2% 79.8% 3.2% 32.9 32.5 27.4 56.1
16 BOS 83.8% 81.7% 2.7% 36.8 34.7 27.6 46.4
17 JFK 78.1% 78.9% 3.4% 45.4 69.1 30.7 46.6
18 BWI 77.6% 79.8% 2.2% 26.9 42.6 23.8 37.519 LGA 84.5% 81.7% 4.2% 35.1 41.1 29.3 51.7
20 SEA 88.5% 86.9% 0.5% 35.3 32.5 25.4 40.0
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
38/126
Section Three: Flight Level Delay Trends
!
ultimate on-time arrival rates for flights leaving those airports. The data confirm that on average,
larger hub airports indeed have higher departure delay rates than smaller facilities.
Table 6b: Key Delay and Cancellation Metrics, by Departure Level Groupings
Source: DOT ASQP Part 234 Reports, Calendar Year 2010
ASQP Flights perYear
% of Total OTD% OTA% Cxl%
0-12,500 11.2% 83.3% 80.8% 2.4%
12,500-37,500 12.2% 83.5% 81.9% 1.6%
37,500-62,500 14.3% 81.6% 81.5% 1.3%
62,500-87,500 8.3% 79.9% 79.4% 2.0%
87,500-112,500 7.9% 79.5% 78.8% 2.6%
112,500-137,500 11.3% 81.1% 79.1% 1.8%
137,500-162,500 6.9% 77.6% 77.4% 1.5%
162,500-187,500 5.7% 83.2% 81.3% 0.8%
187,500-212,500 3.1% 82.4% 81.2% 1.1%
237,500-262,500 3.7% 80.2% 79.9% 1.0%
262,500-287,500 4.2% 78.8% 77.7% 1.7%
312,500-337,500 4.9% 77.5% 76.3% 2.5%
412,500-437,500 6.4% 79.0% 77.7% 2.0%
All Airports 100.0% 81.0% 79.8% 1.8%
Table 6c: Percent of Flights Impacted by Delays, by Departure Level Groupings
Source: DOT ASQP Part 234 Reports, Calendar Year 2010
Size of Airport Carrier Weather AirspaceLate
Inbound
0-12,500 5.1% 0.9% 10.2% 9.2%
12 500 37 500 6 3% 0 7% 9 5% 9 0%
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
39/126
Section Three: Flight Level Delay Trends
larger hubs. Of note in Table 6c is the consistency of carrier-related delays incurred at larger
(hub) airports. Given that crew staffing, catering, maintenance and other infrastructure is
concentrated at hubs, it is not surprising that delays are concentrated in larger airports.
Table 6d: Of Delayed Flights, Minutes of Delay by Cause, by Departure Groupings
Source: DOT ASQP Part 234 Reports, Calendar Year 2010
Size of Airport Carrier Weather AirspaceLate
Inbound
0-12,500 44.3 57.8 28.8 47.9
12,500-37,500 35.3 47.8 27.8 42.737,500-62,500 29.0 41.5 24.9 38.5
62,500-87,500 34.0 45.7 25.5 40.2
87,500-112,500 35.5 52.4 27.9 43.8
112,500-137,500 33.8 37.0 25.9 42.7
137,500-162,500 34.0 48.8 20.3 47.2
162,500-187,500 29.9 28.7 20.5 37.1
187,500-212,500 30.1 54.6 21.1 36.7
237,500-262,500 27.5 35.6 20.5 38.3262,500-287,500 33.4 34.4 20.2 33.4
312,500-337,500 35.2 31.7 24.1 45.3
412,500-437,500 39.5 45.8 21.5 46.8
All Airports 34.0 42.4 24.8 42.3
Of note is the high degree of impact from airspace delays at smaller airports. This is
primarily the impact of ground-delay programs for flights into congested airspace. In contrast,
flights from major airports to small cities generally do not face the same airspace constraints.
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
40/126
Section Three: Flight Level Delay Trends
!
Table 6e: Key Metrics by Aircraft Type
Source: DOT ASQP Part 234 Reports, Calendar Year 2010, Domestic Flights Only
Aircraft TypeCategory
% ofReport
% DepsOntime
% ArrsOntime
CarrierDelay %
CarrierDel Mins
WeatherDelay %
WeatherDel Mins
Late Inb.Delay %
Late Inb.Del Mins
Boeing 767 1.1% 81.8% 78.8% 11.8% 51.7 10.9% 28.4 5.0% 52.5
Boeing 777 0.2% 80.3% 80.5% 11.0% 48.5 11.2% 29.2 3.9% 71.2
Airbus A320 13.1% 83.9% 81.6% 8.5% 31.9 11.4% 27.4 7.9% 42.1
Boeing 717 3.6% 88.1% 86.3% 3.7% 34.7 7.2% 29.5 6.6% 56.5
Boeing 737 28.6% 79.6% 80.9% 10.3% 24.1 9.4% 22.8 10.8% 36.7
Boeing 757 6.4% 81.0% 79.6% 9.5% 39.6 12.7% 27.9 7.6% 44.4
MD-80 Series 9.4% 80.6% 79.2% 9.1% 39.1 12.4% 27.7 8.5% 42.2
Regional Jets 35.0% 81.6% 79.1% 8.1% 41.2 12.6% 27.3 9.1% 45.4
Turboprops 2.6% 80.8% 79.3% 5.9% 36.1 10.3% 26.0 12.3% 50.2
All Types 100.0% 81.4% 80.2% 8.8% 34.0 11.2% 26.4 9.2% 42.3
3.6.4 On-Time Performance vs. Departures
To conclude the discussion of general on-time performance metrics versus annual
departures, we cross-plot the on-time departure performance of major U.S. airports against the
number of annual departures handled. As Chart 7 illustrates, there is a small negative correlation
between the size of an airport (measured in annual departures) but the distribution of performance
among the bulk of U.S. airport facilities is too significant to draw a meaningful conclusion. In
fact, many small airports perform substantially worse than larger facilities on an annual basis.
Chart 7: On-Time Performance vs. Airport Size (Departures)
Top 35 OEP Airports for Calendar Year 2009
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
41/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
42/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
43/126
Section Three: Flight Level Delay Trends
Table 8: Occurrence of Inclement Weather ConditionsIMC = Instrument Conditions; VMC = (Good) Visual Conditions;
Ceiling = ceiling < 1000 feet; Visibility = Visibility < 1 mile
CODE IMC VMC CEILING VISIBILITY
SEA 30.3% 69.7% 6.9% 2.1%
MEM 25.6% 74.4% 5.8% 0.3%
STL 24.4% 75.6% 5.1% 0.3%
ATL 23.7% 76.3% 9.9% 3.0%
MSP 23.6% 76.4% 3.0% 0.5%
IAH 21.6% 78.4% 5.5% 1.6%
DTW 20.1% 80.0% 3.9% 1.0%
PDX 20.0% 80.0% 3.6% 1.6%
SFO 20.0% 80.0% 3.0% 0.1%
CLT 19.8% 80.2% 8.1% 1.6%
IAD 18.6% 81.4% 7.7% 1.8%
Weather has a varying impact on airports in different regions of the country. As Table 8
above shows, airports such as Seattle/Tacoma are frequently impacted by negative weather
conditions, with low ceilings, low visibility and precipitation. As Exhibit E shows in detail,
airports such as Miami, Las Vegas, Phoenix and Honolulu are all impacted by inclement weather
less than 2% of the time. Simply comparing the incidence of weather conditions with on-time
departures reveals a low correlation coefficient of just -.011. As Chart 9 indicates, many airports
with high occurrence of IFR weather routinely outperform airports with lower occurrence.
Chart 9: Occurrence of Bad Weather Conditions vs. Late Departures
(Correlation = -.11)
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
44/126
Section Three: Flight Level Delay Trends
!
Exploratory analysis confirms that the overall frequency of bad weather conditions is not
correlated to low on-time performance, but the variability of that weather is a critical factor.
Airports such as Seattle and Memphis are highly optimized for instrument approaches and radar
separation. Airlines can also plan schedules with confidence in the arrival and departure rates
during bad weather conditions. But when considering airports that have a high variability in
weather conditions, such as mid-Atlantic and Southern airports subject to thunderstorm activity,
we can observe two factors. First, many airports have a significant loss of capacity during bad
weather conditions. Second, some airports have infrequent and unpredictable weather conditions
that can paralyze arrivals and departures for extended periods of time.
3.7 Diversions and Cancellations
Our analysis has thus far focused on delay causes, but delays are both driven by and
causal to cancellations and aircraft diversions. In the next section, we review diversions and
flight cancellations by carrier, to compare the rates and causes of incidents against flight delays.
3.7.1 Diversions
Airlines report diversion data for domestic flight segments to the DOT. Airlines report
when flights return to their origin, for mechanical or weather, and when flights divert en-route.
Flight completion information is included to determine whether the flight ultimately landed at itsdestination. Table 9 shows flight diversions for reporting U.S. airlines during 2010. The highest
diversion rates were reported by American Airlines and its affiliates, reflecting strategic changes
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
45/126
Section Three: Flight Level Delay Trends
Table 9: Diversions by Reporting U.S. Carriers, 2010
Source: Reporting Carriers, Domestic Flights, FY2010, DOT Part 234 (ASQP)
Airline FlightDiversions
Div. Rateper 1,000
Flights w/1 Div.
1 Div. FltsCompleted
Flights w/2 Divs.
2 Div. FltsCompleted
Southwest (WN) 2,166 1.9 2,135 74% 30 23%
Delta (DL) 1,621 2.2 1,600 87% 21 29%
SkyWest (OO) 1,748 2.9 1,721 64% 27 4%
American (AA) 1,957 3.6 1,940 94% 17 41%
Eagle (MQ) 1,073 2.5 1,056 87% 17 35%
US Airways (US) 683 1.7 675 86% 8 0%
ExpressJet (XE) 1,072 2.8 1,065 93% 7 43%
United (UA) 748 2.2 737 88% 11 9%
ASA (EV) 617 1.9 617 83%
Pinnacle (9E) 745 2.9 733 74% 12 33%
AirTran (FL) 631 2.5 627 94% 4 75%
Continental (CO) 620 2.6 616 95% 4 50%
JetBlue (B6) 504 2.5 504 86%
Mesa (YV) 373 2.1 373 80%
Comair (OH) 301 2.0 301 78%
Alaska (AS) 384 2.8 380 44% 4 25%
Frontier (F9) 167 2.0 166 94% 1 0%
Hawaiian (HA) 64 0.9 64 97%
All Reporting 15,474 2.4 15,310 82% 163 25%
3.7.2
Cancellations
In prior reports we investigated cancellation causes, rates and trends during 2010 versus
13
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
46/126
Section Three: Flight Level Delay Trends
!
Table 10: Cancellations by Reporting U.S. Carriers, 2010
Source: Reporting Carriers, Domestic Flights, FY2010, DOT Part 234 (ASQP)
Airline Sched.Flights
TotalCancels
CancelRate (all)
Cancels(Carrier)
Cancels(Weather)
Cancels(Airspace)
Cancels(Security)
Southwest (WN) 1,124,487 11,597 1.03% 4,940 6,223 431 3
Delta (DL) 732,973 14,857 2.03% 6,583 6,961 1,313 0
SkyWest (OO) 599,621 11,932 1.99% 3,801 5,702 2,428 1
American (AA) 540,963 9,146 1.69% 3,572 4,667 905 2
Eagle (MQ) 436,976 12,075 2.76% 2,169 6,246 3,656 4
US Airways (US) 407,111 6,290 1.55% 2,902 2,445 942 1
ExpressJet (XE) 385,077 8,114 2.11% 812 5,577 1,725 0
United (UA) 343,081 5,010 1.46% 2,178 2,479 353 0
ASA (EV) 319,921 7,517 2.35% 5,137 1,560 820 0
Pinnacle (9E) 261,364 7,653 2.93% 5,029 1,690 933 1
AirTran (FL) 248,844 2,674 1.07% 751 1,632 291 0
Continental (CO) 239,271 1,986 0.83% 211 1,677 78 20
JetBlue (B6) 201,434 4,116 2.04% 548 3,512 55 1
Mesa (YV) 174,797 3,439 1.97% 1,397 1,513 525 4
Comair (OH) 147,633 5,645 3.82% 1,686 3,882 75 2
Alaska (AS) 136,950 797 0.58% 313 461 23 0Frontier (F9) 81,966 352 0.43% 89 263 0 0
Hawaiian (HA) 67,649 55 0.08% 38 17 0 0
All Reporting 6,450,118 113,255 1.76% 42,15637.2%
56,50749.9%
14,55312.8%
390.1%
How do the causes of cancellations as reported during 2010 compare with the causes of
flight delays? Carrier-related cancellations represented 37% of all cancellations, while carrier-
related delays were 30% of total reported delays. Weather-related cancellations were 49.9% of
ll i hil j f d l i f d f ll i
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
47/126
Section Three: Flight Level Delay Trends
Table 11: Cancellation Types by Reporting U.S. Carriers, 2010Source: Reporting Carriers, Domestic Flights, FY2010, DOT Part 234 (ASQP)Controllable: Carrier-Related Causes; Uncontrollable: Weather- and Airspace-Related Causes
Airline Cancel Rate Controllable Uncontrollable
Hawaiian (HA) 0.08% 69% 31%
ASA (EV) 2.35% 68% 32%
Pinnacle (9E) 2.93% 66% 34%
US Airways (US) 1.55% 46% 54%
Delta (DL) 2.03% 44% 56%United (UA) 1.46% 43% 57%
Southwest (WN) 1.03% 43% 57%
Mesa (YV) 1.97% 41% 59%
Alaska (AS) 0.58% 39% 61%
American (AA) 1.69% 39% 61%
SkyWest (OO) 1.99% 32% 68%
Comair (OH) 3.82% 30% 70%
AirTran (FL) 1.07% 28% 72%Frontier (F9) 0.43% 25% 75%
Eagle (MQ) 2.76% 18% 82%
JetBlue (B6) 2.04% 13% 87%
Continental (CO) 0.83% 11% 89%
ExpressJet (XE) 2.11% 10% 90%
All Reporting 1.76% 37% 63%
3.8 Airline-Specific Delays and Adaptations
We have demonstrated that controllable delay factors including delays due to carrier-
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
48/126
Section Three: Flight Level Delay Trends
!
immediately that system delay minutes are largely proportional to total flight operations, and but
that there are notable differences between delay patterns among airlines.
Table 12: Delay Composition by Reporting U.S. Carriers, 2010
Source: Reporting Carriers, Domestic Flights, FY2010, DOT Part 234 (ASQP)
System Statistics Delay per Flight Cause of Delay
Airline Flights Delay Min All Flights Impacted Carrier Weather &Airspace
LateInbound
Southwest (WN) 1,124,487 10,029,222 8.9 46.2 28% 17% 55%
Delta (DL) 732,973 8,195,204 11.2 54.8 36% 33% 32%
SkyWest (OO) 599,621 6,211,902 10.4 55.5 22% 27% 50%
American (AA) 540,963 5,565,988 10.3 56.2 36% 33% 31%
Eagle (MQ) 436,976 4,777,239 10.9 54.8 29% 34% 37%
US Airways (US) 407,111 2,961,104 7.3 47.7 29% 40% 30%
ExpressJet (XE) 385,077 4,320,595 11.2 56.5 24% 37% 39%
United (UA) 343,081 2,660,172 7.8 58.9 25% 33% 43%
ASA (EV) 319,921 3,880,916 12.1 66.6 34% 23% 42%
Pinnacle (9E) 261,364 2,614,748 10.0 54.7 35% 29% 36%
AirTran (FL) 248,844 2,250,331 9.0 57.1 18% 32% 50%Continental (CO) 239,271 2,076,570 8.7 49.6 30% 46% 24%
JetBlue (B6) 201,434 2,717,652 13.5 61.3 35% 30% 35%
Mesa (YV) 174,797 1,411,861 8.1 55.7 35% 26% 39%
Comair (OH) 147,633 1,903,634 12.9 56.4 48% 44% 8%
Alaska (AS) 136,950 741,330 5.4 46.7 32% 33% 35%
Frontier (F9) 81,966 722,951 8.8 49.1 22% 29% 49%
Hawaiian (HA) 67,649 216,374 3.2 43.5 73% 2% 25%
All Reporting 6,450,118 63,257,793 9.8 53.8 30% 30% 39%
The next section assesses how airlines have internalized the expected delay patterns observed in
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
49/126
Section Three: Flight Level Delay Trends
To ensure consistency in the data sample, we used the following criteria to narrow our
analysis. We first collected all flight-level data for June 2010 from reporting U.S. carriers on
domestic flight operations. There were a total of 551,687 scheduled flights in the overall data set.
From this set, we excluded all flights that were either (i) diverted, (ii) cancelled, (iii) did not have
a tail number assigned, or (iv) occurred as the first flight of the day. This resulted in a set of
398,723 flights over one month where aircraft turns from a previous flight and to a subsequent
flight could be clearly identified on an aircraft-specific basis.
Our methodology created an authoritative (actual) data set, not one based on interpolation
from published airline schedules. However, our data set was based on domestic flights only, and
to exclude inside turns for international flight operation where an aircraft would operate a
quick international round-trip before returning to the domestic system we excluded aircraft
turns from our analysis that were greater than 240 minutes. We also excluded aircraft turns of
less than 20 minutes, as these flights were universally aircraft swaps not planned in advance. The
final data set had 302,399 flight turns for analysis across the reporting carriers.
The result was the following average turn time by carrier, systemwide during 2010.
Chart 10 below shows the distribution of planned turn times between 20 minutes and 100 minutes
(224,096 total flights). It shows a steady distribution of aircraft turns between 25 minutes and 50
minutes, with a downward slope of aircraft turns for 55 minutes or greater.
The importance of Chart 10 is that the shorter the turn, the more likely a late arrival
(d fi d 15 i t t ft h d l d i l ti ) ill i t f ll fli ht
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
50/126
Section Three: Flight Level Delay Trends
!
Chart10: Turn Time Distribution (20-100 minutes scheduled)
Source: Reporting Carriers, Domestic Flights, June 2010, DOT Part 234 (ASQP)
We now break turn times out by carrier to demonstrate the significant differences among
airlines, and set the foundation to link aircraft turn times to delay and cancellation metrics. Table
13 presents scheduled turn-time statistics by airline, by aircraft type. To isolate aircraft types, we
matched the tail numbers reported for each flight arrival and departure pair against the aircraft
type data in the FAA Registry. We grouped aircraft into three categories: regional (including all
Bombardier CRJ, Embraer 145 and 170/190 series and turboprop aircraft), narrowbody (primarily
Boeing 717 and 737, Airbus 320 series and DC-9/MD-80 aircraft) and widebody and
5,
260
!
23,
858
! 33,328
!
22,
810
!25,
207
!
23,
881
!23,
604
!
15,
447
!
11,684
!
8,
054!
5,
706
!4,
506
!3,
064
!2,
313
!1,
808
!1,
349
!993
!
20! 25! 30! 35! 40! 45! 50! 55! 60! 65! 70! 75! 80! 85! 90! 95! 100!
DomesticTu
rnsJune2010
(224,0
96total)
Turn Time Rounded to Nearest 5 Minutes!
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
51/126
Section Three: Flight Level Delay Trends
Table 13: Scheduled Turn-Time Statistics, by Carriers That Report
Source: DOT ASQP Part 234 Reports, June 2010
Carrier
All Types Regionals Narrowbody 757 & Widebody
Turns Mean St Dev. Mean St Dev Mean St Dev Mean St Dev
Southwest (WN) 77,251 30.0 10.3 30.0 10.3
Hawaiian (HA) 4,797 39.6 25.2 33.5 12.1 112.4 28.1
Eagle (MQ) 26,069 43.2 30.7 43.2 30.7
ExpressJet (XE) 25,821 44.8 28.6 44.8 28.6
SkyWest (OO) 38,781 44.9 28.6 44.9 28.6
AirTran (FL) 17,039 45.2 16.4 45.2 16.4
Frontier (F9) 5,420 48.4 18.0 53.3 11.5 48.4 18.0
Mesa (YV) 10,856 49.7 25.1 49.7 25.1
ASA (EV) 21,510 49.9 31.9 49.9 31.9
JetBlue (B6) 11,280 52.1 22.2 44.4 20.4 56.6 21.9
Pinnacle (9E) 17,193 54.3 32.8 54.3 32.8
Comair (OH) 7,921 58.4 37.1 58.4 37.1
American (AA) 28,550 59.0 20.9 55.5 17.5 77.4 26.6
Delta (DL) 43,174 61.2 29.5 57.5 26.7 73.8 34.8
Alaska (AS) 8,697 61.9 30.8 61.9 30.8
United (UA) 19,317 62.8 28.9 56.5 24.7 73.8 32.3US Airways (US) 23,170 67.4 26.1 60.6 24.2 68.1 25.9 88.9 40.6
Continental (CO) 11,877 68.5 28.2 66.8 27.7 80.8 28.2
Note: Airlines including Spirit, Virgin America and Republic do not report this and other data
Table 14: Turn Time Distribution (20-240 minutes) by Time of Day
Source: Reporting Carriers, Domestic Flights, June 2010, DOT Part 234 (ASQP)
Row Labels American JetBlue Continental Delta Southwest All Carriers
0001-0559 49 490600-0659 76 94 61 51 30 42
0700-0759 69 68 76 61 26 42
0800 0859 62 49 65 58 28 44
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
52/126
Section Three: Flight Level Delay Trends
!
We now connect airline turn times to departure delay times observed. Table 15 groups
carriers into four categories: all carriers, representing all reporting airlines during June 2010;
regional airlines (ExpressJet, Mesa, Pinnacle, Atlantic Southeast, SkyWest and Comair); Majors
(US Airways, Delta, Northwest, United, US Airways, Frontier, AirTran, Alaska and Hawaiian);
and finally Southwest. Southwest is isolated because its turn time and delay metrics are
significantly different from its peers.
Table 15 shows that (1) scheduled turn times for all carriers generally increase during the
evening hours, but delay minutes also increase; (2) regionals are more impacted by delay minutes
than mainline; and (3) Southwests fast turn times are directly connected to follow-on flight
impact as the day progresses.
Table15: Turn Time Distribution (20-240 minutes) by Time of Day, Southwest vs. Others
Source: Reporting Carriers, Domestic Flights, June 2010, DOT Part 234 (ASQP)
In Minutes All Carriers Regionals Only Majors (ex. WN) SouthwestDeparture Turn Delay Turn Delay Turn Delay Turn Delay
0001-0559 49.3 0.0 49.3 0.0
0600-0659 41.6 2.2 30.4 2.7 56.4 1.0 29.5 10.6
0700-0759 42.2 1.3 39.2 1.8 61.7 0.7 26.1 0.9
0800-0859 44.4 1.7 45.6 2.4 55.3 1.0 28.4 1.3
0900-0959 48.2 2.1 51.5 2.9 57.1 1.6 29.4 1.4
1000-1059 46.9 2.9 45.4 4.4 56.0 2.0 29.8 2.0
1100-1159 47.7 3.4 44.9 4.8 57.1 2.2 29.2 3.4
1200-1259 46.3 4.1 42.5 5.7 56.3 2.9 29.9 4.11300-1359 47.8 4.4 44.5 6.0 57.3 3.1 30.6 4.5
1400-1459 47.9 5.3 45.9 7.1 56.6 3.8 31.3 5.5
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
53/126
Section Three: Flight Level Delay Trends
Chart 11: Delay Minutes per Impacted Flight, By Scheduled Turn Time
Source: Reporting Carriers, Domestic Flights, June 2010, DOT Part 234 (ASQP)
How does Southwests turn time and delay pattern compare with a major airline peer?
To conduct a specific comparison, we analyzed Southwests scheduled turn times by time of day
against US Airways. We selected US Airways as a peer for three reasons: (1) similar exposure in
both the western United States (with shared hubs/focus cities at Phoenix and Las Vegas) as well
as prominent operations in Florida, the mid-Atlantic and the Northeast; (2) similar fleet
composition, with extensive narrowbody aircraft; and (3) US Airways operational performance
0.0!5.0!
10.0!15.0!20.0!25.0!30.0!35.0!40.0!45.0!50.0!
20! 25! 30! 35! 40! 45! 50! 55! 60! 65! 70! 75! 80! 85! 90! 95! 100! 105! 110! 115! 120!Carrier! Weather! Airspace! Late Inbound!
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
54/126
Section Three: Flight Level Delay Trends
!
Table16a: Southwest Turn Time Distribution (20-240 minutes) by Time of Day
Source: Reporting Carriers, Domestic Flights, June 2010, DOT Part 234 (ASQP)
Row Labels Sch. Turn Carrier Weather Airspace Security Late Inb.0600-0659 29.5 0.3 0.0 30.5 0.0 58.5
0700-0759 26.1 12.5 2.4 10.8 0.0 14.3
0800-0859 28.4 7.4 6.3 14.4 0.0 18.8
0900-0959 29.4 8.0 6.2 9.6 0.1 16.5
1000-1059 29.8 8.3 3.5 10.4 0.1 20.0
1100-1159 29.2 8.3 1.2 6.7 0.1 26.7
1200-1259 29.9 8.4 2.6 7.4 0.1 26.8
1300-1359 30.6 7.3 2.4 8.6 0.0 26.71400-1459 31.3 7.3 4.5 8.0 0.1 26.0
1500-1559 30.8 7.9 5.5 7.7 0.2 26.5
1600-1659 31.5 8.4 5.0 8.1 0.0 29.1
1700-1759 30.9 7.8 4.4 8.1 0.0 32.1
1800-1859 30.6 7.6 3.5 5.4 0.0 36.9
1900-1959 29.6 8.3 1.9 4.8 0.0 37.3
2000-2059 28.8 8.4 0.8 3.6 0.0 38.1
2100-2159 27.6 9.1 1.1 3.2 0.0 34.9
24 Hours 30.0 8.0 3.3 6.8 0.1 31.0
Table 16b: US Airways Turn Time Distribution (20-240 minutes) by Time of Day
Source: Reporting Carriers, Domestic Flights, June 2010, DOT Part 234 (ASQP)
Row Labels Sch. Turn Carrier Weather Airspace Security Late Inb.
0700-0759 66.7 10.5 0.0 21.1 0.0 3.2
0800-0859 63.0 13.1 0.0 19.8 0.0 3.10900-0959 67.0 13.0 0.7 13.0 0.3 8.4
1000-1059 63.9 19.5 0.0 14.2 0.2 5.9
1100-1159 69 1 10 3 0 1 17 9 0 0 9 6
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
55/126
Section Three: Flight Level Delay Trends
between 20 minutes and 90 minutes. As Chart 12 shows, JetBlue and American have similar
aircraft turn weightings, grouping most aircraft turns between 40 and 65 minutes. Southwest
skews to faster turns, with most occurring between 20 and 40 minutes, while US Airways
schedules longer turns on average.
Chart 12: Turn Time Distribution (20-240 minutes) by Flight Total
Source: Reporting Carriers, Domestic Flights, June 2010, DOT Part 234 (ASQP)
3.8.2
Increases in Aircraft Turn Times, 2005-2010
Based on the trends observed, we believed that a significant improvement in on-time
0%!5%!
10%!15%!20%!25%!30%!35%!40%!45%!
20! 25! 30! 35! 40! 45! 50! 55! 60! 65! 70! 75! 80! 85! 90!American! JetBlue! US Airways! Southwest!
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
56/126
Section Three: Flight Level Delay Trends
!
airline, American, Continental, AirTran, US Airways and Southwest showed significant increases
in turn times. Deltas reduction was likely the result of integrating Northwest Airlines in 2008-
2009, which had used shorter turn time strategies than Delta at its Minneapolis and Detroit hubs.
JetBlue and United showed no meaningful changes.
Table 17: Change in Average Scheduled Turn Time, By Airline, 2005-2010
Source: Reporting Carriers, Domestic Flights, First Wednesday of June, DOT Part 234 (ASQP)
2005 2006 2007 2008 2009 2010 Change
American 54.1 53.3 52.0 56.6 57.6 59.1 9.3%
JetBlue 56.2 55.6 51.5 54.2 51.6 54.8 -2.4%
Continental 62.4 64.0 59.6 63.8 67.8 72.3 15.8%
Delta (a) 64.9 65.1 64.6 64.9 62.3 61.6 -5.0%
AirTran 41.2 41.8 41.7 43.2 42.7 45.5 10.4%
United 63.7 61.0 59.5 62.4 63.7 63.7 -0.1%
US Airways 54.1 59.8 58.8 64.7 67.6 66.7 23.2%
Southwest 26.4 26.3 27.6 29.1 28.6 30.3 14.8%Group 48.2 48.0 46.7 48.7 48.3 50.2 4.2%
(a) Delta turns impacted by Northwest integration 2009-2010
We then isolated the change in aircraft turn times by airport, across all reporting airlines.
The changes ongoing at former Northwest hubs as Delta adapted former Northwest strategies to
its combined network were evident. Detroit, Memphis and Minneapolis all showed 40% orgreater increases in scheduled turn times. Phoenix and Las Vegas showed significant increases
due to changes to US Airways turn strategies and on-time performance focus Increases were
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
57/126
Section Three: Flight Level Delay Trends
Table 18: Change in Average Scheduled Turn Time, By Airport, 2005-2010
Source: Reporting Carriers, Domestic Flights, First Wednesday of June, DOT Part 234 (ASQP)
Airport 2005 2010 Change Airport 2005 2010 ChangeDTW 42.2 66.2 57.0% IAD 57.5 60.7 5.5%
PHX 34.9 52.6 50.8% DFW 53.3 55.9 4.9%
MEM 43.6 62.9 44.2% CLT 62.2 65.0 4.5%
MSP 47.4 66.8 40.9% MIA 63.9 66.5 4.1%
BWI 32.3 41.8 29.1% MCO 44.7 46.5 3.9%
JFK 63.0 77.9 23.7% DCA 51.4 53.0 3.1%
MKE 39.2 48.3 23.4% LAX 54.5 56.2 3.0%
SMF 31.0 38.2 23.2% ATL 64.1 61.6 -3.9%
HOU 26.1 31.9 22.0% SEA 55.3 52.6 -4.9%
IAH 65.3 79.6 21.9% FLL 47.4 44.7 -5.8%
BTV 35.3 42.0 19.1% BOS 54.9 51.7 -5.8%
LAS 37.4 44.5 18.8% LGA 54.4 51.3 -5.9%
EWR 58.8 69.6 18.3% SLC 57.4 53.2 -7.4%
MDW 29.8 34.9 17.2% SFO 71.0 61.2 -13.9%
PHL 50.1 58.7 17.2% PIT 51.2 43.7 -14.7%
ORD 55.9 63.5 13.5% DEN 59.7 50.3 -15.7%
CLE 55.6 59.6 7.1% HNL 134.4 107.9 -19.7%
From this analysis, we conclude that airlines have internalized the risk from late
inbound flights at congested U.S. airports by consciously expanding aircraft turn times.
This buffer we believe explains why flight delays have declined significantly after the peak
in 2007 when both on time performance was at its worst and turn times at their shortest
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
58/126
Section Three: Flight Level Delay Trends
!
3.9 Schedule Padding
Schedule padding reflects strategic delay management by airlines, incorporating into
planned en-route (gate to gate) time the expected delays from controllable, uncontrollable and
follow-on factors. There are three primary reasons to build some level of flight delays into en-
route times, versus incurring a delay only on the impacted flights.
First, DOT press releases and media activity focus on airline arrival performance
measured against the scheduled arrival time, without consideration to mitigating factors (such as
the average en-route time on a given route). It is understandable that DOT seeks a single standard
to apply to all domestic flight operations, but this arrivals-based standard independent of other
factors creates a strong incentive for airlines to manage delays by ensuring all but extraordinary
situations will be internalized into their schedule. Airlines can create whatever on-time
performance they want to achieve by adding scheduled minutes to delay-prone flights, although
there is a costly trade-off in crew pay and aircraft utilization. In turn, placing high in the DOT on-
time list creates an objective standard for an airline to advertise and compare with its peers.
Second, customer flight connections are based on scheduled arrival and departure times.
For connecting itineraries, it is essential that minor variability in the inbound arrival time at the
hub is incorporated into the minimum connect time standards applied. Building a schedule buffer
into each flight minimizes the number of missed connections.
Third, passengers complain about late arrivals, but few complain about landing early,
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
59/126
Section Three: Flight Level Delay Trends
DOT. Data is from the DOT Part 234 ASQP focused on the differences between scheduled and
actual en-route times. We exclude flights delayed due to late inbound aircraft.
Chart 13 presents the average schedule pad by time of day, across the U.S. system for all
reporting carriers. The percentages denote the average of the differences between actual en-route
time and scheduled en-route time for each route. For example, during the 1300-1359 time block,
the average difference between scheduled and actual en-route time was 2.9%, with the actual time
faster. As the day progresses, and as late arriving aircraft compound irregular operations, airlines
build additional buffer into their flight schedules. Table 13 shows that airlines put a roughly 2-4%
buffer in their operation. We then grouped the difference between actual and scheduled en-route
time by the distance of planned flight. Table 19 below shows that as the flight distance decreases,
the schedule pad increases dramatically (due to the variability in taxi times, as we show later).
Chart 13: Schedule Padding (Normal Operations Scheduled vs. Actual Block Time)
All Carriers, Full Year 2010, by Departure Time Block
-1.9%
!
-2.5%
!
-3.0%
!-2.9
%!
-2.8%
!-2.8%
!-2.9
%!
-3.1%
!-2.9
%!
-2.8%
!-2.8
%!
-3.2%
!-3.3%
!
-3.8%
!-3.7%
!
!
0700-0759
! !
0800-0859
! !
0900-0959
! !
1000-1059
! !
1100-1159
! !
1200-1259
! !
1300-1359
! !
1400-1459
! !
1500-1559
! !
1600-1659
! !
1700-1759
! !
1800-1859
! !
1900-1959
! !
2000-2059
! !
2100-2159
!
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
60/126
Section Three: Flight Level Delay Trends
!
Chart 14 below graphs the information in Table 19 with the addition of a trend line, and
without the 100-mile groupings in the table. A pronounced curve is observed with deviation
increasing at transcontinental and US-Hawaii flights.
Chart 14: Spread in En-Route Performance by Distance of Route
Percentage difference between actual/scheduled block time ratios under normal and irregularoperation conditions (Full Year 2010)
0%!
5%!
10%!15%!
20%!
25%!30%
!
0! 500! 1,000! 1,500! 2,000! 2,500! 3,000! 3,500! 4,000! 4,500! 5,000!SpreadbetweenActual
andSchedluedEn-route
Time(Delayedvs.NotDelayed)
Distance of Flight (miles)!
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
61/126
Section Three: Flight Level Delay Trends
Both taxi-out and taxi-in times are impacted by physical factors as well, including (1) the
absolute distance from gates to the runway; (2) runway intersections that force aircraft to hold for
arriving or departing runway operations; (3) intersecting runways that restrict departure flows;
and (4) the amount of ramp and taxiway infrastructure available for run-ups, ground delay
programs and passenger services during extended tarmac waits.
To begin our analysis of taxi-out time variability in 2010, we collected average taxi-out
times for the full year across 13 key airports subject to long taxi times and congestion. We
grouped the average taxi-out times (excluding cancellations and first taxis before gate returns)
that resulted in a successful runway departure. Table 20a provides average taxi-out times,
demonstrating a minimum average of 14.6 minutes at DFW to a maximum of 27.6 minutes at
New York JFK. Taxi times at all airports increase through mid-day and afternoon hours. There
are significant increases during the 12pm-8pm time window at all three New York airports, where
taxi times at JFK routinely exceed 30 minutes.
Table 20b captures the variability of these taxi-out times. There is significant afternoon
variability in taxi times for Atlanta, Boston, Washington Reagan and Dulles, Chicago,
Philadelphia and all three New York airports, while Dallas, Denver, Houston and Los Angeles
remain within a narrower window of taxi times. This is illustrative of two factors: the
appearance of afternoon weather activity that stalls the departure flows from these airports, as
well as normal schedule peaks that are discussed more in Section 4.
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
62/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
63/126
Section Three: Flight Level Delay Trends
What is the importance of Tables 20a and 20b in airline planning? When determining
how large a schedule buffer is appropriate, airlines must assess not only the average taxi time, but
the deviation of taxi-out times from the mean. For example, an airline targeting an 80% on-time
performance with sole consideration to taxi-out time would estimate a taxi-out time at JFK of 58
minutes (mean plus one standard deviation) and add this to the estimated flight time and arrival
taxi-in time in order to calculate the expected block time. As Chart 15 below shows, achieving a
95% confidence interval on flight departures from these airports requires taxi-out time estimates
significantly in excess of the mean.
Chart 15: 95% Confidence Interval Taxi-Out Time, 24 Hours and Peak Hour
95th
Percentile Taxi-Out Time (95% of departures under) by Key U.S. Airport (Full Year 2010)
38.
5
35.
3
37.
0
3
0.
2
2
9.
8 3
9.
2
33.
9
28.
7
55.
3
26
.9
48.
7
33.
3 4
3.
6
44.
0
45.
6
45.
8
33.
0
2
9.
6
54.
5
42.
8
32.
2
78.
7
23.5
58.
8
43.
1
66.
9
ALL DAY! 6PM-7PM!
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
64/126
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
65/126
Section Three: Flight Level Delay Trends
Chart 16: 95% Confidence Interval Taxi-In Time, 24 Hours and Peak Hour
95th
Percentile Taxi-In Time (95% of arrivals under) by Key U.S. Airport (Full Year 2010)
In conclusion, we observe that there is significant variation in taxi-out and taxi-in
times across airports, and that taxi times can vary widely by time of day. At some airports,
including JFK, LGA, EWR and PHL, taxi times during the afternoon are much higher than during
the morning and late evening. At others, including DEN, DFW and IAH, taxi times are more
evenly distributed. Planners must take these differences into account when incorporating
expected taxi times into their scheduled en-route times. But how are these taxi times changing
over time?
29.
0
24.
0
25.
7
23.
5
23.
9
27.
8
25.
0
20.
8
38.2
21.
6
33.
1
26.
4
30.
1
34.
1
30.
1
33.
3
26.
8
24.
6
36.
8
32.
2
24.
7
56.
0
19.
1
40.
1
35.
1
44.
3
ATL! BOS! DCA! DEN! DFW! EWR! IAD! IAH! JFK! LAX! LGA! ORD! PHL!
ALL DAY! 6PM-7PM!
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
66/126
Section Three: Flight Level Delay Trends
!
Chart 17: Change in Taxi Times (In and Out) Excluding Cancellations and Diversions
Airport Set: ATL, BNA, BOS, BUF, BWI, CLE, CLT, DAL, DCA, DEN, DFW, DTW, EWR, FLL, HNL, HOU, IAD,IAH, JFK, LAS, LAX, LGA, MCO, MDW, MEM, MSP, OAK, ORD, PHL, SAN, SEA, SFO, SLC, STL, TPA
Table 22: Average Taxi-In Time by Airport (Minutes per Flight)
Full Year 2010, by Arrival Time BlockAirport Set: ATL, BNA, BOS, BUF, BWI, CLE, CLT, DAL, DCA, DEN, DFW, DTW, EWR, FLL, HNL, HOU, IAD,
IAH, JFK, LAS, LAX, LGA, MCO, MDW, MEM, MSP, OAK, ORD, PHL, SAN, SEA, SFO, SLC, STL, TPA
Averages (Minutes) 1995-1999 2000-2004 2005-2009 2010
Taxi In 5.7 6.3 6.8 7.0Taxi Out 15.2 16.2 17.0 16.2
0!2!4!6!8!10!
12!14!16!18!20!
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
MinutesofTaxiTime
Taxi In! Taxi Out!
S i Th Fli h L l D l T d
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
67/126
Section Three: Flight Level Delay Trends
The variability of taxi times on both arrival and departure, both in means
observed and in the standard deviations therein, are primary drivers of schedule
pads. Taxi times largely incorporate the weather- and airspace-related delay
factors reported by airlines. To achieve 80-95% confidence intervals in block
times, taxi times greater than 30 minutes must be incorporated for key
congested airports.
Taxi times (both in and out) have increased since 1995. This is partially due to
changes and general utilization of runway and en-route assets, but it is also due
to new runway construction and other physical factors.
To complete our analysis of changes in block times since 1996, we compiled scatter
charts for all routes in operation consistently between 1996 and 2010, by any reporting
airline. For each route, we collected the mean taxi-time (out + in) for all operators between
1996-2000 and 2007-2010, and then compared the two averages.
As Chart 18a shows, not all routes had longer taxi-times between 2007-2010 versus
1996-2000, but more than half did. The horizontal axis in Chart 18 represents the change in mean
values: 100% means that the averages in 1996-2000 and 2007-2010 are identical. More than
100% means that the mean 2007-2010 exceeds the mean value 1996-2000.
One can observe that the shorter the flight distance, the greater the (positive) change in
mean taxi times. We attribute this to two factors. First, shorter-haul flights shifted from mainline
S ti Th Fli ht L l D l T d
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
68/126
Section Three: Flight Level Delay Trends
!
15-Year Snapshot
Chart 18a: Average Taxi-Time by Route Distance
Mean Taxi-Time by Distance, Change 1996-2000 vs. 2007-2010 (>100% = 2007-2010 Longer)
Chart18b: 15 Year Change - Average Flight Time by Route Distance
Mean Actual Flight Time by Distance, Change 1996-2000 vs. 2007-2010(>100% = 2007-2010 Longer)
0!
1,000!
2,000!
3,000!
4,000!
5,000!
6,000!
0%! 100%! 200%!
RouteDistance
(Miles)
Change in Taxi Time (Average 2007-2010 vs. Average 1996-2000)!
The shorter the flight, the
more variable the taxi time
has become
S ti Th Fli ht L l D l T d
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
69/126
Section Three: Flight Level Delay Trends
It can be difficult, even with side-by-side comparisons of the changes in flight and taxi
averages, to observe precisely where the increase in block time originates. Cross-plotting the data,
however, shows that taxi-time changes are clearly driving the overall increase in en-route times,
and forcing airlines to increase block times. Chart 18c overlays the data.
Chart 18c: Change in Average Flight Time by Route Distance
Mean Actual Flight Time by Distance, Change 1996-2000 vs. 2007-2010
(>100% = 2007-2010 Longer)
0%#
50%#
100%#
150%#
200%#
250%#
300%#
0# 500# 1,000# 1,500# 2,000# 2,500# 3,000#
Change,
20082010(v
s.
19951997(
DIstance(of(Route(
Taxi#
Flight#
Section Three: Flight Le el Dela Trends
7/28/2019 Stateofusaviation Comprehensive Analysis of Airline Schedules & Airport Delay
70/126
Recommended