Canceled flights. Airplanes idling on runways as pilots wait for the
signal to depart. For the second straight summer, passengers flying
on U.S. airlines endured an increasing number of flight delays, sometimes
having to miss scheduled events or sleep overnight in air terminals.
Why is air traffic approaching gridlock? Reasons given are the increasing
number of passengers and flights as a result of lower fares, severe
thunderstorms, insufficient runway space, airline employee strikes,
and an antiquated air traffic management system. Yet, the national air
system has excelled in keeping big jets from flying into one another.
In 1991, when Lee Berry,
Jim Rome, and other ORNL researchers began analyzing air traffic operations
for the Federal Aviation Administration (FAA), they predicted that the
worst air congestion at the end of the decade would occur over the Cleveland,
Ohio, area. "That has proven to be the case," says Teresa Rose, program
manager for aviation research at the National Transportation Research
According to Rose,
the ORNL group hopes to help the FAA reduce "en route airline congestion"
by developing a computer model. Berry, Rome, and Ron Lee have received
FAA funding to develop a model to predict the probability of airplane
delays for different sectors of air space.
"We will look
at ways to avoid delayed landings caused when airplanes are forced to
take a longer route to better space out planes headed for the airport,"
Rose says. "The model will take into account severe weather, flight
delays, and increased spacing of plane arrivals and departures to allow
air traffic controllers time to adjust to upgraded equipment. Our model
should help airlines select better flight times and routes to reduce
en route delays."
have completed several studies of air traffic congestion. With American
Airlines, they modeled delay propagation by simulating delays and introducing
the concept of a delay multiplier. "A flight that arrives late delays
not only its passengers but also the crew and equipment needed for later
flights, causing 'downstream' delays," Rome says. "A minute of delay
occurring early in the day can cause over 10 minutes of downstream delay."
At major hubs
an airline might have hundreds of landings in a day. Using data from
Northwest Airlines and assisted by Lockheed Martin Management and Data
Systems, Simon Rose, Rome, and Lee performed a cost-benefit study for
NASA to determine whether the airline might save money by swapping landing
slots among its own flights. During the final descent phase of flight,
planes can speed up or slow down by as much as 10 minutes to enable
"Moving a flight
a few minutes can mean the difference between lots of missed connections
and just a few missed connections involving both passengers and crew
members," Rome says. "Changing a flight from 30 minutes late to 20 minutes
late can remove delay costs entirely."
For the study,
cost models and an optimum resequencing algorithm were developed. Using
the results of the study, the ORNL researchers concluded that the U.S.
airline industry would save $75 million a year if this strategy were
operationally feasible. About 30% more would be saved if unused landing
slots were employed.
To increase their
numbers of daily flights (airport capacity), many airports used the
controversial practice of allowing one aircraft to land on one runway
and stop short of a second, intersecting runway, permitting another
plane to simultaneously land on or take off from that runway. In 1999
the Airline Pilots Association (ALPA) opposed this practice as being
potentially unsafe and threatened a boycott unless the safety margins
were increased. Berry recently provided a better estimate of the cost
of eliminating the procedure and, thus, allowing fewer flights. This
estimate improved the basis for making decisions about trade offs between
increased safety margins and reduced airport capacity. The arguments
were presented to the major airlines' operations managers, ALPA, and
the FAA. FAA and ALPA worked out a compromise that both improved safety
margins of such simultaneous operations and retained much of the increased
is being studied by ORNL researchers Joe Cletcher, Gary Mays, Mike Poore,
and Simon Rose. They are applying techniques developed for the nuclear
industry to aviation. The goal is to identify accident precursors by
coding the chains of events that contribute to aviation accidents and
sample of airplane landings at Los Angeles International Airport
shows an example go-around flight that occurs when a pilot is
informed that the airplane is coming in too fast at the wrong
anglewhat is known as a "missed approach."
In another safety
analysis study for the FAA, Berry worked with Austin Digital, Inc.,
to identify situations that could lead to safety problems, such as missed
approaches to runways (e.g., an airplane coming in too fast at the wrong
angle). The collaborators modified digital flight data analysis software
(used for "black box" data) so that it could analyze radar data. They
demonstrated its use to high FAA officials in March 2000. "This tool,"
Rose says, "is the first to use radar data for safety analyses."
Pilots Association (ALPA)