Microburst Recovery For Jet Transport Aircraft
Microburst Recovery For Jet Transport Aircraft
Microburst Recovery For Jet Transport Aircraft
9-2003
This thesis is brought to you for free and open access by Embry-Riddle Aeronautical University – Daytona Beach at
ERAU Scholarly Commons. It has been accepted for inclusion in the Theses - Daytona Beach collection by an
authorized administrator of ERAU Scholarly Commons. For more information, please contact commons@erau.edu.
MICROBURST RECOVERY FOR JET TRANSPORT AIRCRAFT:
A COMPARISON BETWEEN CONSTANT AND VARIABLE PITCH GUIDANCE
TRAJECTORIES
by
Mark Cadmus
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy
submitted. Broken or indistinct print, colored or poor quality illustrations
and photographs, print bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
UMI
UMI Microform EP31887
Copyright 2011 by ProQuest LLC
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106-1346
MICROBURST RECOVERY FOR JET TRANSPORT AIRCRAFT:
A COMPARISON BETWEEN CONSTANT AND VARIABLE PITCH GUIDANCE
TRAJECTORIES
by
Mark Cadmus
This thesis was prepared under the direction of the candidate's thesis committee chair,
Dr. Michael E. Wiggins, Department of Applied Aviation Sciences, and approved by the
Thesis Review Committee. It was submitted to the Department of Applied Aviation
Sciences in partial fulfillment of the requirements for the degree of
Master of Science in Aeronautics.
THESIS COMMITTEE:
r. Francis Ayers
Thesis Member
/Vtflre£u E-JL^JU
4-
Mr. Theodore Beneigh
neigh
Thesis Member
// k/o\
Department Chair, Applied Aviation Sciences Date '
ii
ACKNOWLEDGEMENTS
My sincere appreciation is extended to all those who have assisted in bringing this
project to fruition. The individuals who played a predominant role would like to remain
anonymous, they were; pilots Jeff Jaslow, John Liniger, and Ray Russell; engineers Steve
Ferro and Bill Fairer. Christopher Herbster provided direction in meteorology and gave
freely of his time, as did Bill Baker in reviewing the initial writing.
Special recognition is given to Steve Hampton for bringing the Link Fellowship to
the university and providing many students the opportunity to continue in their chosen
field of research. This project was supported in part by the Link Fellowship.
University, their continual rejection of my advances ensured that plenty of time was
111
ABSTRACT
Year: 2003
The purpose of this research was to compare, in a simulator, the safety of a variable pitch
strategy with the established constant pitch strategy in transitioning through a microburst
mathematical and computer studies of microburst penetrations, the variable pitch strategy
provided a greater recovery altitude than the constant pitch strategy. A Boeing 737 level
computer generated data were collected. "Safety", defined as the maximization of the
minimum altitude experienced by the aircraft during the recovery phase of the microburst
encounter, was statistically greater for the constant pitch maneuver. An improved
microburst model and a flight director steering command are recommended for continued
IV
TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS iii
ABSTRACT iv
LIST OF TABLES x
LIST OF FIGURES xi
Chapter
I INTRODUCTION 1
Definition of Terms 9
Historical Context 13
Eastern 66 14
Allegheny 121 25
Delta 191 35
USAir 1016 40
Additional Accidents 44
v
Case History Conclusions 46
Microburst History 48
Classical 48
Renaissance 49
Modern 52
Microburst Research 53
NMROD 53
JAWS 53
NAS 55
CLAWS 55
MIST 57
AWDAP 57
Microburst Meteorology 60
Physical Properties 61
Size 61
Wind Speed 62
Duration 64
Frequency 64
Environmental Conditions 65
Atmospheric Properties 65
vi
Motive Forces 66
Dry Microbursts 69
Wet Microbursts 70
Flight Instruments 73
Aircraft Behavior 74
Lift 76
Dynamic Stability 78
Speed Stability 83
Performance 85
F-factor 87
Aircraft Categories 88
Thrust 92
Pitch 92
Configuration 94
vii
Climb Phase 103
Evaluation 103
Design 112
Participants 116
Instruments 122
Data 125
Procedures 126
IV RESULTS 134
V DISCUSSION 143
viii
Descriptive Statistics 149
VI CONCLUSIONS 158
REFERENCES 163
APPENDICES
IX
LIST OF TABLES
Table Page
Microburst Encounter 45
4 Altitude and Airspeed of Simulator Trials for Constant and Variable Pitch
Maneuvers 136
6 Analysis of Variance for Altitude with Pilots and Maneuver as Factors .... 138
Factor 140
Factor 141
11 Post Hoc Comparison between Pilots for Altitude in Variable Pitch Maneuver
142
x
LIST OF FIGURES
Figure Page
10 Boxplot of pilot versus altitude for the constant pitch maneuver 145
12 Boxplot of pilot versus altitude for the variable pitch maneuver 150
13 Regression plot of variable pitch maneuver: Airspeed predicting altitude ... 153
xi
22 Pilot 1, constant pitch maneuver, run 4
xii
45 Pilot 2, variable pitch maneuver, run 3 209
Xlll
1
CHAPTER 1
INTRODUCTION
Microbursts are a violent form of low-level wind shear and pose a substantial
threat to aircraft in flight. Wind in the microburst descends at high velocity and spreads
radially outward, forming an area of horizontal and vertical wind shear. Aircraft
penetrating this wind shear experience a reduction in climb capability. This performance
degradation can be catastrophic when the microburst is encountered at the low altitudes
and airspeeds required during the takeoff and approach phases of flight. Low-level wind
shear and microbursts are responsible for the deaths of 665 people in 25 airline accidents
since 1964 (McCarthy, 1996, p. 9). The preponderance of the fatalities, 512, occurred
Avoidance of the microburst is the safest measure for arriving and departing
aircraft. In recent years there has been considerable improvement in the detection and
have occurred, resulting in fatalities, and inadvertent microburst encounters are likely to
happen again. The escape maneuver is the final defense in surviving a microburst
encounter.
maximum performance of the aircraft. The current escape maneuver, keeping "the
airplane flying as long as possible in hope of exiting the windshear [sic]" (Federal
Aviation Administration [FAA], 1996, p. 1), does not produce an optimum trajectory. An
alternate strategy, exiting the wind shear in the minimum amount of time, decreasing the
phase. Extension of the research to the approach to landing phase is a natural evolution.
The advent of the flight data recorder (FDR) led investigators to consider the
hazards of low-level wind shear to aircraft. An Iberian DC-10 equipped with an enhanced
digital flight data recorder (DFDR) crashed while landing at Boston's Logan
International Airport on December 17, 1973. Investigators were able to construct the
approach environment and flight path of this ill-fated tri-jet from the 96-parameter
from the presence of wind shear. Grossi (1988) of the National Transportation Safety
Board (NTSB) declared, "there is little doubt that without the DFDR data, this
investigation would not have yielded this level of insight into the windshear [sic]
phenomenon, and in fact may not have identified it as a factor" (p. 459).
Using the data obtained from the DFDR, five appropriately rated pilots flew 48
simulated approaches. In part, these simulations were conducted to "examine the flight
conditions that confronted the flight crew" (National Transportation Safety Board
[NTSB], 1974a, p. 12). The flight simulator trials demonstrated the serious problem that
wind shear posed to this aircraft (p. 20). "The examination of DFDR data, including the
data reproduced in the DC-10 flight simulator, provided more positive evidence of the
wind conditions along Flight 933's final approach profile" (p. 19). The knowledge gained
from the data generated by this accident led to the first recommendations on wind shear
to aviation safety. (Grossi, 1988, p. 459). These recommendations were directed toward
Specialists in meteorology and atmospheric science augmented the data generated from
FDRs and flight simulators, refining old theories, and developing new theories. The
accident investigation of a Boeing 727 in New York became the genesis in the
heavy rain on June 24, 1975, when it entered a horizontal wind shear coupled with a
vertical downdraft. This combination of wind caused the aircraft to crash short of the
runway with 113 fatalities. "After an exhaustive analysis of the FDR data and eyewitness
accounts, [Fujita] called this windsystem [sic] the 'downburst'" (Fujita, 1985, p. 2). The
downburst, "a localized, intense downdraft with vertical currents exceeding a downward
speed of 12 fps or 720 fpm" (Fujita, 1976, p. 50), is strong enough to blow down a jet
This concept was considered controversial (Rosenfeld, 1999) and not until further
of these accidents brought new data and new research into the field. Many of the ensuing
developments were a benefit to the aviation community, including ground based sensing
and forecasting, pilot education, and practical pilot training. In combination, these
4
paramount.
Low-level wind shear is the parent category of downbursts, which are further
dry or wet, depending on the presence of rain at the surface. Dry microbursts occur
contact-is common due to the large spread between temperature and dew point. In the
eastern states, the atmosphere in the summer months tends to have a higher relative
predominance of wet microbursts (Atkins & Wakimoto, 1991, p. 471; see also Nelson &
atmospheric hydrostatic pressure can initiate the microburst. This pressure differential
Precipitation drag can also induce a microburst by accelerating the air in the vertical
plane. Evaporative cooling provides negative buoyancy to the air, which can cause the
formation of both wet and dry microbursts (Stull, 2000, pp. 340-341). Even with the
are on an experimental basis via the Internet from the National Environmental Satellite
In the initial stages of the microburst air descends from the cloud base creating a
vertical shaft of wind. This wind induces a horizontal ring vortex on the perimeter of the
wind shaft that descends with the vertical column of wind (Fujita, 1986, p. 56). Several of
these horizontal vortices may develop. When the vertical wind contacts the ground it
spreads radially into an outflow (see Figure 1). The vortex ring encircling the outflow
also expands, increasing the wind velocity near the ground (Caracena, Holle, & Doswell,
1989, p. 12). The outflow front is oftentimes the only visible indication of a microburst,
as colloidal particles carried by the outflow wind are distinguished from their
surroundings.
Figure 1. Idealized microburst flow with descriptors. Note. From Pilot Windshear Guide
(p. 8), by FAA, 1988, Washington, DC: FAA.
The outflow is of particular importance in classifying the downburst phenomenon.
Damaging outflow winds extending greater than 4km are considered macrobursts, while
those winds that do not exceed 4 km are microbursts (Fujita, 1985, p. 8). Microbursts
have tighter wind shear gradients and are of stronger intensity than macrobursts, and
stability. The combination of horizontal and vertical wind shear reduces lift through a
reduction in relative airflow and angle of attack for the aircraft. This loss is compounded
by the speed instability that may be present during a microburst encounter. In the region
of speed instability the aircraft requires additional thrust to offset the increased drag of
the slower speed. Speed instability thus reduces the thrust available for increasing altitude
or airspeed.
trim state (Cook, 1997, p. 119). Microburst winds excite both the short and long period
modes of longitudinal stability. The short period mode is a quick oscillation in the pitch
attitude of the aircraft and has little debilitating effect on the flight path. The long period
mode, or phugoid, causes oscillations in both the airspeed and the altitude of the aircraft.
The oscillatory nature of the phugoid can cause the airspeed to decay very rapidly.
The altitude variations are also impairing and can cause a "premature impact with the
ground short of the runway" (McCarthy, Blick, & Bensch, 1979, p. 48). The phugoid is
lightly damped in transport aircraft and can be excited by the variable winds associated
with a microburst, as documented in numerous studies (e.g., Frost, Turkel, & McCarthy,
1982; McCarthy et al., 1979; McCarthy & Norviel, 1982; Sherman, 1977). Combined
7
with the performance reducing wind shear, the phugoid oscillation makes the microburst
Federal Aviation Administration (FAA) contracted with the Boeing Company to develop
an escape profile (FAA, 1988, p. ii). The resultant procedure is known as a constant pitch
maneuver as it constrains the initial pitch attitude of the aircraft to 15°. With a high pitch
attitude and resultant low airspeed, the FAA procedure increases the exposure of the
simulators and mathematical models (e.g. Dogan & Kabamba, 2000; Hinton, 1988; 1989;
testing theories (Ramsey, 1992, p. 11-2). As flight simulators are able to model a
theories, Boeing used aircraft simulators to develop procedures for escape maneuvers
(Higgins & Roosme, 1977; FAA, 1987, p. 12), and in developing the Pilot Windshear
Guide for the FAA (FAA, 1988, p. ii). Researchers at the National Aeronautics and Space
survivability, during the takeoff phase, of the variable pitch escape maneuver compared
8
with the FAA recommended constant pitch maneuver. The principle investigator in these
simulator trials advised, "extension of the work to the approach-to-landing case is also
necessary" (Hinton, 1988, p. 10). Of the 29 air carrier accidents attributable to wind
shear, as identified by McCarthy, only 3 were in the takeoff phase (1996, pp. 8-9). With
the preponderance of accidents occurring in the landing phase, it follows that the research
A flight simulator study comparing the constant pitch guidance strategy with the
variable pitch guidance strategy, in terms of altitude loss during the approach to landing
abort maneuver, is the first phase in providing insight into the applicability of the variable
and flight safety of jet transport aircraft exposed to the microburst phenomenon. This
guidance strategy through a microburst encounter provides for greater safety than the
transport, with a microburst wind field program was employed to test these conclusions
in a dynamic environment.
variable pitch guidance strategy with the established constant pitch guidance strategy in
phase of flight. The safety of the maneuver was statistically evaluated in terms of altitude
Definition of Terms
speed of 12 fps or 720 fpm at 300 ft above the surface. This value corresponds to
Macroburst - A large downburst with its outburst winds extending in excess of 4 km (2.5
Microburst - A small downburst with its outburst of damaging winds extending only 4
km (2.5 miles) or less. In spite of its small horizontal scale, an intense microburst
could induce damaging wind as high as 75 m/sec (168 mph) (Fujita, 1985, p. 8)
Outburst Center - The nadir point of a downburst where the vertical air current hits the
surface and spreads out violently. The fastest spreading flow is seen in the
direction of the cell motion. Environmental flows, such as sea breeze and adjacent
cells distort the outburst current. Depending upon the flight path relative to an
Phugoid - Lightly damped low frequency oscillation in speed coupling into pitch attitude
Recovery altitude - The lowest altitude, above ground level, recorded by the simulator
Wind Shear - A change in wind speed and/or wind direction in a short distance resulting
The research was conducted with the awareness of several limiting factors.
Simulators replicate an aircraft to the best that technology has to offer at the time it is
built and subsequently upgraded. The simulator response is based on both objective and
subjective data, and is hampered by latency, transport delay, and noise in the system.
Simulators accurately replicate an aircraft's response when it is operated well within the
performance envelope. As the parameters of the envelope are approached, the response of
the simulator losses fidelity and utilizes a more subjective data routine. In microburst
research much of the data required are at the edge of the envelope.
Avoidance of the microburst is certainly the safest maneuver, but out of necessity
for data acquisition, evasion of the microburst was not practiced in the simulator.
Additionally, the element of surprise was neither present nor considered-every approach
assumed, the participating pilots being generally aware of where the microburst began,
Flight below glide slope was not penalized; however, the limits of survivability
were defined by deviation below the altitude corresponding to field elevation and
The simulated microburst included neither turbulence, nor the effects of rain. This
does not deviate from the observed environment where quite often the microburst is dry
shearing action (FAA, 2003a), the significance to aviation lying in its degrading effect on
aircraft performance, and hence flight safety. Low-level wind shear, that which occurs
within 500 meters of the surface, is particularly dangerous for departing and arriving
aircraft (International Civil Aviation Organization [ICAO], 1987, p. 1). The most violent
form of low-level wind shear, the microburst, is "strong enough to blow down a jet
Microbursts diminish the lift, stability, and the climb capability of aircraft. The
tight wind shear gradients in the microburst lead to rapid changes in the wind vector and
may exceed the inertial capabilities of an aircraft to maintain flight (Caracena et al., 1989).
Initially, the escape was based on the traditional go-around procedure of holding airspeed
and if necessary, allowing a decay to stick-shaker speed to avoid terrain (FAA, 1979,
^J7.a.5). After several microburst accidents, the escape procedure changed from airspeed
to pitch control. The pitch attitude of the aircraft is now set at 15° and raised or lowered
as required to respect intermittent stick shaker (FAA, 1988, p. 46). The advent of
maneuvers differ from the FAA escape procedure and, in mathematical and flight
simulation, yield less altitude and airspeed loss, providing for a greater probability of
survival in the event of an inadvertent microburst encounter (e.g., Dogan & Kabamba,
2000; Hinton, 1988; Miele et al, 1987; Mulgund & Stengel, 1992b).
13
Historical Context
aircraft accidents. It was not until the mid 1970s that the phenomenon was first postulated
as a causal factor in the deaths of 113 people (Fujita, 1976). The concept was not well
received (Rosenfeld, 1999), and many in the industry clung to previous beliefs,
discounting the ferocity of the downburst. As more accidents attributed to these winds
occurred, more information became available, and the concept of the microburst took
hold. A retrospective analysis indicates that low-level wind shear, the parent category of
microbursts, is responsible for the deaths of at least 665 people in 29 American air-carrier
The sensationalism of aircraft accidents obscures the fact that low-level wind
shear has been documented throughout history. In ancient times, Aristotle considered the
phenomenon of wind shear in his discourse Meteorology (Berlin translation), and during
the Renaissance, an Oxford don relayed an accurate description of the microburst and its
debilitating effects on maritime activities (Bohun, 1671). The modern era brought a new
and understood.
Public and Congressional concern over the spate of microburst induced aircraft
accidents released grants to the FAA, the National Science Foundation, the National
Oceanic and Atmospheric Administration (NOAA), and others, to initiate the rigorous
study of the downburst (National Research Counsel [NRC], 1983, p. 1). These projects
carried whimsical titles-NIMROD, JAWS, CLAWS, and MIST-which belied their most
dynamic nature of the downburst. Each accident has enlarged the knowledge base and
became the catalyst for the downburst theory, though it was not the first aircraft to
Eastern 66
On the summer afternoon of June 24, 1975, an Eastern Airlines Boeing 727-225
was approaching JFK international airport as a scheduled flight from New Orleans.
Numerous scattered thundershowers delayed inbound aircraft, and after holding, Eastern
66 was vectored for an instrument landing system (ILS) approach to runway 22L. Slight
deviations around rain showers had the 727 intercepting the localizer, while a company
Eastern 902, a Lockheed L-1011, reported to the final controller "...we had ... a
pretty good shear pulling us to the right and ... down and visibility was nil..." (NTSB,
1976, p. 3). The Boeing crew, listening on the same frequency, was incredulous of the
pilot report transmitted, a crewmember stating: "I wonder if they're covering for
The L-1011 encountered a wind shear that reduced its airspeed by 24 knots. A
positive climb was not established until over 200 feet of altitude was lost and abnormal
amounts of pitch and power were employed (Fujita, 1976, p. 23). The wide-bodied jet
started climbing just 60 feet above the terrain (Fujita, 1985, p. 37). Eastern 902 was not
15
the first to encounter or report the wind shear; Flying Tiger 161, a DC-8 aircraft, had just
The effect of the wind shear on the DC-8 was witnessed by a Pan Am B-707
captain who "thought that the pilot must have been like a cat on a hot tin roof, trying to
save his airplane" (cited in Fujita, 1985, p. 36). The Flying Tiger pilot stated that he
estimated conditions to be so severe that he would not have had the performance required
to execute a missed approach, hence he elected to carry out a landing (NTSB, 1976, p. 5).
As they were taxiing, the captain of Flying Tiger 161 reported to JFK tower, "I just
highly recommend that you change the runways and... land northwest, you have such a
tremendous wind shear down near... the ground on final" (cited in NTSB, 1976, p. 5).
The tower controller decided no change in landing direction was necessary, as the
surface weather report was indicating winds 210° at 7 knots, almost aligned with the
runway. The Flying Tiger captain commented, "I don't care what you're indicating. I'm
just telling you that you have such a dangerous wind shear on the approach that you
should change the traffic to land to the northwest" (Bliss cited in Moldrem, 1996, p. 303).
Neither Eastern aircraft, both on approach frequency, were privy to the comments
made by the DC-8 captain transmitting on tower frequency. Eastern 902, in the go-
around, was asked if they would classify their encounter with the wind shear as severe, to
which they responded 'affirmative' (NTSB, 1976, p. 52). The next transmission was
Descending through 500 feet, Eastern 66 entered an area of heavy rain, the
windshield wipers were positioned to high, but the visibility remained restricted. At a
lower altitude the captain reported the airfield in sight (Fujita, 1976, p. 41), and with a
16
relatively high indicated airspeed of 150 knots, the jet was only seconds from the runway.
Nearing the threshold, the winds changed abruptly, and the flight path and airspeed
decayed rapidly. The first officer, who was flying, called for takeoff thrust to arrest the
descent. The command was issued too late, and the aircraft continued descending, with
The aircraft succumbed to its mortal wound, and sliding through additional
lighting towers, disintegrated piece by piece. The main wreckage area came to a rest on
Rockaway Boulevard, 1400 feet from the initial contact point and 1000 feet shy of the
runway 22L threshold. The official report determined "the accident was not
survivable..." (NTSB, 1976, p. 39). In an incredibly gallant effort by fire and rescue
personnel, who were on scene within 2 minutes of the aircraft accident, 11 of the 124
Early in the investigative process the role of weather was speculated in the demise
of Eastern 66. Time magazine initially reported the accident under the title 'A Fatal Case
of Wind Shear' (1975, July 7, p. 9). The aviation oriented periodical, Aviation Week and
Space Technology, proclaimed the NTSB "were pursuing wind shear as one of the most
likely factors in the Eastern Airlines Boeing 727 crash..."(1975, June 30, p. 26).
Speculating low-level wind shear the most credible culprit, Eastern Air Lines retained an
that was pushing through the mid-western states. Braniff 250, a BAC 1-11 aircraft, broke
Fujita was able to demonstrate that the BAC 1-11 had just transitioned the fine-line, or
wind-shift line, at a time, location, and altitude most favorable for the development of
horizontal and vertical vortices (NTSB, 1968, p. 35). In his report to BAC, Fujita advised
against flying through this area, and rather prophetically also "against flying through
thunderstorms in areas of heavy precipitation where vertical draughts (sic) were bound to
Later, while investigating the outbreak of tornados that occurred on April 3-4,
1974, Fujita documented debris fields that did not have a rotational component, yet were
obviously a result of high-speed wind damage. "Some distance away from the tornado
paths, trees in the forests were blown over in radial directions, as if they had been blown
outward" (Fujita, 1976, p. 44). From these observations the concept of the downburst
emerged. This theory, accounting for tangible evidence, postulates that a strong
downdraft colliding with the ground spreads in an outburst of damaging winds. Armed
with this knowledge, Fujita was provided "with the courage to investigate the Eastern 66
accident" (Fujita, 1985, p. i). To be determined was whether Eastern 66 penetrated one of
these downbursts and was subsequently blown into the ground, or if a more benign
based on the data acquired from the FDRs of the penetrating aircraft. These studies were
initiated to examine the flight conditions that probably existed at the time, and to
18
determine the difficulties that a flight crew has in recognizing the development of an
unsafe condition (NTSB, 1976, p. 18). The analog flight data recordings of Eastern 66
and the Flying Tiger DC-8 did not provide the detailed information required to determine
exact wind velocities. The DFDR from Eastern 902, being digital and recording more
parameters, provided the basis for dissecting the wind into vertical and horizontal
approaches were conducted, of which 18 resulted in impact with the approach lights
(NTSB, 1976, p. 19). When applying power, most pilots did not add enough, and they
were reluctant to interrupt their scan to verify engine pressure ratio (EPR). Additionally,
several pilots used a pitch attitude lower than commanded by the flight director,
commenting that the backpressure required on the control column was more than they
The simulator studies did confirm the difficult situation in which the flight crew
Relieved they might have crashed during actual flight" (NTSB, 1976, p. 20). Aside from
any other issues, the meteorological conditions on approach overwhelmed the ability of
The NTSB determined in part, "the probable cause of this accident was the
aircraft's encounter with adverse winds associated with a very strong thunderstorm
located astride the ILS localizer course, which resulted in a high descent rate into the
non-frangible approach light towers" (NTSB, 1976, p. 39). The Safety Board did not
conditions experienced by Eastern 66. Detailed analysis of the FDRs, along with weather
sustained two separate headwind gusts of 25 and 28 feet per second (fps) as it entered an
area of vertical wind. The headwind then decreased to 7 fps while a vertical wind of 21
fps occurred (Fujita, 1976, p. 41). This caused the aircraft to descend below the glide
slope at 300 ft above ground level (AGL) and into the approach light stanchions.
From the mapping of this weather pattern emerged confirmation of Fujita's new
theory, and substantiation of vertical winds greater than previously held possible.
Introducing an operative meteorological term into the lexicon of aviation, Fujita coined
the word downburst: "a localized, intense downdraft with vertical currents exceeding a
downward speed of 12 fps or 720 fpm at 300 ft above the surface" (Fujita, 1976, p. 50).
This downward velocity corresponds to a descent rate typical of what transport category
In conjunction with the term downburst came the term outburst center; "the nadir
point of a downburst where the vertical air current hits the surface and spreads out
violently" (Fujita, 1976, p. 50). An aircraft traversing the outburst center experiences a
the experience that befell Eastern 66. Not all were convinced of these unorthodox ideas,
1946, and in Ohio in 1947. From these studies, downdrafts were hypothesized to decrease
intensity from 10 fps at 4,000 ft altitude to zero velocity at ground level (Byers &
Braham, 1949). According to this theory, vertical winds dissipated rapidly with height,
and a cushion of air existed near ground level. This cushion would prevent an aircraft
from being driven into the ground by wind (Melvin, 1986, p. 49).
downdrafts exhibit on aircraft. Significant vertical winds at low altitude could drive a jet
airliner into the ground, thus dismissing the fallacy of a cushion of air.
Fujita's paper Spearhead Echo and Downburst near the Approach End of a John F.
Kennedy Airport Runwayf New York City. This publication, available through the Eastern
Airlines Flight Safety Department or the University of Chicago, was popular enough to
necessitate an additional printing just six months after the original 2000 were published
(Fujita, 1985, p. 45). Many airlines incorporated Fujita's research and publication into
their own flight training departments (NTSB, 1986, p. 52). Slowly the knowledge gained
In light of the new theories of extreme vertical winds, which have the potential to
dramatically degrade aircraft performance, the Air Line Pilots' Association (ALPA)
petitioned the NTSB to reevaluate a previous air carrier accident. The circumstances
surrounding the accident of Pan American 806 were similar to those of Eastern 66, and
ALPA saw an opportunity to exonerate the flight crew who were held accountable in the
Strategically situated, this tiny tropical island became a refueling depot for the early
jetliners making the run between Hawaii and New Zealand. Pan American flight 806, a
long range 707-32IB, was one such aircraft scheduled for the quick stop on the night of
Cleared for the ILS approach runway 05, the aircraft captured the localizer some
20 miles out. After being advised of a 'bad' rain shower over the airport with winds 030
at 20 gusting 25 knots, Pan Am 806 was given landing clearance (NTSB, 1974b, p.2).
Clipper 806 was unable to establish a stabilized approach, first sinking well below
the glide slope, then climbing slightly above, and when briefly on glide slope soon
ballooning well above. The stabilizer trim was run nose pitch down at this time and the
aircraft descended well below the glide slope, leveling off briefly at 300 ft AGL. The
aircraft then lost about 8 knots of airspeed and flew into the jungle environment at 140
The aircraft was determined to be in good operating condition prior to impact, and
the investigative team concentrated on human factors issues. The Safety Board reasoned
illusions in flight and procedural errors were accomplices in this accident. The initial
probable cause, as issued by the NTSB, was "the failure of the pilot to correct an
excessive rate of descent after the aircraft had passed decision height" (1974b, p. 19).
There was no mention of weather as a causal factor in the original accident report,
it was implied through the statement "visual illusions produced by the environment [rain]
may have caused the crew to perceive incorrectly their altitude..." (NTSB, 1974b, p. 19).
22
ALPA, noticing the references in the accident report to a 'bad' rain shower and the
degradation in airspeed and altitude that Pan Am 806 experienced, conjectured that the
new theories of downdraft and outburst center might have played a role in the demise of
the B-707. Just weeks after the publication of Fujita's findings in the Eastern 66 accident,
and two years after the initial accident report on Pan Am 806, ALPA petitioned the
During the second investigation, the FDR was reexamined in conjunction with the
cockpit voice recorder (CVR) and engineering performance data. As was the case with
Eastern 66, any discrepancy between the theoretical performance capability and the
actual performance of the aircraft, as derived from the FDR and CVR, was attributed to
external forces. The second investigation found very little adverse winds encountered
until about 51 seconds prior to impact. The wind, increasing in velocity, was some
combination of head wind and updraft, and this became a decreasing headwind (or
combination downdraft) just seconds later. Another increase in headwind and updraft was
then encountered, followed by a lull in the wind; in the final 4 seconds of flight the
These last winds were severe enough that the aircraft would not have been able to
sustain level flight under the application of full power, about 57,000 pounds thrust
(NTSB, 1977, p. 12). It was during this time that the Boeing 707 experienced a 1500 fpm
rate of descent only 178 feet above the trees. The Safety Board asserted that the "accident
could have been avoided had the crew recognized, from all available sources, the onset of
the high descent rate and taken timely action" (p. 22).
23
The new probable cause, as determined from the majority of the board members,
changed little from the original. Paraphrasing the 1974 report, the inclusion of why the
aircraft experienced an excessive descent rate was the only change. "The probable cause
of the accident was the flightcrew's [sic] late recognition and failure to correct in a timely
penetration through destabilizing wind changes [italics added]" (NTSB, 1977, p. 27).
wind'. As before, the obloquy was placed on the flight crew as their late recognition and
Kay Bailey, the acting chairman of the Safety Board, disagreed with the
conclusions drawn by the majority members. Convinced that wind shear was a major
factor in the explanation of the accident, his letter of dissent proposes, "the probable
cause of the accident was the aircraft's penetration through destabilizing wind changes
and the flightcrew's [sic] late recognition and failure to correct in a timely manner the
resulting excessive descent rate" (NTSB, 1977, p. 29). While not exonerating the flight
crew, the Chairman does acknowledge the reduction in performance that wind shear has
on aircraft performance.
The NTSB did not conduct simulator studies of this accident. In a rather self-
serving statement, they acknowledge the problem is dynamic and "would probably
difficulties and the ability of the crew to recognize in a timely manner the onset of an
changed direction and velocity in both the horizontal and vertical. From the description of
the winds, it is probable that Pan Am 806 entered an area of outburst winds and
continued into a downburst. The winds were characterized as (see Figure 2): (1) a
headwind with downdraft, followed with another (3) headwind and updraft, finally
ending with decreasing (4) headwinds and a downdraft of up to 1,700 fpm (NTSB, 1977,
p. 12).
500 ft t
Approx
Series of Horizontal Scale
Microburst Vortices
500 ft
Microburst
Figure 2. Probable winds encountered by Pan Am 806. The glide slope at the time of the
PAA 806 accident was propagated at 3.25°, with an average airspeed of 150 knots this
corresponds to a descent rate of 861 fpm. The downburst was descending almost twice as
fast at 1700 fpm. In reference to the glide slope at 150 knots the downburst has a relative
velocity of 839 fpm. If the flight time from point 1 to point 4 is one minute, the glide
slope will have descended the aircraft 861 feet while the downburst will have descended
an additional 839 feet, hence the upward glide slope incline with respect to the downburst
in the illustration. Note. Microburst winds from Pilot Windshear Guide (p. 10), by FAA,
1988, Washington, DC: FAA.
25
The winds experienced by Pan Am 806 were within the parameters comprising
Fujita's definitions, though the NTSB did not use the lexicon, downburst or outburst.
Reevaluating the accident, and changing the probable cause to include destabilizing
winds, the NTSB raised the prospect that previous aircraft accidents may have been
the reinvestigation of the Pago Pago accident. The FAA was beginning wind shear
Eastern 66 accident. These included ground and airborne based sensing, and wind shear
penetration capability of an airplane (NTSB, 1976, p. 40). This research was still in its
infancy and the potential for a wind shear related accident had not diminished.
Allegheny 121
Airlines Flight 121 crashed while attempting a go-around maneuver. Witnesses to the
accident corroborated the FDR data, both indicating that the aircraft was in a climb
Flight 121 departed Windsor Locks, Connecticut on June 23, 1976, for the short
trip to Philadelphia. After a routine cruise, the crew prepared for an ILS approach to
runway 27R. When the DC-9 was still about 15 miles out, the airport visibility decreased
from 6 to 2 miles, the captain commented that it was probably due to the small rain
shower a few miles west of the field. Assuming they could land before the cell reached
the airport, the flight crew continued the approach (NTSB, 1978, p. 2).
The winds at the airport were initially reported on the automatic terminal
information service (ATIS) as 260° at 10 knots. When given landing clearance, Flight
121 was issued winds 230° at 25 knots. Three seconds later, tower advised a different
aircraft that the winds were 210° at 35 knots. The captain of Flight 121 heard this
transmission, and commented to the first officer, "thirty-five, let's go around" (cited in
Activating the go around button on the throttle quadrant, the captain followed the
flight director command bars up to a 15° pitch attitude while the JT-8D engines were
spooling to the thrust setting requested. Flaps were moved from 50 to 15, and the landing
gear was retracted. As the airspeed dropped 5 knots below reference landing speed (VREF)
the flight director command bars lowered to a pitch setting of about 10° in response. The
ground proximity warning system (GPWS) triggered a pull up alert as the aircraft
continued descending toward the ground. Allegheny 121, unable to climb through the
wind shear, struck the right side of runway 27R four thousand feet beyond the threshold.
An airline captain waiting for takeoff witnessed the event, as did the Philadelphia
tower controllers. The observant captain noticed that the DC-9 hit in a nose up attitude of
about 10° just 38 feet from his aircraft: "Flight 121 appeared to stop flying, descended to
the ground with the nose up, struck the ground to the right of runway 27R, and then slid
along the ground..." (NTSB, 1978, p. 4). The air traffic controllers also confirmed a nose
There was no post accident fire, the tail section, including the engines, separated
from the fuselage shortly after impact, taking away a significant heat source from the
27
main wreckage area. Though the aircraft was destroyed, as were three taxiway signs,
During Allegheny's approach, the small cell, on which the captain previously
commented, had grown into a level 4 intensity thunderstorm with a top of 37,000 feet.
The ensuing rains decreased the visibility on runway 27R below approach minimums, the
runway visual range (RVR) varying between 1000 and 4000 feet. The winds also were
The maximum wind speed recorded was 41 knots at 1708. At 1712, the wind
speed was 36 knots. The direction of the wind was from the west from 1701 to
1705, from the southwest from 1706 to 1712, from the north from 1716 to 1717,
from the northeast from 1718 to 1721, and from the east from 1722 to 1733. (p. 6)
The meteorological conditions were highly dynamic and produced an equally dynamic
The final flight path of Allegheny 121 was a roller coaster ride of varying altitude
and airspeed. The aircraft descended from 551 ft to 88 ft, climbed to 371 ft and then
descended to 136 ft, which it held for several seconds before settling. The airspeed was
similarly chaotic, increasing from 157 to 162 knots then decreasing to 117 knots and
increasing again to 153 knots. The FDR ends the airspeed trace with the aircraft breaking
The NTSB proposed various wind models to explain the reduced performance
experienced by the DC-9. These models were developed using the established technique
to program the derived winds into their Flight Development Motion Base simulator. This
simulator, replicating a DC-9, was programmed with the accident aircraft's equivalent
weight and performance. Seven pilots flew the simulator in all but one of the wind
models, Model 4b. Most test runs were able to avoid contact with the terrain when
following flight director commands. Model 5a was not traversed in 5 out of 9 attempts at
the accident EPR setting of 1.83, but when thrust was increased to the maximum setting
of 1.93 EPR the runs were successful (NTSB, 1978, pp. 15-17).
The NTSB chose not to evaluate Model 4b, even though it would account for the
that such high downdrafts so near the ground-which would be required to produce this
Conservative in their statements and research, the Safety Board did not mention
the possibilities of either a downburst or an outburst center in their final report, though
the meteorological conditions, combined with the performance of the aircraft, and
witness statements, suggest that such a phenomenon did influence Allegheny 121 (see
The majority of the Board found "the probable cause of this accident was the
aircraft's encounter with severe horizontal and vertical wind shears near the ground as a
result of the captain's continued approach into a clearly marginal severe weather
condition" (NTSB, 1978, p. 29). Phillip Hogue, a member of the Safety Board, dissented,
stating, "the probable cause of the accident was severe wind shear encountered as the
aircraft performance was not realized. Successful simulator runs were only achieved with
strict adherence to pitch attitudes derived through a speed command system, that
temporarily sacrificed indicated airspeed below the takeoff safety speed (NTSB, 1978, p.
25). Flight below the takeoff safety speed (V2) is not a normal airline procedure;
accordingly the Safety Board recommended the FAA "establish a joint Government-
industry committee to develop flight techniques for coping with inadvertent encounters
AC 00-50 Low Level Wind Shear (NTSB, 1986, p. 157). In the event of a downburst
encounter, the change instructed the pilot to immediately increase thrust to maximum and
trade any airspeed above V2 for altitude. If the aircraft continued at an unacceptable
descent rate, the pilot was advised to gradually increase the pitch attitude and temporarily
The updated circular, AC-00-50A, was disseminated to all airlines in the United
States through their respective FAA principle operations inspector, whose task was to
ensure that the new information was reflected in each air carrier's operations, procedures,
and training programs (NTSB, 1986, p. 157). A wind shear escape maneuver
demonstration, performed yearly in the training simulator, was mandated for the pilot-in-
command. As a demonstration exercise, in which the escape was assured, this training
may have instilled a false confidence that all wind shear encounters could be negotiated
(p. 53). The training discounted the avoidance principle, and suggested adherence to
United States, the crew of Pan Am 759 prepared for the ensuing takeoff with little
hesitation. The weather radar was illuminating areas of precipitation along the departure
path, and as a precaution to expected wind shear, a maximum performance takeoff was
planned. The air-conditioning packs were turned off to allow for greater engine thrust, the
flaps were set at their minimum takeoff setting of 15, and the advice from the captain to
the first officer was "let your airspeed build up on takeoff (NTSB, 1983, p. 103). These
measures, outlined in the FAA publication AC 00-50A, were applied to ameliorate the
The Boeing 727-235 was accelerating in a rain shower down runway 10 at New
Orleans International Airport on the afternoon of July 9, 1982. Once airborne, the wind
quickly changed from a headwind, to a left crosswind, and then into an increasing
tailwind. The aircraft rose to an altitude of about 100 feet and then slowly settled, striking
tree tops % mile from the runway end before plowing into a residential area at maximum
-thrust. Along with the aircraft, six houses were destroyed, and five were damaged, 145
persons on board and 8 persons on the ground lost their lives (NTSB, 1983).
The flight was airborne less than two minutes and covered a little over a mile. In
that time the crew over-boosted the engines in a desperate attempt to fly out of the wind
shear. As the altitude was diminishing the airspeed was increasing. The final command
from the captain was "come on back you're sinking Don-come on back" (NTSB, 1983,
p. 112). This action would have traded airspeed for altitude; that is traded kinetic energy
(NTSB, 1983, p. 20), though ideally this excess could have been converted into a higher
altitude to avoid terrain contact. The escape procedure of the time recommended "if
severe wind shear is encountered on takeoff, the pilot should immediately confirm that
maximum rated thrust is applied and trade the airspeed above V2 (if any) for an increased
because of insufficient time, or procedures, the flight crew was not able to bring their
afternoon. Even though wind shear was anticipated, the NTSB found "the captain's
decision to take off [sic] was reasonable in light of the information that was available to
him" (NTSB 1983, p. 72). Confirming a go decision, the low-level wind shear alert
system (LLWAS) was not issuing any warnings. However, immediately after the crash,
the system warned of a wind shear in the same quadrant as the remains of Pan Am 759
(p. 54).
periphery of an airport. The desired spacing is 3km, however local terrain, zoning laws,
or other constraints may dictate different spacing requirements. A wind speed difference
of 15 knots between a periphery sensor and the center field sensor will trigger a wind
beyond the sensors. As microbursts are relatively small in geographic scale they can
occur between sensors, and are only registered when the outflow winds have impinged
upon an anemometer. This is an historical alert, as the microburst is well developed and
may be several minutes old by the time it is sensed. New systems, combined with
expansion), and provide a more reliable warning of microbursts (FAA, 2003a, f 7-1-26
2b). This combination of systems is now available at many airports throughout the United
States.
The infrastructure in place at New Orleans International Airport was not adequate
to warn the Pan Am crew of a microburst. Only after the accident did a warning of low-
Two meteorological models of the wind field that influenced flight 759 were
conducted by Caracena and Maddox (NTSB, 1983, p. 28). Though differing in some
aspects, both investigations revealed the likelihood of a microburst encounter (p. 30).
According to Fujita (1983a), a microburst began just as 759 initiated the takeoff
roll and lasted until one minute after the crash. The microburst was centered 700 feet
north of the runway and 2,100 feet east of the midfield sensor. The aircraft encountered a
17 knot headwind, followed by a 31 knot tailwind with a 4.1 knot downdraft. The first
obstacle, a grouping of trees, was hit at 50.7 feet above the ground with a rate of climb of
384 fpm.
33
The NOAA report proposed that 759 flew through a weak to moderate microburst
with a wind shear of 39 knots and a down flow of 7 fps (4.1 knots) at 100 ft AGL
(Caracena, Maddox, Purdom, Weaver, & Greene, 1983). The center of the NOAA
In analysis of the takeoff performance for the B-727, the Boeing Company
diminished to about 10 knots at the point of initial impact. The vertical winds "showed a
steadily increasing downdraft from the 35 feet AGL point to about 5 seconds before
impact. At this point, the downdraft remained at about 25 fps until tree contact" (NTSB,
1983, p. 57).
The results of the Boeing static engineering analysis suggested that had the pilots
held their indicated airspeed, by pitch management, the aircraft could theoretically have
maintained a 95-foot altitude, eventually flying out of the microburst (NTSB, 1983).
encounter. The Safety Board concedes the difficulty a pilot would face in recognizing this
emergency.
The probable cause of the accident was the airplane's encounter during the liftoff
and initial climb phase of flight with a microburst-induced wind shear which
imposed a downdraft and a decreasing headwind, the effects of which the pilot
would have had difficulty recognizing and reacting to in time for the airplane's
descent to be arrested before its impact with trees. (NTSB, 1983, p. 72)
34
Inadequacy in the existing infrastructure was also acknowledged. "Contributing to the
accident was the limited capability of current ground based low level wind shear
detection technology to provide definitive guidance for controllers and pilots for use in
The Safety Board felt that though avoidance was the most positive form of
(NTSB, 1983, p. 61). The Board was critical of previous flight simulator wind shear
training, stating it "may tend to instill an unwarranted sense of security to the flightcrews
[sic] rather than stressing wind shear avoidance" (p. 67). In demonstration of this
casualness with wind shear was the captain's comment "let your airspeed build up on
takeoff (p. 103), insinuating a technique of penetration and keeping to schedule could
The recommendation from the NTSB did little to change the microburst training
administered by the airlines. Guidance from the FAA was in the form of Advisory
Circular 00-50A, Low Level Wind Shear. This document, last updated after the Allegheny
121 accident, did not include the latest microburst findings. According to the FAA, "wind
shear is not something to be avoided at all costs, but rather to be assessed and avoided if
severe" (1979, f 7a), severity being a qualitative evaluation based on the judgment of the
The guidelines to identify and escape from a microburst encounter were equivocal
and this would be causal in the next air disaster. Indeed, many of the problems that
The temperatures in central Texas exceeded 100 °F, on the afternoon of August 2,
1985. The high temperature was providing the energy to build several air mass
thunderstorms. Delta 191, a flight from Fort Lauderdale (FLL), deviated around one such
thunderstorm with a top of 50,000 feet during the arrival to Dallas Fort Worth
International Airport (DFW). A much smaller storm, a growing cumulus with a top
reaching 23,000 feet, lay between their aircraft, a Lockheed L-1011-385-1, and the
This smaller cell was maturing quickly, the first officer remarked, "lightening
coming out of that one.... Right ahead of us" (NTSB, 1986, p. 131). There was no
discussion or attempt, at this point, to abandon the approach, and Delta 191 proceeded
Descending below 1,000 ft AGL, the captain advised the first officer, who was the
pilot flying (PF), "watch your speed.... You're gonna [sic] lose it all of a sudden, there it
is" (NTSB, 1986, p. 133). The aircraft had entered a microburst and the airspeed, which
had been slowly increasing, dropped from 173 knots indicated airspeed (kias) to 120 kias
in 20 seconds. The captain commanded a go-around 10 seconds later, but the aircraft
never achieved a positive climb gradient. Ground contact occurred at 169 kias, with the
aircraft bouncing through a field, over an interstate highway, and onto airport property,
The aft fuselage separated from the aircraft and escaped the post crash fire. Most
of the 29 survivors came from this section. Of the 163 persons aboard, 134 passengers
and crew were killed (NTSB, 1986, p. 6). Additionally, the driver of a pickup truck,
The LLWAS at DFW was operational at the time of the accident; tower
controllers noticed a system alert about 10 to 12 minutes after the accident when "all
sensors were in alarm" (NTSB, 1986, p. 24). As with Pan Am 759, the LLWAS activated
An earlier flight, American 351, entered the area of wind shear and lost 22 kias
several minutes prior to Delta's encounter. This occurrence was not relayed to air traffic
control as required per FAR 121.561, the B-727 captain testifying, "a windshear [sic] of
20 knots at 2,500 feet at [the] airspeed I was at is negligible and certainly would not
interfere with the safety of anyone's flight" (cited in NTSB, 1986, p. 19).
It is not known whether the presence of either ground advisories or pilot reports
would have persuaded the crew of Delta 191 to delay the approach. It appears there was a
conviction to continue even when the wind shear was acknowledged. This reluctance to
hold until weather conditions were more favorable was also evident with the crew of Pan
Am 759.
Attempting to explain this behavior, the NTSB speculated, as early as the Pan Am
759 accident, that wind shear training might instill a false sense of security through
repeated successful encounters (NTSB, 1983, pp. 67-68). The Delta Air Lines Systems
Manager also held this view; "simulator windshear [sic] training might possibly be a
subtle form of 'negative training' because it could lead pilots to conclude that adherence
based on the procedures contained in FAA AC 00-50A Low Level Wind Shear (NTSB,
1986, p. 53). If wind shear was anticipated, the pilot was expected to fly a stabilized
airspeed loss below VREF was attained with thrust application (FAA, 1979). A go-around
was recommended, "if the airplane is below 500 feet AGL and the approach becomes
unstable" (f 7b-1). This guidance had the potential to put the aircraft and crew in a
dangerous situation.
Many factors combined to cause this accident. The lack of an LLWAS warning
and the lack of pilot weather reports (PIREPs) deprived the flight crew of current
information. There were salient clues, however, that foretold of possible microburst
development, and it appears that the captain may have been attuned to this as witnessed
by his forecast of the dramatic loss of airspeed. The training to continue the approach,
rather than hold until conditions improved, ensured that eventually a microburst would be
penetrated.
indicated that the aircraft probably entered a microburst (NTSB, 1986, p. 35). Though the
crew of 191 applied the maximum thrust setting possible in an attempt to escape, they
were not as aggressive in employing a positive pitch attitude. Unable to extract the
performance necessary, the aircraft crashed. In theory, "the airplane physically had the
performance capability to fly a path that missed the ground" (p. 37). The discrepancy
between reality and theory in escaping from this, and from other accident microbursts,
mitigation steps (McCarthy, 1996). This symposium was eventually funded by the FAA
in 1986 and brought about the Wind Shear Training Aid (WSTA) curriculum. The
training aid was preemptive to many of the concerns addressed by the NTSB in their
The National Transportation Safety Board determines that the probable causes of
the accident were the flightcrew's [sic] decision to initiate and continue the
lightning; the lack of specific guidelines, procedures, and training for avoiding
and escaping from low-altitude windshear; [sic] and the lack of definitive, real-
time windshear [sic] hazard information. This resulted in the aircraft's encounter
80)
The proposed plan was inclusive enough that the only additional operational
recommendations by the NTSB, to the FAA, were for principle operations inspectors to
The Integrated Wind Shear Program Plan, drafted in 1986, of which WSTA is a
part, was the first comprehensive attempt to mitigate the problems of the microburst. In
addition to improving surface and airborne wind shear detection, the plan included
training for airline management as well as pilots. The training aid for pilots included
of the actions proposed in the Safety Recommendations issued by the Safety Board since
1988, front piece). This advisory circular supercedes AC 00-50A Low Level Wind Shear,
which had not been updated since the Allegheny accident. Microburst information and
By 1988, the FAA Administrator was urging the use of the non-regulatory WSTA
for complying with the requirements of Part 121 of the Federal Aviation Regulations. In
1991 the International Civil Aviation Organization (ICAO) incorporated wind shear
training, as outlined in the WSTA, into Annex 5 and 6. Thus, by the early 1990's, most
operators of jet transport aircraft around the world were using the FAA training aid and
During the 1970s, and into the mid-1980s, wind shear accidents were occurring
about every 18 months. After Delta 191, the next microburst accident would be almost 10
years later. The increase in safety was due to a variety of factors, the WSTA was
implemented in large scale and there existed standard procedures, endorsed by the major
infrastructure had expanded to include the ASR-9 radar, which was able to discriminate
precipitation intensities and display these areas to air traffic controllers on their
with the FAA mandating airborne wind shear warning systems per FAR 121.358.
The failure of the infrastructure to provide timely warnings, as in the case of Pan
Am 759 and Delta 191, would again surface. While infrastructure chiefly aids in
US Air 1016
The late afternoon flight on July 2, 1994, from Columbia, South Carolina (CAE)
to Charlotte, North Carolina (CLT), 80 miles away, required just minutes to complete for
US Air 1016, a Douglas DC-9-30 aircraft. Though thunderstorms were not reported on
the arrival ATIS, scattered thunderstorms were present in the area. The DC-9's radar was
depicting two separate cells in the terminal area. The cell on the south end of the airport
was contouring and moving northward, bringing heavy precipitation; the tower
Established on a visual approach to 18R, US Air 1016 noticed the rain was now
between their aircraft and the runway, the captain commented to the first officer, "chance
of shear" (NTSB, 1995, p. 158). This was soon confirmed by the LLWAS alert in the
northeast boundary, as reported on frequency by the local controller. Soon after, 1016
entered the rain. The first officer registered an increase in airspeed, the captain observed
another increase, and a go-around was commanded. The aircraft was at 200 ft AGL and
147 kias when thrust was increased to 1.82 EPR and a normal go-around was initiated.
The flaps were raised from 40 to 15 and the pitch attitude was increased to 15°. The
captain, who was the pilot not flying (PNF), instructed the first officer to decrease pitch,
41
"down, push it down" (p. 164). The aircraft climbed for a few more seconds, to about 350
feet AGL, and then began a steady descent into the ground (NTSB, 1995).
The aircraft broke into four main pieces upon contacting the terrain, less than a
half-mile from the airport. Of 57 persons on board, 37 died, 16 received serious injuries,
departing CAE, the crew received a copy of the CLT weather, forecasting a thunderstorm
(NTSB, 1995, p. 18). When the flight crew arrived in the CLT area they were able to
visually identify the storm that appeared as a contouring cell on their radar. Further
analysis indicates the cell was of severe enough intensity to form a radar shadow,
attenuating the left side of the storm (Smith, Pryor, & Prater, 2000, p. 57). The same area
of the storm, on the northeast boundary of the airport, triggered the LLWAS alert. The
wind shear alert was responded too on the flight deck with a non-pertinent word (NTSB,
1995, p. 162).
weather; US Air 806 was 'sitting tight' (NTSB, 1995, p. 162) and company 797
transmitted, "[departure] wouldn't sound like a good plan" (p. 162). Aircraft were
landing though. The preceding flight, a Fokker model FK-100, reported a smooth ride on
approach (p. 161). The NTSB acknowledged in the Eastern 66 accident, "pilots
commonly rely on the degree of successes achieved by pilots of preceding flights when
they are confronted with common hazards" (NTSB, 1976, p. 34). With clues for and
less than 30,000 feet (NTSB, 1995, p. 49). Though small in stature, the storm had a radar
reflectivity of 65 Decibels and had generated at least three cloud to ground lightning
strikes (Smith, Pryor, & Prater, 2000). The microburst, centered 1.85 km east of the
accident site, was 3.5 km in diameter. According to NASA, the maximum wind velocity
change was 86 knots along the north-south axis, with a vertical wind velocity of 23 fps
(14 knots) along the flight path (NTSB, 1995, p. 48). Douglas Aircraft estimated the
vertical winds along the flight path to be initially 10 fps, increasing to 25-30 fps. The
DC-9 experienced a 61-knot horizontal wind speed change: a 35-knot headwind shearing
to a 26-knot tailwind in 14 seconds (NTSB, 1995, pp. 46-48). Both analyses indicate a
performed a mathematical simulation (flight simulation was not performed) using data
from the FDR, the NASA derived wind field, and the theoretical aircraft performance.
The simulation was able to avoid ground contact with gear retraction, firewall power, and
a sustained 15° pitch attitude. Under these constraints the minimum altitude of the aircraft
was 335 ft AGL (NTSB, 1995, p. 50). The DC-9 could have successfully flown through
the wind shear encounter if the simulated missed approach procedure had been used, or if
the wind shear escape maneuver of maximum effective pitch attitude and firewall thrust
action. The reality presented a different situation. During the go-around the EPR had not
been set to the maximum 1.93, but was about 9% less (NTSB, 1995, p. 97). This was
43
cautioned in the Eastern 66 simulation "most of the pilots actually added less thrust than
they thought they had added" (NTSB, 1976, p. 19). Also, the pitch attitude was not held
at 15° (1995, p. 97). Again, in the Eastern 66 simulation it was noted that the pilots did
not rotate to a high enough angle needed to stop the rate of descent (1976, p. 19). The
Though the LLWAS performed to its design tolerance and issued a wind shear
alert, the NTSB concluded that TDWR derived information would have been beneficial
to the crew (1995, p. 118). The NTSB also felt that inadequate procedures in the CLT
Tower prevented the flight crew from receiving critical weather information (p. 118). The
The airborne wind shear alert system installed in the DC-9 was designed to
activate a warning to the flight crew when it detected severe wind shear (NTSB, 1995, p.
annunciator only. About ten seconds after 1016 initiated the go-around, the system should
have gone into warning alert, it did not, and investigators were unable to determine why
NTSB found that air traffic control procedures and the airplane's wind shear warning
The accident was a result of the flight crew's decision to continue the approach
into convective activity, their failure to recognize the microburst in a timely manner, and
the inability to establish an escape configuration. The lack of real-time wind shear hazard
information from air traffic control was also cited as an element in the probable cause
The flight crew's inability to identify the microburst in a timely manner prompted
the recommendation to "reevaluate the Windshear Training Aid based on the facts,
conditions, and circumstances of this accident" (NTSB, 1995, p. 123). Additional training
was suggested for identifying convective activity and microbursts (p. 124). Simulator
was also proposed that the escape procedure be used in place of a go-around maneuver
below 1000 feet AGL when conditions conducive to wind shear were present (p. 123).
Though the clues were apparent, and acknowledged, that wind shear was a real
possibility, the crew continued. The failing of the infrastructure led them into a
microburst. Thereafter the escape, theoretically possible, was incumbent upon the flight
Additional Accidents
Microbursts are neither a new nor a rare event; nor are they confined to any single
geographic locale. With the late acknowledgement of the microburst in 1983 by the
NTSB, came the question of whether this overlooked phenomenon might have been
(1996) from the sources of Fujita and others, suggests that many previous accidents
display the characteristics of a microburst encounter (see Table 1). Microburst accidents
are not limited to jet aircraft. A notable case study by Poellot, Borho and Bassingthwaite
(1997) presents the en route accident of a Piper Navajo in North Dakota. However, this
table and study concerns large jet transport category aircraft and not general aviation.
45
Table 1
American Jet Transport Accidents and Incidents as a Result of Wind Shear-Microburst Encounter
of the accident reports, the parent cloud can be relatively small. Microbursts can occur
wherever convective activity exists, as such there are few regions immune to this type of
meteorological activity.
Around the world, there have been aircraft accidents attributed to microbursts. In
May 1976, Royal Jordanian Air Lines encountered a microburst while attempting to land
at Doha, Qatar. Ironically, the aircraft slid tail first into the fire station garage (Fujita,
1985). Unfortunately, 45 people perished. In Faro, Portugal a chartered Martin Air DC-10
Accident Prevention, 1996). The aircraft was destroyed and 56 people died. Microburst
related accidents have been documented in Africa, Australia, India, Japan, Mexico, and
The substantial loss of life from the microburst provided the stimulus for
investigation. While some accidents added to the death toll in large numbers and some
people escaped fatalities, all accidents were financially costly. The burden on the aviation
industry accelerated the search for a pragmatic solution. The result was a highly
sophisticated infrastructure.
cause. Before the accidents clues were available, which in hindsight seem obvious. There
was a decision to continue the approach, sometimes when dangerous conditions became
increasingly evident. In the attempt to escape, aircraft of different make and model
Eastern 66 had reports from company 902 about wind shear, they had previously deviated
around thunderstorms, and were aware of convective activity in the area and on the
approach. PA A 806 was advised of a 'bad' rain shower on the airport. ALG 121 was
following an Eastern flight that executed a missed approach because of high winds. Pan
Am 759 anticipated the wind shear as evident in the discussion and actions of the crew.
Delta 191 also anticipated the wind shear, when the captain forecast losing speed kall of a
sudden'. US Air 1016 received an LLWAS alert, was painting the storm on the radar, and
anticipated the shear. In all the cases there was an indication of potentially severe wind
shear.
The clues were unheeded until too late. Eastern 66, unable to maintain glide slope,
initiated the go-around seconds before impact. Allegheny received an update of 35 knot
winds before deciding to go around, Delta was also below the glide slope when go-
around became evident, and US Air 1016 had two separate airspeed spikes before they
attempted a go around.
In the go-around, and even in the takeoff regime, the aircraft exhibited similar
performance. Power was applied too little or too late, pitch was sacrificed at some point
in the maneuver, and all aircraft contacted the terrain at a speed well above stall. Eastern
66 impacted 33 knots above stall speed, Pan Am 806 at 30 knots, Allegheny 121 at 45
knots, Pan Am 759 at 27 knots, Delta 191 was 64 knots above stall speed, and US Air
1016 was 32 knots above the stall speed when it crashed. Clearly, the full performance
capabilities of the aircraft were not being utilized. This performance decrement continued
with the adverse winds of Eastern 66, and continuing through the destabilizing winds of
Pan Am 806, the horizontal and vertical wind shears of Allegheny 121, and the
microbursts of Pan Am 759, Delta 191, and USAir 1016. Though previous accidents were
most certainly a result of inadvertent encounters with microbursts, and dismissed under
aviation meteorology. This accident provided the catalyst and clues for Fujita to form his
theories, and though progressive and unorthodox, they were not conceived in a vacuum.
Microburst History
First through myth, and then science, man has endeavored to understand the chaos
of nature and create order in his world. Both methods of understanding describe visible
events precipitated by invisible forces; a thunderbolt can be viewed as the wrath of Zeus
Classical
The result of lectures at the Lycaeum around 336 b.c.e., Meteorology describes a
coherent world of orderly winds blowing from one of 12 directions. Modern readers can
still identify the winds of Aristotle; "Zephyrus is the wind that blows from ... where the
sun sets at the equinox" (Aristotle, Berlin Trans, n.d., 363b). This discourse in
Meteorology continues with the nature and properties of the wind, informing the reader
that contrary winds may not blow while apposing winds can and do. "Winds that are not
diametrically opposite to one another may blow simultaneously," (Aristotle, 364a) this
can be quite advantageous as "... different winds and blowing from different quarters, are
favourable [sic] to sailors making for the same point" (Aristotle, 364a).
Renaissance
Not disparaging the authority of Aristotle, Bohun (1671) describes winds that are
certainly neither advantageous nor favorable. As documented from his own astute
observations and those of contemporary sailors, Bohun notes the existence of variable
and most dangerous winds. Describing these tempestuous winds, Bohun uses the
colloquial term tornado, which had, since 1625, been in common usage amongst
navigators to describe violent thunderstorms of the tropical Atlantic, with torrential rain
and sudden and violent gusts of wind (Oxford English Dictionary, 1971).
So variable and unsteady are the tornado-winds, so little obliged to any certain
law, that they commonly shift all the points of the compass in the space of an
houre (sic), blowing in such suddain (sic) and impetuous gusts, that a ship which
was ready to overset on one side, is no lesse (sic) dangerously assaulted on the
other; sometimes they shift without intermission ... Let a fleet of ships saile (sic)
as near as they can without falling fowl on each other, and they shall have severall
Bohun (1671) additionally notes that winds need not be parallel to the ground,
"sometimes you shall have a suddain (sic) puffe (sic) of wind, driven from between two
clouds, with a violent displosion (sic) of the air; that descends almost perpendicularly to
the Earth" (p. 18). This wind, he noted, occurred oftentimes with the onset of rain, "I
have oftentimes obsev'd (sic), that stiffe (sic) gusts of wind happen immediately before
50
rain" (p. 20). With one of three drawings in a book of 302 pages, it is curious to note that
Bohun chooses to illustrate a wind remarkably similar to the present understanding of the
microburst (see Figure 3). Even if Bohun placed no name on the microburst, he described
its effects. "The Portugues [sic] in their discoveries of the Orientall [sic] Indies, lost 9
ships out of 12, which were overset by the prodigious inpetuosity [sic] of these suddain
Explaining the formation of microbursts rather clearly, Bohun (1671) states that a
tornado-cloud may create a tempest "by its pressure; when the cloud distills not by
degrees in pluvious drops, but rushes down impetuously all at once, driving before it a
swift torrent of air, which falls as from a precipice, and threatens the oversetting of ships"
(p. 249). This is so very similar to the current understanding. "Dry air evaporates rain
falling from above, and the cooling caused by evaporation creates a large bomb of cold
air that barrels down toward the Earth along with the remaining precipitation. ... the
Bohun also states about these winds, "the lesser the cloud appears at first, the
tempest will last the longer" (1671, p. 250). Fujita writes "the parent clouds which induce
microbursts are not always thunderstorms. Quite often, isolated rain showers spawn
relatively strong microbursts" (1985, p. 70). Arguably, Bohun documented the effects
and some of the properties of downdrafts and microbursts. While Bohun was able to
describe this specific tempest in general terms, he failed to label it discernibly and the
years pushed his writings into obscurity. It was then left for Fujita to rediscover and name
the downburst and microburst. How Fujita accomplished this would be remarkably
familiar to Bohun.
51
Figure 3 Wind resembling microburst from Bohun's book of 1671 Note From A
discourse concerning the origine and properties of wind With an historicall
account of hurricanes, and other tempestuous wind. (p. 19), by R. Bohun,
1671,Oxford, England* W Hall
52
Modern
In his discourse on wind, Bohun relied on his own astute observations along with
the logs and observations of sailors to document the phenomenon of which he wrote.
Likewise, Fujita used his own observations and the various logs and observations of
others to formulate his theories. Instead of relying on sailors and written logs, Fujita had
the opportunity to utilize the electronic logs of aircraft, complex instruments, and
Fujita had the opportunity to investigate many aircraft accidents resulting from
wind shear and microbursts. The knowledge gained through the FDRs and aircraft
performance painted a physical picture of the environment in a specific place and time.
providing data for analysis and pragmatic use. Unquestionably, Fujita is the father of the
Many notable scientists refined the downburst theory. Caracena developed the
concept of the vortex ring generated by the initial downburst, which accounts for many of
the dynamics observed in the microburst (Caracena et al., 1989). McCarthy provided the
meteorological expertise for the WSTA, and performed many studies of wind shear
forecasting techniques (Atkins & Wakimoto, 1991). Indeed, many share in the
NIMROD
Research On Downbursts (NIMROD). The initial proposal was submitted to the National
Science Foundation (NSF) after the Eastern accident. The additional funding resulting
from the accident expanded the project, which became operational in spring 1978. Using
new techniques afforded by Doppler radar, the team led by Fujita was able to
skepticism from many scientists that any downward momentum of air could continue
larger downdrafts; this necessitated a clear division between the localized phenomenon
and that of wider reach. From these beginnings came the terms microburst and
macroburst. During the course of the investigation, 50 microbursts were observed and
JAWS
Following the success of NIMROD in 1978, Fujita along with McCarthy and
Wilson, from the National Center for Atmospheric Research (NCAR), proposed a large-
scale project. The moniker chosen by NCAR staff was every bit as creative as the
preceding study. Being a collaboration of the University of Chicago and NCAR, and
based around the Stapleton Airport in Colorado, the Joint Airport Weather Studies
documented (Fujita, 1985, p. 55). Microbursts were in fact quite common. The low
probability of encountering a microburst came from their small size and short life span,
Doppler radars, 2 pulsed Doppler lasers, 27 portable automated mesonet (PAM) stations,
a high density LLWAS and a pressure jump sensor array. This structure was placed
strategically around Stapleton Airport, then the fourth busiest airport in the United States
(McCarthy, 1983).
which were no less impressive. Three basic areas were to be addressed: low-level
convective storm winds, aircraft performance in wind shear conditions, and wind shear
detection and warning techniques. The focus remained "to explore quantitatively the
Had the researchers any doubt of the significance in their work, it would have
disappeared July 9, 1982, in the middle of their field observations. On this day, Pan
American Flight 759 crashed in New Orleans. The accident was quickly attributed to a
microburst, "as a result the JAWS researchers felt [equally] that the microburst
phenomenon should be understood as quickly as possible for the sake of aircraft safety"
Pan Am 759 was the first accident to be formally classified as resulting from an
encounter with a microburst. This recognition legitimized the theory and shifted its
55
acceptance from the fringe into the mainstream, with a resulting increase in credibility
and funding for research (Fujita, 1985, p. 53; NASA, 2002, f 5).
The NTSB requested that the information gleaned from the JAWS project be used
to quantify the low-level wind shear hazard and to evaluate the effectiveness of the
LLWAS. Additional use of the data was to develop training aids to emphasize the peril of
convective weather to safe flight, and to develop realistic microburst models for use in
NAS
The outcry from citizens, over the Pan Am 759 accident, prompted congress to
pass public law 97-369 mandating the FAA contract with the National Academy of
Sciences (NAS) to examine ways to mitigate the risk of wind shear (NRC, 1983, pp. ix-
x). The blue-ribbon panel recommended near, middle, and long-term solutions,
essentially all of which have been implemented to varying degrees (McCarthy, 1996,
p.2). This was not a field study but a collection and assimilation of previous research. The
National Research Council (NRC) published their findings as Low-Altitude Wind Shear
and Its Hazard to Aviation. The blue-ribbon panel advocated an integrated wind shear
program, which set a course for the FAA, and "was instrumental in coming to grips with
a national problem facing the safety of the flying public" (McCarthy, 1996, p. 6).
CLAWS
It was the safety of the flying public that prompted the FAA to request a
microburst real-time forecast and warning service at Stapleton Airport (NTSB, 1986,
pp.33-34). On May 31, 1984, a United Airlines B-727 encountered a microburst during
its takeoff roll, and, after becoming airborne, struck the localizer antenna. The pressure
vessel of the aircraft was breached, fortunately with no injuries, and the aircraft returned
The Classify, Locate, and Avoid Wind Shear (CLAWS) Project was formed in
response to the FAA request (McCarthy& Wilson, 1985). CLAWS was planned, funded,
and implemented in just 7 days, and lasted from July 2 to August 15, 1984. Doppler radar
was used to issue warnings of microbursts and probable wind shear; 35 microburst
advisories were issued for the airport, prompting 7 aircraft to abandon the approach or
delay takeoff. In addition to the Doppler radar warnings, a daily microburst probability
forecast was issued, achieving an accuracy of approximately 80%. Wind shift advisories
and convective initiation advisories were also issued from the data provided by the test
instruments.
The success of CLAWS was the ability to quickly implement the research efforts
warnings may have prevented an accident. At least one pilot displayed his gratitude; "by
just having this available-note we were in a heavily loaded 737 in the critical approach
phase-this warning in advance may have just saved and aircraft from being forced into
the ground short of the runway" (cited in McCarthy & Wilson, 1985, p. 254).
An important and unanticipated value of CLAWS was the decrease in air traffic
delays caused by severe weather. The project also produced effective microburst
Both JAWS and CLAWS, based on the eastern slope of Colorado, primarily
studied dry microbursts; those microbursts produced in a dry environment in which the
rain evaporates before reaching the surface (Caracena et al., 1989). The two most recent
57
accidents, PAA 759 and DAL 191, had occurred in moist environments. The
atmosphere, differ from the dry environment (Wakimoto & Bringi, 1988).
MIST
Laboratories Operational Weather Studies), tested algorithms for wind shear detection by
Doppler radar. Utilizing the same equipment, meteorological concerns were addressed by
the National Science Foundation (NSF) in the MIST (Microburst and Severe
Thunderstorm) offshoot (Dodge, Arnold, Wilson, Evans, & Fujita, 1986, pp. 417-419).
During June and July of 1986, MIST employed 5 Doppler radars, 41 PAM
stations, 30 mesonet stations, 5 LLWAS networks, and 2 rawinsonde sites. With this
microbursts in a wet region (Wakimoto & Bringi, 1988). The MIST project
complemented the dry microburst data obtained during JAWS. Scientists now had a full
AWDAP
Significant public and political focus on the problem of wind shear was generated
by the accident of Delta 191. The Congressional House Committee on Science and
Technology responded by funding wind shear research at NASA. On July 24, 1986, the
Airborne Windshear Detection and Avoidance Program (AWDAP) was absorbed into the
FAA's National Integrated Windshear Plan (NASA, 2002), thus creating a joint research
hazard level, to develop remote sensing of wind shear, and to design and develop a means
The first objective was met with the development of a wind shear hazard index
predicting impending flight path deterioration. The index, or F-factor, is based on the
total aircraft energy and its potential rate of change through a horizontal and vertical wind
(Proctor, Hinton, & Bowles, 2000, p. 482). Hazardous wind shear was determined,
through research, to be present when values greater than 0.1 were generated (p. 483).
To evaluate the concept of the F-factor, along with meeting the remaining two
objectives, NASA would flight test a variety of sensors and displays. Doppler radar was
proving a success in ground instillations, and with substantial modification, NASA was
able to develop an airborne Doppler radar. This would be tested on their 737 aircraft
along with a lidar system, and an infrared radiometry sensor. Additional enhancements to
the 737 included an improved in situ reactive wind shear warning system and a VHF data
Flight testing and validating the equipment in an operational setting was a major
function of the research project. The testing took place in Denver in July of 1991 and
1992 to evaluate dry type microbursts, and in Orlando in June of 1991 and August of
1992 for evaluating wet type microbursts (NASA, 2002). The fully manned and
instrumented aircraft was initially guided toward a microburst with the aid of ground
based TWDR, several miles from penetration the airborne sensors would be used and
internal guidance would commence. The aircraft was flown at a minimum of 750 feet
AGL and 210 kias for storms greater than F-factor 0.1, with 0.15 storms being avoided
The testing confirmed that Doppler radar is the most effective in depicting both
(NASA, 2002). Lidar worked well in the dry atmosphere, but was attenuated by the rain
and did not provide sufficient warning for wet microbursts. The infrared radiometry
system was a disappointment in all cases, being unable to distinguish the necessary
temperature changes.
Key successes of AWDAP were the development of the F-factor and Doppler
radar. The F-factor, displayed on the sensor screens, provided a crucial quantitative
analysis in aiding the decision making process and is now on many new jet aircraft
(Procter et al., 2000, p. 485). Doppler weather radar has also made the transition from
the equipment and theories in a real-world flying laboratory, through actual microbursts
with a jet transport aircraft, NASA provided basically unassailable data, and probably
unattainable by industry.
tragedy. The Eastern 66 accident provided the catalyst for the downburst theory, and
eventually freed funds for the NIMROD project; Pan Am 759 increased the funding for
the JAWS Project, the incident of United 663 prompted CLAWS, and the Delta 191
accident was the catalyst for the FAA to join with NASA in an airborne detection
Escape procedures are also inexorably tied to the aircraft accidents. Iberian 933,
the DC-10 which crashed in Boston, prompted the first publication by the FAA on wind
shear (Grossi, 1988). The Allegheny 121 accident added to this advisory circular escape
guidance (NTSB, 1986, p. 157). Not until several accidents later, most notably United
663 and Delta 191, were definitive procedures advanced in the form of the WSTA. The
NTSB recommended an evaluation of the WSTA after the accident of US Air 1016
(NTSB, 1995, p. 123). Like knowledge, escape procedures were refined through the
Microbursts are not a new phenomenon, but the implications to society are. In a
relatively short time span, the theory was postulated, researched, and procedures
implemented to mitigate its effects. Many notable scientists contributed to this effort in
diverse and creative ways. The data that they collected, the observations made, and the
theories presented enhanced the knowledge of microbursts, the safety of aviation, and the
science of meteorology.
Microburst Meteorology
Unlike its cousin the tornado, the microburst is not visible, and telltale signs of dust or
rain rings rising from the surface are often the only indication of its existence. Also
unlike the tornado, the microburst does not require a storm environment to develop.
Summer skies, even those appearing innocent, frequently contain the ingredients to
downward flow of wind creates a pressure gradient of strong torque which may manifest
Size
extending only 4 km (2.5 miles) or less (Fujita, 1985, p. 8). It is very difficult to classify
downbursts as to size, and to have an arbitrary 4 km size limit confines the term to
artificial boundaries (McCann, 1994, p. 532). The term microburst is now colloquially
Microburst winds are generated in the meteorological mid-layer, at about the 500
millibar (mb) level, which corresponds to about 18,000 ft MSL. The core of the
microburst is generally less then 1 mile in diameter, with the horizontal outflow 2.5 miles
in diameter, the spread beginning 1000 to 3000 ft AGL (FAA, 2003a, f 7-1-26). The
Microburst winds are a localized phenomenon in both space and time. One of the
strongest microbursts ever recorded had wind gusts greater than 130 knots, while just 2.3
miles away the winds were light and variable at 5-6 knots (Fujita, 1983b, p. 6). This
occurred August 1, 1983, at Andrews Air Force Base, just five minutes after the
presidential aircraft had landed, with President Reagan on board (Fujita, 1983b).
62
Wind Speed
While some microburst may produce hurricane strength winds, most have a wind
speed of 12-14 ml sec (27-31 mph) (Fujita, 1985, p. 63; Proctor, 1985, p. 257). Peak
winds occur about five minutes after the initial horizontal divergence at the ground,
typical horizontal differential speeds are 24 m/s (~ 54 mph) over a distance of 1800 m or
about 1 nm (Proctor, 1985, p. 257). The wind at 75 m (~ 250 ft) AGL contains the highest
The greatest horizontal wind shear and downdraft velocity exists when the core
downdraft radius is small (Proctor, 1985, p. 264). The small size and high speed
correlates into a tight wind gradient. Downdrafts of 30 m /s (~ 6,000 ft/min) with shears
of 167 km /hr (~ 90 knots) can occur in the microburst (FAA, 2003 a). The maximum
down-flow speeds are in the lower levels, below 1 km (3,280 ft) (Proctor, 1985).
73), which acts to enhance the outflow speed near the ground. The stretching ring vortex
generates much faster outflow winds than can normally be accounted for in the downdraft
(Fujita, 1983b, p. 28). In fact, maximum outflow speeds occur just as the vortex-ring
reaches the ground (Proctor, 1985, p. 258). These vortex rings produce strong shears over
a scale of several hundred meters (p. 264). This is particularly dangerous for aircraft, as
witnessed when Delta 191 flew through several stretching ring vortices (Fujita, 1986, pp.
35-44).
As the ring vortex stretches, it breaks apart due to expansion. Sections may
The streamlines from the majority of microbursts flow outward with little or no
horizontal curvature. There are exceptional cases in which downdrafts are observed to
while about 10 % of rotating microbursts are anti-cyclonic (Fujita, 1985, p. 74). This
mini-cyclone may act as a hydrometeor funnel to fuel the downdraft (p. 74).
Underneath the downburst, a dome of high pressure exists (Proctor, 1985, p. 258).
This dome is induced by stagnation pressure of the outflowing winds. The central high-
pressure dome is encircled by a low-pressure ring, which acts as an accelerant for the
diverging wind. Beyond the low-pressure ring lies an encircling ring of high pressure,
outside of which the pressure drops to the normal atmospheric level. The microburst
winds initially accelerate toward the low-pressure ring, and then slow as they approach
the high-pressure ring, after which they again accelerate toward the reduced pressure on
The relationship between pressure and temperature is intertwined in the ideal gas
law. In most microbursts the down-flow wind was observed to be cooler than the
environment, and some reactive wind shear alert systems use this heuristic as a warning
threshold (NTSB, 1995, p. 13). In NIMROD and JAWS, however, 40% of the
microbursts observed were warmer than their environment (Fujita, 1985, p. 65).
Evaporation plays a large role in keeping the air inside a microburst cool, and hence
negatively buoyant, however, the downward momentum of the air may drive warm
Duration
The life of a microburst can be calculated as the duration of Vi peak wind speed,
that is the time from when the wind is half of its greatest value until it drops below this
value on its return to the environmental norm. This varies between 1 and 8 minutes with
an average of 3 minutes (Fujita, 1985, p. 65). The build up of wind and subsequent
dissipation may add considerably to this time. Caracena et al. (1989), notes the
periodicity of vortex ring instability may increase the life of the microburst six fold
(p. 13). Generally, however, rapid growth and decay of the microburst occurs on the order
This should not lead to a false assumption that once a microburst occurs the event
is over. A series of microbursts can take place at a similar location (Caracena et al., 1989,
p. 14); they often occur in families (Cummine, 1997, p. 268; FAA, 2003a) and are
Frequency
In the central and southern United States, as many as 100 microbursts a year may
occur in a county-size area (McCann, 1994, p. 533). Predominant in spring and summer,
an estimated 3,510, with a wind speed of 75 knots or greater, take place in the United
States^* times more frequent than tornados (Fujita, 1985, p. 78). In the 42 days of the
NIMROD project, 50 microbursts were observed, while 186 microbursts were observed
in the 86 days of the JAWS project, and 62 microbursts occurred in the 61 days of the
specific region (Fujita, 1985, p. 68). During the MIST project, peak microburst activity
occurred at 15:00 local time with a lesser peak of unknown origin at 12:00 local time
(Atkins & Wakimoto, 1991, p. 472). The JAWS project obtained similar results, with wet
microbursts peaking between 14:00 and 15:00 local time, and dry microbursts peaking
between 14:00 and 16:00 local time (Fujita, 1985, p. 69). Microbursts exhibit this diurnal
variation with strong correlation to maximum surface temperatures (Atkins & Wakimoto,
1991, p. 472).
Environmental Conditions
Though not a rare event, the microburst needs a specific environment in which to
emerge. Convective type clouds provide the clues indicating the atmosphere may be ripe
for microburst development, however they are not an affirmation that an occurrence is
imminent. For a microburst to spawn, the atmosphere must produce motive forces while
Atmospheric Properties
Often, microbursts are associated with thunderstorms, but any low or middle layer
convective cloud, with the right conditions, is a suitable parent (Fujita, 1985; FAA,
2003a). Altocumuli, and clouds with little vertical development, are able to spawn
microbursts as intense and violent as large thunderstorms. The relationship between radar
reflectivity of the cloud and the strength of a microburst is not apparent, "weak showers,
whose drops evaporated before reaching the ground, sometimes produced intense
66
microbursts" (Proctor, 1985, p. 257). It was also found that some microbursts "were
Of course this does not portend that thunderstorms are not a vehicle for
microburst development. The accident histories, especially Delta 191 and US Air 1016,
contain the conditions conducive to microburst development with a deep mixed layer,
high lapse rate, and precipitation to fuel strong microbursts (Caracena et al., 1989, p. 25).
The presence of these conditions, however, does not guarantee their development.
that develop from secondary outflow boundaries (McCann, 1984, p. 537, 539), with the
faster moving boundaries more conducive for microburst generation (p. 538). This was
the environment that trapped Delta 191: the 50,000 ft thunderstorm, which was
on the approach course (Fujita, 1986, pp. 9-15; NTSB, 1986, pp. 58-59).
Motive Forces
Microbursts have their origin in the precipitation entrained in the mid- and upper-
layer of the atmosphere. In the upper-layer, the precipitation is often in frozen form;
either hail or ice crystals. As the mass of individual particles accumulates, through
various processes, it eventually exceeds the ability to be suspended at that layer. When
layers, the once frozen particles melt, creating rain. This change of state absorbs the
latent heat in the surrounding atmosphere. The cooler air, now much more dense than its
environment, becomes negatively buoyant and accelerates downward with the rain. If a
layer of dry air exists in a lower level some of the rain will evaporate, further cooling the
In the mid-layer of the atmosphere, where the temperature may already be above
freezing, a downburst can initiate when dry air from the mid to upper atmosphere is
entrained into a convective cloud. The dry air mixes with the saturated air in the cloud,
cooling off the local air relative to the surrounding air, thus creating negative buoyancy.
The entrained dry air initiates a downward velocity (Cummine, 1997, p. 269).
In either the mid level or upper level cloud, negative buoyancy in the downdraft is
primarily generated below the freezing level. The freezing level is therefore a measure of
Radar scans confirm an acceleration of the downdraft at the freezing level (Fujita,
1985, p. 16). It is apparent that the heat required for melting is an important energy
source for initiating a large downward acceleration (p. 18). Evaporative cooling is often
enough to keep the parcel colder than its surroundings and therefore negatively buoyant
(Leech, 1985, p. 308). If a parcel of downward moving air is warmer than its
environment, the speed decreases, but if it has enough kinetic energy it can reach the
surface before decelerating to zero velocity (p. 307). This may explain the variable
within the storm, the forcing mechanism is the amount of evaporation and melting of the
water and ice as it is pulled down by gravity (Wolfson, 1990). The cooling of the air
below the melting level occurs primarily from the evaporation of rain and secondarily
from the melting of hail (Proctor, 1985, p. 258), and this is the mechanism for most
liquid, and downburst strength does exist, but is less critical than environmental lapse rate
for temperature and humidity (Proctor, 1985, p. 258). The lapse rate is a measure of how
much negative buoyancy the water-saturated air can gain through melting and
evaporation. If the lapse rate is lower than 5.5 °C /km, microburst probabilities are nil
(McCann, 1984, p. 533). Thus, temperature inversions with negative lapse rates form a
barrier to downdrafts, diminishing the strength of the outflow. It is rare for a microburst
1996, p. 68).
Two distinct types of microburst exist-dry and wet. Each has a preferred region
within the United States. The Dry microburst is predominantly found within the western
states, while the wet microburst occurs in the more humid regions of the country,
especially in the southeastern states (Atkins & Wakimoto, 1991, p. 471). There is
substantial overlapping in the geographical regions, and the microbursts need not be
exclusive to any particular locale. Of the microbursts observed in the Denver area JAWS
project, 83% were dry, while only 36% of the microbursts in the Chicago area NIMROD
Regions frequented by warm and dry conditions in the lower atmosphere, with a
nearly saturated and well-mixed layer at about 500 mb, favor the development of dry
microbursts (Atkins & Wakimoto, 1991, p. 470). This environment is typical during
The high bases (500 mb) of the convective clouds in the saturated layer sit atop a
deep dry adiabatic layer with temperature dew point spreads approaching 30 °C
(Caracena et al., 1989, p. 15; Fujita, 1985, p. 71). The high cloud bases and dry lower
environment allow time for rain to evaporate (Fujita, 1985, p. 71). The evaporation of
falling precipitation droplets causes the subsiding air to become negatively buoyant
The strong surface winds associated with the dry microburst are a result of
When the lapse rate below 500 mb is approaching, or greater than, dry adiabatic
(9.76 °C/km) (Hallowell et al., 1996, p. 67), and conditions are slightly unstable to stable,
dry microburst formation is possible (Nelson & Ellrod, 1997, p. 263). The
thermodynamic forcing in the dry environment is much greater than in a wet environment
altocumulus clouds and are often accompanied by virga (Nelson & Ellrod, 1997, p. 262).
However, the anvils of thunderstorms can produce high-level virga resulting in a dry
microburst. These dry microbursts occur a distance from the parent hailstorm, and are
difficult to detect because they are not embedded within the large radar echo of the
Wet Microbursts
days when the environment is potentially unstable; hence these microbursts are
frequently associated with severe storms (Atkins & Wakimoto, 1991, p. 478).
temperature structure of a dry adiabatic sub-cloud layer from the surface to about 850
mb, topped by a more stable layer. The moisture profile from the surface to 500 mb is
nearly saturated, and is capped by a dry layer at mid-level (Nelson & Ellrod, 1997, p.
263; Caracena et al., 1989, p. 16). The precipitation core must interact with the dry layer
for microbursts to form, if this precipitation core contains ice, the microburst will be
The dry layer acts to change the state of the precipitation, absorbing the latent
heat in the atmosphere while adding to the negative buoyancy of the air. The strongest
downdrafts are associated with very dry air aloft near the melting level (Atkins &
Wakimoto, 1991).
microbursts. In more stable environments, precipitation in the form of ice facilitates the
strongest microbursts (Atkins & Wakimoto, 1991, p. 480). This can lead to large
temperature drops (p. 472) and negative atmospheric buoyancy, which accelerates the air
downward.
71
A strong downdraft conveys rain toward the surface at a much faster rate than it
can fall at terminal velocity through still air. As the downdraft approaches the ground it
decelerates in the vertical, allowing a heavy load of water to accumulate above the
ground. A wet microburst may at first appear as a darkened mass of rain descending
through light rain (Caracena et al., 1989) In the humid regions of United States, such as
Louisiana and Florida, practically all microbursts are accompanied by heavy rain (Fujita,
1985, p. 70). The precipitation in a wet microburst has high radar reflectivity, as opposed
states the diameter as less than 1 nautical mile, while Proctor (1985) seems to narrow the
diameter down in computer simulation to between 250 m and 4,500 m. Thus the
any event, the relationship between radar reflectivity and microburst strength is not
apparent (1985).
Avoidance of the microburst remains the most effective means of survival. At this
time, ground based systems surpass airborne systems in detecting microburst events.
Though wet microbursts are more easily discerned by remote sensing, all types represent
a danger, as witnessed in the accident record. "Wherever and whenever it occurs, and
regardless of its type, a microburst can cause an airplane crash, and should be taken
Many similarities, and some differences, occur with both types of microbursts.
Table 2 examines and compares some of the more common phenomenon and
Table 2
response. These responses tend to couple, one affecting the other, so a loss of airspeed,
for example, will reduce the lift, affect the longitudinal stability, and decrease the total
energy. To understand the events that an aircraft experiences when transitioning through
Flight Instruments
Most aircraft rely on air data, at least in part, to determine altitude, vertical speed,
and airspeed (Stengle, 1984, p. 199). In the microburst, pressure varies from atmospheric;
the Andrew's microburst had pressure variations of +0.4 mb (-0.12 inches Hg) to -0.2
mb (~ 0.06 inches Hg) (Fujita, 1983b, p. 4). This represents a difference in the indicated
barometric altitude of 120 ft below actual altitude to 60 ft above actual altitude. Vertical
speed indicators (VSI) are primarily pressure driven and can also be susceptible to error
(FAA, 1988, p. 27). Airspeed indications also rely on pressure sensations, but are less
dynamic pressure (United States Air Force [USAF], 1983, pp. 4: 15-16).
the microburst may induce sudden gusts which cause rapid fluctuations in the angle of
attack sensors (FAA, 1988, p. 28). In modern jet transport aircraft the angle of attack
sensors along with the Pitot-static system are inputs into an air data computer (Wild,
1996, p. 10: 21). Thus, a variety of aircraft instrument systems may be influenced by the
microburst.
74
Aircraft Behavior
particularities of that microburst, but where in the storm the aircraft penetrates, with what
speed and altitude, in what configuration, and other factors. Hence, only a general case is
considered here-the NTSB reports are replete with specific case studies.
an increasing headwind, which will increase the lift and energy of the aircraft. The
headwind will give way to a downdraft, which will decrease the lift and energy of the
aircraft. As the downdraft is escaped the aircraft will experience a tailwind, which
continues the diminution of lift and energy (Bristow, 2003, pp. 262-263). The wind
change may be significant enough to cause the wing to stall (McCarthy, Blick, & Bensch,
1979). Two thirds of the microburst is a hostile environment to the lift and energy of the
aircraft.
Figure 4. Effect of microburst on flight path. Note. From Influence of wind shear on the
aerodynamic characteristics of airplanes (p. 17), by D. D. Vicroy, 1988, NASA technical
paper 2827, Washington D.C.: NASA.
When ring vortices are introduced to the microburst (Figure 1), a series of
increasing and decreasing up and downdrafts, along with headwind and tailwind
components, influence the aircraft (Figure 2). Therefore, depending on the spatial
relationship of the aircraft trajectory to the microburst, the initial headwind increase may
not be present. This anomaly may hinder the ability to recognize a microburst encounter,
without turbulence. The flight crew of US Air 1016 may not have recognized the
microburst they were in due to the lack of turbulence (NTSB, 1995, p. 108). Survivors of
the Eastern 66 accident recalled minimal turbulence during their approach (NTSB, 1976,
p. 4), while Delta 191 reportedly experienced significant turbulence (NTSB, 1986, p. 18).
Turbulence complicates aircraft control and may induce unequal span-wise loading on
the wings, so that one wing stalls prior to the other (Melvin, 1986, p. 55). The control
problems of turbulence, though significant, are not the major factors affecting aircraft
performance.
in the dynamic stability of the aircraft (ICAO, 1987, p. 59). Longitudinal stability may be
excited, and even resonate, from the upset (McCarthy et al. 1979; Sherman, 1977).
Additionally, negative speed stability can present a control and performance problem
during the approach, possibly requiring a descent regardless of the altitude deficit
microburst (Vicroy, 1988, p. 13), but aircraft energy will also be affected. The airspeed
decrease reduces kinetic energy, and the altitude degradation will decrease the potential
energy of the aircraft. Total energy can be increased with added thrust, thereby mitigating
the effect of the energy absorbing microburst (Proctor et al., 2000). Thrust may become
trading potential energy for kinetic energy. A base amount of kinetic energy is required to
Lift
direction normal to the free stream velocity of the air (AIAA, 1992, f 1.7.2.8). Though
engine thrust may act in a direction which enhances lift, particularly at a high angle of
L = C L 0/2 P V 2 S) (1)
where CL is the coefficient of lift, rho (p) is the air density, V is the airspeed, and S is the
and at subsonic speeds depends mainly on wing geometry and angle of attack (Barnard &
Philpott, 1995, p. 20). Angle of attack is often denoted with the Greek letter alpha (a) and
represents the angle between the relative wind and a reference line, usually the chord line
or the fuselage centerline (Cashman, Kelly, & Nield, 2000, section 1). Increasing the
angle of attack will increase the coefficient of lift, the relationship being linear below the
stall angle of attack (Anderson, 1997, p. 212). The stall angle of attack is the value of a
for maximum usable lift (Chambers & Grafton, 1977). Larger angles of attack than the
stall alpha materially affect lift and may hamper longitudinal stability, to the extent that
The initial headwind in the generalized microburst will increase the airspeed and
hence increase the lift. As the headwind diminishes, the lift from airspeed decays. Further
decaying the lift is the downward flow. In the downdraft, the relative wind now strikes
the aircraft from above, thus decreasing the angle of attack (Figure 5) and reducing the
value of CL-
Wind ^ ^ ^
Downburst
i
^ _ | " | o e n . » s » i 2 a . ^ ^ . M , „ ; ,"wrfl "^.. r
—-—atfEsu- Previous Wind
Figure 5. Result of downburst on angle of attack. Note. Aircraft from The Boeing 737
Technical Site, by C. Brady, 2003, http://www.b737.org.uk/dimensions_737200.gif.
The loss of lift due to the change in a is dependent on the wing design,
configuration, and the initial angle of attack. When at high angles of attack, the reduction
in a produces less of a decrement in performance, than when the initial a is low. This
arises from the slope of the CL versus angle of attack curve for typical transport category
aircraft. For a Boeing 727, with flap 15 at approach speed and 5° a, a one knot down flow
will decrease the lift by 4.5 %, while at an initial a of 15°, a one knot down flow will
In addition to the loss of lift, the aircraft will become entrained in the downward
moving air. The vertical velocity of the downburst will subtract from climb performance
As the down flow is passed, and the outflow is encountered, the airspeed drops by
the magnitude of the tail wind, causing a loss of lift (Higgins & Baker, 1986, p. 43). The
airspeed, being a quadratic term in Equation 1, materially affects the lift generated. A
change in the lift force (Fujita, 1985, pp. 20-21). This alleviation in lift is common to all
The aerodynamic effects most influenced by the microburst are rapid changes in
lift and pitching moment (Weishaupl & Laschka, 2001, p. 265). Pitching moment
translates into the stability of the aircraft and can degrade performance.
Dynamic Stability
the aircraft's stability. There are five dynamic stability modes: three lateral-directional
modes and two longitudinal modes. The lateral-directional modes include the Dutch roll,
the spiral, and the roll mode. The longitudinal stability modes are the short period pitch
oscillation (SPPO) and the phugoid (USAF, 1980, p. 7.1). Transport aircraft are required,
per the airworthiness standards of FAR 25.181, to be highly damped in the SPPO and
Dutch roll mode between 1.2 times the stall speed (Vs) and the maximum allowable
79
speed (VMo) (FAA, 2003b). The phugoid and spiral mode stability, however, are very
Stability modes are excited whenever the airplane is disturbed from its
equilibrium trim state (Cook, 1997, p. 119). A microburst encounter will upset the
equilibrium trim of an aircraft and may cause a resonant response in the phugoid mode
The phugoid mode is an oscillation about airspeed coupling into pitch attitude and
height with a relatively constant angle of attack (Cook, 1997, p. 120). A small
disturbance in speed leads to a reduction in lift. As the lift is reduced the aircraft descends
and accelerates, when the aircraft accelerates it increases lift and climbs. When the
aircraft climbs its speed decays, reducing lift, and causing the sinusoidal series to
continue. During the oscillations, drag gradually lessens the amplitude until the motion
eventually damps out (p. 121). Jet transport aircraft are designed with minimal drag, so
properties. It is commonly assumed that the change (A) in angle of attack is zero, that is
Aa = 0 and that the thrust (T) is equal to the drag (D) of the aircraft, that is T = D and that
compressibility is negligible (Cook, 1997; McCarthy, Blick, & Bensch, 1979; VonMises,
1959). The frequency of the phugoid (C0p) in radians per second is given by:
Through the equation, it can be seen that the natural frequency of the phugoid is inversely
The phugoid may be stable and damped, or it may be divergent and aperiodic.
When damped the oscillations do not continue indefinitely but eventually fade out. The
length of time for one cycle divided by the time taken for the total number of cycles to
decay is the damping ratio. A heavily-damped oscillation has a damping ratio of 0.3 or
expressed as:
CP = (1/V2)(D/L) (3)
Where D is drag and L is lift (McCarthy et al., 1979, p. 11). Thus for any given lift, the
less the drag, the less phugoid damping available. A Boeing 737 in the landing
A perturbation in speed brought about by microburst winds can easily excite the
phugoid and start the airplane on its "roller coaster" type ride. The efficient design of the
modern aircraft lessens the phugoid damping, so that there is a greater deviation from the
nominal flight path than would occur with a well-damped mode (Stengel, 1984, p. 201).
If the microburst wind occurs at the same, or similar frequency, of the phugoid, a
resonant response can occur; causing correspondingly larger airspeed upsets (McCarthy
Hertz), which is a period of about 38 seconds. This period is within the temporal scale of
an airplane's traverse through a microburst. It is probable that a wind gust could initiate a
phugoid and then excite it at its resonant frequency (Frost, Turkel, & McCarthy, 1982, p.
1). The velocity perturbation at the resonant frequency for a B-727-200 series aircraft is
close to 20 decibels. If such an aircraft encounters a horizontal gust of only 4 knots wind
speed, at the angular frequency of 0.026 Hz, the aircraft will respond with an airspeed
deviation of approximately 40 knots (McCarthy et al, 1979, p. 10). That is, the speed will
be 100 kias at the high point on the phugoid wave, while at the point of minimum altitude
600 ft AGL the aircraft, a Boeing 727-200, encountered a headwind gust of 25 knots,
which soon subsided to 20 knots and then in four seconds dropped to a 5-knot headwind
(NTSB, 1976, p. 17). The time period from initial upset to the steady state was 19
seconds. This is a half sine-wave of frequency 0.026 Hz (McCarthy et al., 1979), with an
amplitude of 15 knots. McCarthy et al. (1979) conclude that the JFK accident is
"associated with the airplane's encounter with a horizontal wind containing high energy
at the airplane's critical phugoid frequency, which caused a sudden extreme variation in
Examining the microburst accidents, one predominant theme manifests itself: The
question, 'why would an aircraft trying to climb impact terrain at a speed well above its
stall speed, and in many cases, above its go-around speed?' Eastern 66 hit at about 130
kias, close to its reference speed for the approach (NTSB, 1976, p. 7), but V s for their
82
configuration of flap 30 was about 97 kias. Pan Am 806 flew into the jungle at 140 kias
while VR E F was 135 kias (NTSB, 1977, p. 23). Allegheny 121 impacted the ground at 155
kias while the reference speed was 122 kias and V s was 110 kias (NTSB, 1978, p. 2, 15).
PAA 759 had a V2 speed of 151 kias and impacted the trees at 149 kias, following an
airspeed increase of 18 knots in about six seconds, while V s was still lower at 122 kias
(NTSB, 1983, p. 33, 59). Delta 191 had a VREF of 137, a V s of 105 kias, and impact
occurred at 169 kias (NTSB, 1986, p. 7; Fujita, 1986). Finally, USAir 1016 had a VREF of
121 kias with flap 40, a go-around target speed of 128 kias with flap 15, and impact
At the top of a phugoid oscillation the aircraft will initiate a descent with
increasing airspeed even while flying at the maximum angle of attack (Melvin, 1986, p.
52). Gera found (1980) that even a closed loop system was not damped in phugoid
oscillation by the pitch attitude, and that attitude control in stabilizing the divergence was
ineffective (p. 11). Melvin (1986) found the phugoid continued after the shear boundary
and the oscillation was not preventable by the pilot (p. 52). McCarthy et al. (1979),
hypothesized that the phugoid may result in airspeed oscillations of a nature that would
be difficult to control, possibly leading to stall, or other disastrous results (p. 29).
Sherman (1977) found the phugoid could become aperiodic and unstable from a wind
shear encounter (p. 1). In the aperiodic mode, the phugoid continually diverged from the
equilibrium, and in the unstable dynamic system the phugoid continued to grow about the
stability. Slowing the airplane down increases stability (Sherman, 1977, p. 9).
83
Alternatively, holding airspeed constant in wind shear decreases changes in airplane
stability, however, for wind shearing from tailwind to headwind, the increasing of flap
hazard to aviation in general, a wind profile that resonates one aircraft type may not
resonate another (Stengel, 1984, p. 201). Sherman (1977) found the wind gradient the
most important factor-not the wind speed (p. 13). The higher the speed of the airplane the
smaller the wind gradient required for the onset of unstable conditions. Therefore,
transport aircraft, with their higher approach speeds, are much more susceptible to the
performance restraining effects of wind shear then many other aircraft types (p. 8).
The hazards of the microburst include loss of lift and the effects of the phugoid
mode, which can drive an airplane into the ground even with an increasing airspeed. "It is
emphasized that it is the combination of the phugoid excitation, and the severe
downdraught [sic] that makes the downburst such a treacherous phenomenon for an
Speed Stability
Speed stability arises from the aircraft's response to total drag. Total drag, for a
subsonic airplane, can be conceptualized as the sum of boundary layer, or profile drag
and trailing vortex, or induced drag (Anderson, 1997, p. 73). Profile drag increases with
speed, while induced drag decreases with speed. The minimum point in the summation of
these two forces is the minimum drag speed and also the point of neutral speed stability.
84
Neutral speed stability is the point of change between negative speed stability and
positive speed stability. That is, speed errors will grow below this speed, but die out at
higher speeds (Etkin, 1972, p. 480). From the neutral point, a reduction in speed will
increase drag. This increase in drag will grow as the speed decreases until the stall speed
is reached and the aircraft stops flying (Barnard & Philpott, 1995, p. 327). In the region
of negative speed stability, a change in speed will cause the aircraft to diverge from its
trimmed state.
Positive stability occurs at speeds greater than the neutral point. The aerodynamic
forces on the aircraft cause the airplane to respond in pitch toward the original trimmed
airspeed (Higgins & Baker, 1986, p. 43). The aircraft will pitch down and accelerate to
recover a loss of airspeed and pitch up and decelerate to regain the original trimmed
airspeed (ICAO, 1987, p. 59). In the region of positive speed stability, a change in speed
will cause the aircraft to converge toward its original trimmed state.
Takeoff and landing speeds are close to neutral speed stability for jet transport
aircraft. The Boeing 727-200, at 140,000 pounds and flap 30, has a VREF of 128 knots
which is the minimum drag speed (Higgins & Patterson, 1979, p. 3). Flight below this
speed requires additional thrust to overcome drag. Colloquially this is known as the
in normal flight an increase in speed is attained with an increase in thrust (Hurt, 1965, pp.
353-357). Airspeed degraded by wind shear below VREF will therefore require more
thrust to maintain the lower speed then was required to maintain VREF. If acceleration
back to the landing reference speed is desired, an additional amount of thrust is required.
In the region of reverse command, wind shear can quickly saturate the thrust available.
85
Performance
The hazard from wind shear arises from the maximum performance capability of
the aircraft being temporarily exceeded by the downdraft environment (Higgins &
Roosme, 1977, p. 15). Climb potential and airspeed are important facets of performance
The ability to climb is based on excess thrust (Hurt, 1965, p. 152), and therefore is
dependent on the drag curve and the maximum thrust available curve. Two speeds
become evident, the speed corresponding to where the difference between thrust and drag
is a maximum, and the speed corresponding to where the product of speed and thrust
subtract drag is a maximum (Higgins & Patterson, 1979, p. 4). The first speed is the best
angle of climb speed and produces the greatest gain in altitude for a given horizontal
distance, the second speed is based on time and is the best rate of climb speed, it is the
greatest altitude gained in a unit of time (Barnard & Philpott, 1995, pp. 213-218; Dole,
At 140, 000 pounds with flap 30 and gear down, the Boeing 727 has a best angle
of climb speed of 124 kias with maximum thrust, producing a 1,650 fpm vertical
velocity. The best rate of climb is achieved at 140 kias enabling a climb of 1,750 fpm.
VREF, between these two speeds at 128 kias, generates a rate of climb of 1,690 fpm
airspeed into altitude through zooming (Dole, 1988, p. 86; Hurt, 1965, p. 150). When
flaps are raised the climb ability improves (Higgins & Roosme, 1977, p. 8). For the 727
at 140,000 pounds, the rate of climb at VREF increases 500 fpm when transitioning from
86
flap 40 to flap 30. When the landing gear is raised and the flaps are moved to 25 the rate
of climb increases 300 fpm (Higgins & Patterson, 1979, p. 9). This last increase is not as
great because the aircraft is configured for higher speed operations, but is being operated
at low speed. If the airspeed were allowed to accelerate to 160 kias, the best climb speed
Accelerating the aircraft is energy intensive. For the 727 at 140,000 pounds and
flap 30 at VREF the climb rate is 1,650 fpm. If an acceleration rate of 2.5 knots per second
is desired, the climb capability will be cancelled out completely (Higgins & Patterson,
1979, p. 6). In a wind shear environment, if the pilot wishes to regain lost airspeed, the
climb rate must be sacrificed. At VREF and 100% thrust, the 727-200 will lose 650 fpm
for each knot per second acceleration (p. 7). Boeing recommends not accelerating in a
wind shear because of the great loss in climb rate (Higgins & Roosme, 1977; Higgins &
Patterson, 1979; Boeing Windshear Task Force, 1985; Higgins & Baker, 1986).
Just as accelerating the aircraft reduces the climb rate, zooming the aircraft will
increase the climb rate, if only momentarily. When the aircraft zooms and uses its kinetic
energy it will be left at a low airspeed with a resultant decrease in climb performance
(Webb, 1990, p. 206). Once the airspeed has been decayed it is very difficult to regain.
The problem arises at speeds below the neutral point, because drag decreases excess
thrust (Proctor et al., 2000, p. 484) and consequently impedes climb ability. Though
transforming the kinetic energy into potential energy may be beneficial in the short term,
there might not be enough climb performance remaining at the lower airspeed to arrest
the descent rate in a severe downdraft (Higgins & Roosme, 1977, p. 9).
87
The role of energy is an important concept in the performance of aircraft-from the
kinetic energy that governs lift, to the potential energy of climb, and the trading of the
F-factor
The F-factor is an attempt to quantify the microburst hazard for aircraft in flight,
it is a numerical index derived from the aircraft total energy and its potential rate change
Where Ux is the component of atmospheric wind directed along the flight path, and dUx is
the derivative with respect to time, g is the gravitational constant, w is the vertical wind,
The equation is based on air mass kinetic energy, as an airplane's ability to climb
is a function of airspeed, and not groundspeed. A descending tail wind will decrease the
energy state of the aircraft and so will create larger values of the F-factor. The term dUx
is a function of the meteorological event and the aircraft trajectory (Proctor et al., 2000,
p. 483). The aircraft trajectory is further a function of the thrust to weight ratio of the
particular aircraft, which makes the F-factor unique to each make and model. Typically,
twin-engine aircraft have the highest thrust to weight ratio in the transport category and
so will have a lower F value for a given microburst than will three or four engine aircraft.
To avoid spurious and nuisance warnings the F-factor is averaged over the
distance of one kilometer by means of integration. Values greater than 0.1 of the
averaged F-factor are considered hazardous by the FAA. A "must alert" threshold is
established for aircraft equipped with the F-factor matrix at 0.13 (Proctor et al, 2000, p.
484).
In determining various microburst severities, the F-factor was computed for the
Delta 191 and US Air 1016 accidents. The Delta L-1011 experienced the averaged F-
factor of 0.23, with an instantaneous F-factor greater than 0.35. The DC-9 of US Air
value at 0.27, while the FDR derived winds are greater than 0.3 (Proctor et al., 2000,
pp.486-487; NTSB, 1995, p. 48). These microbursts were up to three times greater than
The F-factor is a useful tool in determining how a microburst will affect a specific
aircraft. Its application lies not so much in constructing past encounters, but in
quantifying future events while aiding the flight crew in the decision making process.
Aircraft performance is impeded in wind shear and the F-factor combines many of the
form.
Aircraft Categories
Each aircraft has different inertial and aerodynamic characteristics; wind fields
that are hazardous for one aircraft type may be less hazardous for another type. Aircraft
with high airspeed and wing loading (e.g. transport category aircraft) appear to be more
sensitive to gradients in head/tail wind, while aircraft with low airspeed and wing loading
(e.g. general aviation aircraft) are more adversely affected by downdrafts (Stengel, 1984,
p. 201). General aviation aircraft are also much lighter and have less inertia to overcome
performance-wise. The type of power plant also aids in this, as a reciprocating engine
will provide very quick acceleration while a jet engine takes time to spool up. The JT-8D
turbojet engine requires about 8 seconds to develop full power from an idle power
setting. Propellers also develop a localized airflow, aiding lift and decreasing alpha.
Additionally, general aviation aircraft have different approach speeds and phugoid
values, both in frequency and damping. For a multitude of performance reasons, general
other. A change in airspeed will change the lift force, which may excite the stability,
affecting the altitude and airspeed, which will further effect the climb performance and
handling qualities, ad infinitum. These variables are dependent not only on the type of
aircraft, but also the microburst environment. The value in a static analysis allows one to
encounter, cannot completely reveal the performance capabilities of the aircraft (Stengel,
1984, p. 200). To counter this limitation, computer simulations, flight simulations, and
even real world microburst encounters have been carried out (e.g. NASA, 2002). These
simulations and real world experiences have provided additional insights into
substantial, and as witnessed through the accident record, can be catastrophic. To counter
encounters continue though, as the case of US Air 1016 demonstrates, so that the escape
which pilots are called upon to weigh hazards against mission objectives. Because
the future is uncertain, there will be instances when the pilot presses on even
though hindsight will prove that to have been the wrong choice." (Stengel, 1984,
p. 198)
The purpose of the escape maneuver is to provide the performance necessary to fly out of
a microburst.
Recognizing the microburst is the first action required in commencing the escape
maneuver, the Boeing Company Windshear Task Force (1985) found that in a typical low
level wind shear encounter about 5 to 15 seconds were available for recognizing a
especially when below 1000 feet AGL (United Air Lines [UAL], 1991, p. A-29). An
• Throttle position not correlating with normal position for extended period of time
escape procedure should be executed without delay (American Airlines [AA], 1990, p.
3A-25).
independent wind shear study in 1982 (Ireland & Simmon, 1986, p. 27). Commencing
after the accident of Pan Am 759 in New Orleans, this two-year research led to the
development of the WSTA (p. 28). Eventually funded by a grant from the FAA, the study
lost its independence and developed into a consortium including the Boeing Company,
and Heliwell Incorporated (FAA, 1988, f 3). With the three major jet aircraft
manufacturers present, it is not surprising that the guidance across the various aircraft
procedure is to simultaneously set maximum thrust and rotate to an initial 15° pitch
attitude. The flight path is controlled as necessary with pitch, while the configuration of
the aircraft remains unchanged during the recovery (FAA, 1988, p. 48; UAL, 1991, p. A-
mask an impending wind shear (Boeing Company, 1985, p. 7); therefore, a throttle
Applying thrust introduces energy into the system and mitigates the degrading
effects of wind shear (Visser, 1997, p. 5). Engine overboost is permitted to avoid ground
contact, but should be discontinued when flight safety has been assured (FAA, 1988, p.
46; UAL, 1991, p. A-29; AA, 1990, p. 3A-25). The JT8D engine (installed on the DC-9,
B-727, and B-737) is capable of providing a 10% increase in thrust when firewalled.
Though the Nl speed and exhaust gas temperature (EGT) would exceed operating
(Boeing, 1985, p. 04.20.16A). Boeing warns "[this] should only be considered when all
other available actions have been taken and ground contact is imminent" (p. 04.20.16A).
Pitch
pitch attitude of 15° nose up (FAA, 1988, p. 46). The initial attitude may be bound by the
lower value of stick shaker or stall buffet, which should always be the upper limit of pitch
attitude. When stick shaker or buffet stops, the attitude should be increased in 2°
increments up to 15°, and beyond if the flight path is unacceptable (p. 46).
Wanting to prevent premature arrival at stick shaker, United limited the pitch
attitude to 15° in their simulator trials (Melvin, 1986, p. 51). Though a range of recovery
attitudes provide good performance, the FAA maintained the 15° limit because it is easily
operators using target pitch attitudes may use a pre-calculated value in place of the initial
attitude (p. 45). The slow 2° change in pitch attitude, when required, replicates a more
optimal technique, designed to avoid arrival at the stall too quickly, and it also diminishes
the chances of an over-rotation in pitch (Melvin, 1986, p. 51; Boeing Company, 1985, p.
10).
The 15° attitude is only the initial target. If the flight path continues to deteriorate
satisfactory flight path is attained or intermittent stick shaker is reached (FAA, 1988, p.
46; UAL, 1991, p. A-29; AA, 1990, p. 3A-25). As in normal operations, vertical flight
path control is maintained with pitch attitude. It is recommended not to use more pitch
than is necessary to control the vertical flight path (AA, 1990, p. 3A-25; UAL, 1991, p.
A-29). Though jet transport category aircraft have climb capability, in still air, at
intermittent stick shaker, the high pitch attitude results in high drag and a minimal climb
rate (Higgins & Patterson, 1979, p. 9). Additionally, if all the airspeed is bled off in the
maneuver there will be no reserve to soften the impact with terrain if ground contact
There are a number of difficulties of flying near stick shaker speed, and pilots
have had little training in this area (Webb, 1990, p. 206). Problems occur if the airspeed
decay rate is too great and the airplane stalls before a pitch change is possible; this may
turbulence or shear. The effects of heavy rain are also problematic, causing an increase in
the airplane's stall speed and leading to the stick shaker speed being underrepresented,
with a possible stall occurring before warning is given (NRC, 1983, p. 56). Additional
performance concerns of flight near the stall angle of attack are the airspeed and altitude,
which decay rapidly in this high drag region of flight. Near the stall angle of attack
p. 202). It is therefore advised to delay the onset of stick shaker as long as possible and
then only when ground contact appears imminent (Melvin, 1986, p. 51).
The stick shaker is generally calibrated per the stall angle of attack. On the
Boeing 737-200 "the stick shaker is actuated at a relatively fixed angle of attack for a
specific flap setting" (Boeing, 1985, p. 40.40.02). Flying at the stick shaker is a maneuver
that excites the phugoid because the angle of attack is held fairly constant. That there is
no change in angle of attack (Aa = 0) is the major assumption for the reduced-order
model. Hence, the phugoid is easily propagated in this flight attitude, leading to the
problem previously discussed, mainly an oscillation in airspeed and altitude which may
cause the aircraft to crash even with an increasing airspeed (Melvin, 1986, p. 57).
Configuration
advice when escaping an encounter is to maintain configuration (FAA, 1988, p. 47). The
FAA has acknowledged that a performance increase may be available with the extension
of flaps (1988, p. 47). In normal operations flaps are raised during a go-around or
departure, this reversal of procedures may lead to confusion. It is felt the risk of moving
the flaps in the wrong direction is greater than the risk of encountering a shear so great
that a flap change is needed for recovery (UAL, 1991, p. A-29; AA, 1990, p. 3A-26).
A performance increase occurs after landing gear retraction (FAA, 1988, p. 47).
through a rise in drag as the gear approaches the body of the aircraft and the gear doors
open and close. Therefore, the increase in performance from gear retraction may be offset
by the initial decrease in performance and it is recommended to leave the gear in its
requirement, per FAR 91.183(b)(c), to report the encounter to air traffic control (FAA,
event may prevent a subsequent accident. It is also advisable to land at the nearest
suitable airport in point of time if the engines have exceeded their design tolerances
the effects of the microburst, it was chosen partly for simplicity and ease of recall (FAA,
1988, p. 45). More complex techniques may make better use of airplane performance,
admits the FAA in their publication AC 00-54, Pilot Windshear Guide (p. 45). This is
born out in a number of independent studies demonstrating optimal and near optimal
procedures increasing the altitude of penetrating aircraft (e.g. Dogan & Kabamba, 2000;
Hinton, 1988; 1989; Miele, Wang, Tzeng, & Melvin, 1987; Mulgund & Stengel, 1992a).
seek to minimize the altitude loss in a microburst encounter while keeping the airspeed
above stall.
Alternate Guidance Maneuvers
microburst detection and avoidance. Escape procedures have not had the benefit of
equivalent scrutiny. United's wind shear efforts from 1982 to 1984 may have
unintentionally led to this shortfall in research. It was United's feeling that differences of
opinion within the industry were impeding development of wind shear training and that a
consensus was required for a training package to be developed (Ireland & Simmon, 1986,
p. 28). With the endorsement of the major airframe manufacturers, and subsequently the
FAA, there has been little incentive to develop alternate procedures. As such, this area of
study has become the realm of individuals in university settings, and to a much lesser
minimizing the altitude loss within the constraints of airspeed (Visser, 1996; 1997;
Mulgund & Stengel, 1992b; Miele et al., 1987). Optimization requires a global
knowledge of the microburst winds; that is the wind components at all points in the
aircraft's trajectory must be known in advance (Mulgund & Stengel, 1992b, p. 2).
Guidance studies and laws assume only local information (Miele et al., 1987, p. 485).
The state of art of microburst detection limits the knowledge of the wind field so that
performance over the constant pitch maneuver in simulated cases (Miele et al., 1987;
Penetration landing laws attempt to maintain an approach profile, and land the
aircraft in the microburst winds. This strategy uses varying thrust and pitch. Lateral
escape maneuvers use a global knowledge of the microburst and avoid the most severe
trajectories evaluate the performance of the aircraft in the vertical plane, with global, or
local knowledge of the wind field being used depending on the study.
Penetration landing makes sense only if the wind shear encounter occurs at lower
altitudes (Miele, Wang, & Melvin, 1988, p. 153). At low altitude the aircraft may only
have to traverse a section of the microburst, whereas an aborted landing may lead to
greater hazard, traversing the whole of the shear region at low airspeed and altitude (p.
154).
Penetration landing guidance uses pitch control to maintain the nominal glide path
while thrust control augments the approach by keeping the aircraft from running out of
airspeed (Psiaki & Park, 1989, p. 1131). When the control laws are used with global
knowledge of the microburst, full thrust is commanded at the headwind section (p. 1132).
certainly the accident case histories bear out the dangers of trying to land in a microburst
wind. If the initial altitude is high enough the abort landing is clearly a safer maneuver
adverse conditions, however, global knowledge of the microburst is not yet available and
there is a danger of steering into the core rather than away (Melo & Hansman, 1990, p.l).
An incorrect turn towards the microburst core is more hazardous than straight flight;
lateral maneuvers are therefore limited to the availability of precise information about the
microburst (p. 6). If lateral maneuvering is employed, the optimum bank angle, per
computer simulations, is limited to 10° (Melo & Hansman, 1990, p. 3; Visser, 1996).
advance warning is provided (Visser, 1996, p. 115), in which case the most adverse
microburst winds can be avoided. As yet, the obstacles include the uncertainty of the core
Longitudinal maneuvers constitute a practical reality with current technology. The two
basic longitudinal maneuvers are the constant pitch maneuver, previously discussed, and
the variable pitch maneuver, which seeks to optimize the flight path.
The basic philosophy of the variable pitch guidance maneuver is the minimization
of the time spent in the shear environment and thereby the reduction of the effects of
wind shear on aircraft performance. Minimizing the time in the shear is accomplished by
sacrificing altitude for speed (Bray, 1986, p. 13). This differs from the constant pitch
trading speed for altitude, in hope of exiting the shear (FAA, 1988, p. 45).
Variable Pitch Guidance Maneuver
minimax equation; this term being derived from minimizing the maximum value of
longitudinal guidance strategy (Visser, 1997, p. 1). The maneuver is robust with respect
to uncertainty in microburst strength, with little sacrifice in altitude (p. 11). This guidance
easily outperforms constant pitch guidance in terms of both altitude and energy
management (p. 7), while climb rate guidance was found to be minimally useful (p. 6).
The target altitude maneuver incorporates three phases; a descending flight phase
to a target altitude, the maintenance of horizontal flight, and ascending flight after the
aircraft has passed through the shear region (Miele et al., 1987, p. 483). In providing
uniformity with the constant pitch guidance strategy this technique is referred to herein as
the variable pitch guidance strategy, as pitch attitude continually changes throughout the
maneuver.
Similar to the constant pitch maneuver, when the wind shear is acknowledged, the
thrust setting is commanded to a maximum value (Miele et al., 1987, p. 485) introducing
energy into the system. Thereafter, the only control variable available to the pilot is angle
of attack, which is manipulated indirectly by pitch attitude. Controlling pitch, the pilot is
able to alternately trade altitude for airspeed, or potential for kinetic energy.
Descent Phase
The Chebyshev solutions of optimal control trade altitude for airspeed in the
initial phase of the microburst (Visser, 1996, p. 117; Visser, 1997, p. 4). This may appear
decrease the angle of attack at the outset of a microburst (Dogan & Kabamba, 2000;
Melvin, 1986; Bray, 1986). The constant pitch technique fails to exploit the energy gain
The initial altitude represents an energy component that can be converted to speed
with which to extend endurance in the shear (Bray, 1986, p. 16). The aircraft is guided to
Kabamba (2000) or by flight path or angle of attack as in Miele et al. (1987). Optimally
these controllers take into account the microburst winds with pitch attitude being changed
accordingly. The rate of descent is very high for the dive portion of the flight, averaging
2300 fpm. The descending flight path is flown entirely in the shear portion of the
trajectory (Miele et al., 1987, p. 493), thus maximizing the airspeed and minimizing the
time spent in the shear. The aircraft is leveled out at a predetermined optimal altitude and
Level Phase
The horizontal branch is flown partly in the shear and partly in the after shear
portion of the trajectory (Miele et al., 1987, p. 493). After the dive phase, the aircraft is at
a relatively high airspeed and uses this kinetic energy to maintain the target altitude. As
the airspeed decays, the pitch is adjusted to maintain altitude until the shear boundary,
101
which is ideally at the point of minimum airspeed. In the event that stall a is reached
prior to the shear boundary, the altitude is allowed to decay. In optimized studies this
does not happen, as the extent of the microburst is known and altitude and speed are
judiciously controlled.
The less time spent at the stall angle of attack, the better the performance (Bray,
1986, p. 18). Flight at maximum a is not efficient, and "it is very bad to use up available
The optimal altitude is chosen through the solution of the Chebyshev equation.
Miele et al. (1987) reformulated this as a Bolza type integral, and using the velocity of a
strong, but realistic shear of 140 ft per second, simplified the equation for the target
Where ho is the initial altitude of the encounter, and AWX is the change in the horizontal
component of wind velocity, or wind shear. Units are in feet, so that a shear on the order
of 140 feet per second at an altitude of 1000 feet will yield a target altitude of 400 feet.
From Equation 5 it is apparent that the target altitude is a function of the microburst
severity and the altitude of the encounter. Greater microburst intensities lead to higher
target altitudes. Higher initial altitude also leads to higher target altitude, and it provides
greater opportunity to convert potential energy into kinetic energy; hence, identifying the
microburst being the strongest to date, a simplified target altitude equation emerges. By
keeping the target altitude slightly higher than the optimal altitude, Miele et al. (1987)
introduced a simplified guidance strategy. Assuming the worst-case scenario, the aircraft
With Equation 6, neither global nor even partial knowledge of the shear is required. This
near optimal guidance works well for moderate to severe microbursts, but is overly
The simplified case of using one of the most intense wind shears provides a
shears are between 37 fps and 43 fps (Fujita, 1985, p. 63), the frequency of a greater wind
decreasing exponentially as the wind speed increases. The relationship between the
^probability of an encounter and the wind speed for the JAWS data indicate (p. 64):
Where P is the probability, and W is the wind shear in meters per second. Equation 7
suggests that a velocity of 140 fps (~ 46 m/s) has a probability of 0.000116627. This low
value, of 1 occurring per 8,574 microbursts, translates into a robust altitude floor for the
just reaches its limit as the high shear region is exited (Visser, 1997, p. 2). When the
path.
Climb Phase
Once the aircraft has transitioned through the shear it is desirable to gain altitude.
Miele et al. (1987) use a flight path corresponding to the steepest climb condition in
quasi-steady flight for their modeled aircraft, a B-727-200 advanced, giving a path
inclination of 7.431° (p. 481). Dogan and Kabamba (2000) use a pitch attitude of 15° once
the energy drain has abated (p. 421). Visser (1997) considers a climb from target altitude
in terms of an aircraft's instantaneous available climb performance (p. 6). As the danger
of the microburst is past, the technique for climb-out resides with the discretion of the
operator.
Evaluation
maneuver based on altitude guidance. Visser (1997) maintains "large altitude excursions
tend to raise the anxiety levels experienced by pilots" (p. 9). Additional benefits of the
variable pitch maneuver include the reduction of the probability of the phugoid mode,
controls less likely to saturate, and improved control of the flight path.
Bray (1986) found that the constant altitude, variable pitch attitude demonstrates
improved performance compared with constant pitch (p. 18), and that superior
performance in low-level wind shear involves controlling the flight path of the aircraft to
minimum altitudes (p. 20). This conclusion was supported recently in Dogan and
Kabamba (2000); "even if pitch guidance is used with the intention of immediately
increasing altitude, the minimum altitude reached during the escape maneuver is very
likely to be lower than it would be if dive or altitude guidance [variable pitch] were used"
(p. 425).
Takeoff Case
The benefit of the variable pitch maneuver is also apparent in the examination of
the takeoff phase. Similar to the approach to landing case, the takeoff case commands a
constant altitude through the wind shear with a climb-out when the shear's boundary has
been traversed (Bray, 1986; Hinton, 1988; Melvin, 1986). Unlike the approach case, the
Optimal trajectories for the takeoff case are characterized by an initial decrease in
angle of attack (Melvin, 1986, p. 53) with a push-over to a linear flight path (Bray, 1986,
p. 18). The linear, or horizontal flight path, is controlled with pitch, and as airspeed
decays a gradual increase in angle of attack occurs, similar to flaring, until the shear ends
(Melvin, 1986, pp. 53-54). These flight paths are superior in survivability when
microburst simulation, the constant pitch maneuver was not survivable when the variable
pitch recovered at 60 feet (p. 18). Hinton (1988) produced similar results, with a constant
pitch maneuver impacting the terrain, while the enhanced flight-path-angle strategy
(variable pitch) lost only 4 feet during the microburst encounter (p. 8).
If the aircraft is above the minimum altitude, a dive should be initiated. The
minimum altitude used for the studies varied from 200 feet (Bray, 1986, p. 17), to 100
feet (Hinton, 1988, p. 4). Diving the aircraft and then maintaining a level altitude
105
minimizes the time in the shear and the time spent at stick shaker speed. The activation of
stick shaker should be postponed because the airplane phugoid mode is excited in this
Though the minimal altitude is different in the takeoff case, the flight phases and
the philosophy of exiting the shear in the minimum amount of time are similar to the
abort landing case. Each maneuver ideally contains a descent phase, which is completely
in the shear, a level flight phase, which is partly in the shear and partly beyond the
boundary, and a climb phase, which occurs outside of the microburst. Since thrust is at a
maximum the only control available to effect a flight path change is pitch attitude. From
the control perspective, there is no difference in the variable pitch guidance strategy
As optimal studies take into account the whole of the microburst, comparing them
to a maneuver which does not, provides little insight. Since optimization in a real world
environment is not yet realized, the value resides in the development of simplified
guidance, which approximates the optimal trajectory (Miele, Wang, & Melvin, 1988, p.
154). The more functional comparison of safety is between the constant pitch guidance
maneuver and the variable pitch guidance maneuver, which does not take global
Qualitative Assessment
maneuvers (FAA, 1988, p. 45). The initial pitch to 15° does not exploit the energy
possibilities of the aircraft and may lead to premature arrival at the stick shaker. Flight at
stick shaker speed is inefficient and ALPA's Airworthiness and Performance Committee
opposes early arrival at stick shaker speed (Melvin, 1986, p. 51). "Deliberately flying to
the stick shaker angle of attack when ground impact is not imminent is extremely
dangerous" (p. 58). Melvin additionally reports that "many pilots have been encouraged
to rapidly increase the angle of attack to its limiting value in hopes of magically escaping
a wind shear. In reality it reduces their chances for escape" (p. 54).
The variable pitch guidance maneuver is optimized for one wind shear intensity;
below this value it will be too conservative and allow a lower altitude than the constant
pitch maneuver, while greater intensities will cause a pitch-up to the stick shaker prior to
escaping from the microburst. Greater shears than the optimized value of 140 fps (~ 46
m/s) have been recorded; the Andrew's microburst generated wind speeds greater than
190 fps (~ 62 m/s) (Fujita, 1983b, p. 6), and the US Air 1016 microburst was calculated
to have a maximum wind velocity change of 145 fps (~ 47.6 m/s) along the north-south
axis (NTSB, 1995, p. 48). From Equation 7 it can be inferred that though these strong
microbursts are rather anomalous, they are present and constitute a great hazard to
aviation.
A limiting assumption of the variable pitch guidance maneuver is that all low-
level wind shear is a microburst. A sea-breeze front, for example, may prompt the aircraft
to descend and accelerate into relatively still air at a low altitude. Though this is not a
great problem, it may increase the damage of a bird strike or violate noise abatement
procedures.
Quantitative Assessment
terms of efficiency, was computed using the optimal trajectory as the criterion. The
guidance trajectories all used the variable pitch maneuver, though target altitude and
pitch were controlled differently depending on the extent of the wind shear knowledge-
global, local, or none. With the optimal trajectory set at 100% efficiency, the guidance
trajectories with some local knowledge of the wind shear were 90% to 99% effective,
while the escape maneuver without local knowledge was 82% to 90% effective, and the
constant pitch maneuver was 73% to 79% effective (p. 500). A maximum angle of attack
maneuver was also evaluated, which produced an efficiency of 42% to 51% (p. 500),
demonstrating that premature arrival at the stick shaker is not an efficient strategy (e.g.
Melvin, 1986).
confirmed trading altitude for airspeed in the initial phase of the escape maneuver
reduced the risk of crashing (p. 425). Their version of the variable pitch maneuver, which
they termed h-guidance, was similar in concept to the maneuver described in Miele et al.
(1987), the chief differences being the dive controller and the climb phase. In a computer
simulation, the constant pitch guidance cleared the ground by 102 ft (33.45 m) in a
moderate to severe microburst, while the variable pitch maneuver generated a minimum
altitude greater than 183 feet (60 m) (p. 422). When the microburst intensity was
increased by 25% the constant pitch guidance caused a crash, while the variable pitch
microburst strength, when the command altitude (target altitude) was low (Dogan &
Kabamba, 2000, p. 422). The probability of a crash was computed using the Monte Carlo
method with variable pitch having less probability of a crash than other maneuvers when
crash, with confidence parameter 5 at .05, is about .25 with constant pitch guidance and
as low as .12 with variable pitch guidance (p. 423). The variable pitch guidance
maneuver provides greater authority over the minimum altitude reached during the
The special case of propeller driven aircraft was investigated by Mulgund and
Stengel (1992a), and an increase in performance was demonstrated with a variable pitch
maneuver. The minimum altitude for the variable pitch maneuver was 52 feet higher than
the altitude generated by the constant pitch maneuver (p. 7), which in this study was 17°
rather than 15°, as it was found to provide better performance (pp. 5-6). Though the
minimum airspeed of the variable pitch maneuver was lower by 2 knots, the total energy
was higher by 5.7% and the maximum angle of attack was lower by 1.5° (p. 7). The
performance increase for different aircraft types demonstrates the robustness and value of
The same authors, Mulgund and Stengel, examined a jet transport aircraft's
performance through a microburst in a later work (1992b). Again they found that an
optimal maneuver of varying pitch attitude provided a higher minimum altitude than the
constant pitch strategy. For the Boeing 737-100, the minimum altitude was 400 ft for the
variable pitch maneuver and 350 ft for a 15° constant pitch maneuver (p. 7).
109
When lateral escape was considered, the variable pitch maneuver still
outperformed the constant pitch maneuver in altitude by about 80 ft (25m) (Visser, 1997,
p. 8). The specific energy of the variable pitch maneuver also was greater than the
constant pitch maneuver during the microburst encounter in which the core was not
penetrated (p. 8). Vissefs study (1997) demonstrates that the variable pitch maneuver is
superior in conserving altitude and energy whether the microburst core is penetrated or
avoided.
encounters-the variable pitch maneuver out performs the constant pitch maneuver in
these scenarios. The takeoff case is similar. Hinton (1988) compared five recovery
strategies and found the variable pitch strategy (his enhanced flight-path-angle strategy)
commanding the best overall performance (p. 10). In a strong shear, of 84 knots, the
variable pitch maneuver cleared the ground by 52 feet, while the constant pitch strategy
caused an impact with the ground (p. 25). In weaker shears, the variable pitch was not as
good as some other strategies, but it always out performed the constant pitch maneuver
Simulator Assessment
To validate his earlier findings, Hinton (1989) conducted a flight simulator trial of
the strategies examined, using NASA's Visual Motion Simulator, replicating a Boeing
737-100 aircraft (p. 6). The performance of the piloted simulation was generally less than
that of the batch computer simulation for any given recovery strategy. The constant pitch
strategy was 36 to 57 ft less in the real-time simulation than that of the computer
simulation (p. 8). The variable pitch strategy was the most irregular, being 100 to 104 ft
110
in the computer simulation and varying in the flight simulation from 114 ft to 29 ft (p. 8).
The poor correlation was attributed to errors in pilot tracking, variations in aircraft and
microburst state parameters, and a slightly lower performance of the simulator computer
(p. 9).
The average minimum altitudes during the flight simulation were 79.8 ft for the
constant pitch strategy, 82.2 ft for an acceleration strategy, and 86.9 ft for the variable
pitch strategy (Hinton, 1989, p. 7). These takeoff case scenarios were not statistically
A statistically significant difference in minimum altitude was found between pilots at p <
.01 (p. 29), and also in the root mean square (RMS) of the pitch error between pilots at
With a 10% increase in the wind shear velocity, a significant difference in escape
strategies was present at p < .01 (Hinton, 1989, p. 30). In this analysis, the constant pitch
maneuver mean altitude was 4 ft with a standard deviation of 9.3 ft. The acceleration
guidance altitude was 0 ft with a standard deviation of 0 ft, while the variable pitch
maneuver performed the best, with a mean altitude of 29.2 ft and a standard deviation of
of simulators is that they allow pilots to experience dangerous wind shears in a safe
environment, ideally with the knowledge and skills transferring to a real world
environment. Simulated microburst encounters can help crews coordinate their escape
efforts in critical situations (Trevino & Laituri, 1989, p. 6). As a research tool, simulators
provide a unique insight by introducing the human element, which often points to the
Ill
are unique in evaluating conditions of flight and their utilization for such endeavors is
appropriate.
case be considered (p. 10). The mathematical models and computer simulations suggest
that a more favorable strategy than the constant pitch maneuver exists for escaping from a
microburst encounter. "Static analyses are not necessarily conservative, nor do they
reveal the full potential for successful wind-shear penetration" (Stengel, 1984, p. 200). A
simulation of the variable pitch guidance strategy may therefore be beneficial. "Enough
proposing of, analysis of, and simulation of microburst encounter guidance strategies has
been carried out. The time has come to test these strategies in a real aircraft or, at least, in
factor of safety than the constant pitch guidance maneuver through the same microburst
as examined in a flight simulator of a large jet transport category aircraft. In this context,
RESEARCH METHODOLOGY
Evaluating whether the variable pitch guidance maneuver exhibits a greater factor
of safety than the constant pitch guidance maneuver was determined, in this study,
through statistical significance of the difference in recovery altitude between these two
escape maneuvers. Qualified pilots flying an FAA approved airplane simulator performed
the maneuvers. The test runs involved a simulated flight along an ILS approach path
through a microburst wind shear. Recovery altitude was recorded by the simulator
computer and was the primary measure of safety. The data were evaluated by an analysis
maneuver.
Simulation was the key method in this study, as the aerodynamic factors that are
involved in microburst encounters are confounding and numerous; while the safety
Design
This comparative study examined the difference in recovery altitude between the
constant pitch maneuver and the variable pitch maneuver. Recovery altitude was defined
as the lowest altitude, above ground level, recorded by the simulator computer, of the
aircraft during the microburst escape maneuver. The escape maneuver was that flight
phase from initiation at 800 ft AGL until the aircraft's airspeed stabilized beyond VREF +
10 out of the microburst environment, or until the completion of data recording for that
run, whichever occurred first. In practice, the escape maneuvers were completed prior to
study (1989) of microburst penetration in the takeoff phase. Hinton found the standard
deviation of the pitch-hold strategy to be 58.4 feet, while the flight-path angle strategy
(variable pitch maneuver) produced a standard deviation of 58.3 feet (p. 28). Assuming
the difference between the means of the maneuvers is significant at one standard
deviation, and specifying an alpha value of .05 and a beta value of .20, the group sample
size is estimated, by the power of F Test, to be 17 runs per maneuver (Neter, Wasserman,
& Kutner, 1990, p. 1151). Any difference larger than one standard deviation would
require fewer trials. The estimation necessitates the sample size be equal across groups.
group size, and 16 trials per maneuver for a total of 32 trial runs were planned for the
experiment. One pilot subject dismissed himself prior to data collection, and so 15 runs
were substituted for each maneuver, 5 per individual to maintain symmetry in the cell
frequency. With 15 trials per maneuver, the minimum difference between means to
generate a statistically significant value at the .05 level increased by less than 15 feet
(Neter, Wasserman, & Kutner, 1990, p. 1151). This compromise was deemed acceptable
for the experiment; though more trial runs were desirable they were not thought
The approaches flown for this study amounted to 42. Of these, 4 were flights
through the microburst wind shear terminating with a landing on the runway, 15 were
constant pitch maneuvers through the microburst wind shear, 3 were training maneuvers,
and the remaining 20 were variable pitch maneuvers through the same microburst as the
approach through the wind shear to a landing. This enabled the subjects to familiarize
themselves with the effect of the microburst on the aircraft performance and response.
Each subject flew one of these approaches prior to the experimental procedures for that
day, with the result that the pilot who flew two days also performed two landings in the
shear. The other maneuvers that were not statistically evaluated were the training
maneuvers. The practice maneuvers were supervised, but no data recording occurred.
The constant pitch maneuver was performed a total of 15 times, though symmetry
was not attainable due to time constraints. The difference was 1 trial per pilot subject.
The variable pitch maneuver was performed a total of 20 times with the breakdown of 5,
6, and 9 trials per subject. Statistical procedures accounted for the discrepancies in cell
Generally, the smaller the difference to be determined, the larger the sample size
required. The difference in means between the two maneuvers studied was assumed to be
similar to the study conducted by Hinton (1989). In actuality, the difference in means was
much greater, so the sample size, though less than desired, was greater than required. In
the context of the type and scope of the experiment, the statistics generated are robust to
The statistics employed to evaluate the data were selected to reduce the error
maneuvers was determined through an ANOVA with significance predefined at the .05
level.
115
The escape maneuvers and pilots were the independent variables; trajectories
from the escape maneuver produced specific values of the dependent variable: altitude.
An ancillary dependent variable, airspeed, was also recorded for regression analysis with
altitude. The lower limit of altitude was considered ground contact, while the upper limit
was not confined. The lower limit of airspeed was the stall speed for the particular
aircraft configuration, and no upper limit was imposed. In all cases the simulator
controlled the lower limit of altitude, and generated a stall for the lower limit of airspeed.
It appears that sustained ground contact, that is, the lower limit of altitude, did not occur
Internal validity was controlled using the same flight simulator with the same
wind data as each previous run. Pilot subjects were also compared within the group for
by recovery altitude.
External validity was enhanced with the use of an independently certified flight
simulator and a realistic microburst model derived from the analysis of the Delta 191
microburst accident.
Reliability was evaluated through the examination of the data. The variance of
recovery altitude for each pilot subject was compared, through a test of significance,
within the group. No significant difference between pilot subjects in recovery altitude
strengthens the assumption that the variance in the data was due to the escape maneuver
and not the individual. The demographics of the pilot group increases the reliability, as
each pilot subject had different training and experience levels, yet performed similarly in
the maneuvers; indicating that the data were not dependent on time nor place.
116
Participants
Airline transport pilots, type rated on equipment, and with operational airline
experience, performed the maneuvers from the left hand seat. Support pilots occupied the
right hand seat and performed PNF duties. Support pilots held the same qualifications as
the left seat participant for the type aircraft used. Crew coordination was observed to
The three pilot subjects were drawn from qualified instructors, employed by
Flight Training International, who were available during the slotted simulator times. They
were paid their normal contractual rate for the duration of the flight, whether acting as PF
or PNF. In the interest of confidentiality the pilots are represented by number, identifying
The simulator engineering manager for simulator 737 # 4 at United Airlines set
the required simulator parameters, repositioned the flight, initiated the computer sub-
occurred during data collection. After each approach, the aircraft was repositioned and
Experimental Device
The use of simulators to explore aircraft performance in wind shear has been well
and appropriate corrective action (e.g. NTSB, 1976, p.18; NTSB, 1977, p. 15). The Royal
Aeronautical Society (1995) contends "the fidelity of windshear [sic] modelling [sic] has
developed significantly in the last few years such that simulators are very effective tools
117
for training flight crews in the techniques necessary to combat these phenomena" (f
2H.2). In addition, the FAA mandates flight simulator training of wind shear encounters
under part 121.409 for air carriers with turbine-powered airplanes (FAA, 2003b). With
endorsements from the NTSB, the Royal Aeronautical Society, and the FAA, it is
assumed that a simulator is the appropriate device for exploring aircraft performance in a
microburst environment.
Denver, Colorado was used for this investigation. Replicating a Boeing 737-291 with
JT8D-17 engines, simulator 737 # 4 was specifically qualified for wind shear training by
the National Simulator Program Manager (NSPM). The simulator was re-certified 15
degree-of-freedom system, was engaged during all flights. The simulator was fully
The FAA classifies the aircraft chosen, a Boeing 737-200, as a large jet transport
category aircraft. This model, powered by the Pratt and Whitney JT8D-17 turbofan
engine, produces 32,000 foot-pounds of thrust at sea level. The maximum take off weight
The wing loading is 119.38 pounds per square inch at take off, and the thrust to weight
ratio is 0.27. The aircraft is a typical twin-engine short-range airliner with conventional
handling characteristics. Figure 6 illustrates the dimensions and three view drawing of the
Boeing 737-200.
118
Figure 6. Boeing 737-200 aircraft dimensions and views. Note. From The Boeing 737 Technical Site, by C. Brady, 2003,
http://www.b737.org.uk/dimensions_737200.gif.
119
Flight instrumentation on the 737-200 is primarily electromechanical. To enhance
situational awareness of the flight crew, an Allied Signal Mark 5 Enhanced Ground
Proximity Warning System (EGPWS) has been incorporated into the avionics. This
computer-based system with geographical database warns of controlled flight into terrain
(CFIT) and low-level wind shear conditions. The reactive wind shear system includes
warning occurs when the computer senses that a difference between the aerodynamic
Reactive systems place the aircraft within the microburst before a warning is
issued to the flight crew. In this study, the warning occurred slightly after the PF initiated
the microburst recovery, and well after the effects of the microburst were apparent to a
well-trained crew.
To meet the requirement of wind shear and microburst training, United Airlines
has developed models representative of known accident scenarios. These wind shear
models "must be supported or properly referenced in the ATG [Approval Test Guide]"
2, 1985. Colloquially this has become the Delta microburst, in reference to the
accompanying accident, Delta 191. The winds derived from analysis were representative
of a severe microburst as defined by the F factor. The analysis of the accident (Fujita,
1986) indicates a 27-knot headwind changed to a 40-knot tailwind in about one mile, and
the maximum downdraft was about 2880 feet per minute (28.4 knots). The aircraft
transitioned the downburst and then entered a roll vortex. This changed the vertical
120
component of wind to include both up and down drafts. A crosswind changing direction
It was calculated that the center of the parent microburst, which spawned these
winds, was located 1000 feet to the left hand side of the aircraft, and 12,000 feet before
the runway threshold (NTSB, 1986, p. 59). The outflow from the thunderstorm was about
11,000 feet (3.4 km) in horizontal diameter and was assumed symmetrical. The NTSB
(1986) concluded that based on the outflow diameter, the winds met the criteria of a
microburst (p.59). The L-1011 passed close to the center of the microburst, and was in
The Delta microburst is modeled for use in the simulators at United Airlines per
the ATG and is labeled UAL-7, severe on approach. Table 3 represents the computer
wind plots for UAL-7, the microburst used in this research study. The microburst is
initialized when the aircraft descends through 1,200 feet AGL, and the wind effects
continue for 21,000 feet horizontally thereafter. The wind, as modeled, is independent of
altitude; each approach will experience the same wind, regardless of vertical
Only the UAL-7 severe on approach microburst model was used in this
experiment, as was only one simulator, 737 # 4, thus providing internal consistency in the
experiment.
for vertical wind corresponds to a downdraft, and a negative crosswind value is indicative
of a wind from the left, while a positive value is from the right. Headwind and crosswind
0 0 0 0
1000 -4 0 -10
6000 8 960 1
8000 27 -1080 5
8300 22 -840 5
9200 12 -540 5
10,000 1 -2100 5
11,200 18 -840 15
11,700 0 -2880 20
12,600 -6 1320 -5
17,000 -40 0 0
21,000 0 0 0
The test instruments included two separate maneuvers: the constant pitch
maneuver and the variable pitch maneuver. The constant pitch maneuver represents the
current authorized procedure, while the variable pitch maneuver has shown promise in
computer simulations (see Bray, 1986; Dogan & Kabamba, 2000; Hinton, 1988, 1989;
Melvin, 1986; Miele, et al., 1987; Mulgund & Stengel, 1992a, 1992b; Psiaki & Stengel,
The constant pitch maneuver, published by the FAA (1988) in Advisory Circular
00-54, Pilot Windshear Guide, delineates the approval of the maneuver for specific
aircraft. Boeing's model 737 is in this list of approved aircraft (p. ii). The constant pitch
maneuver dictates the thrust being set to go-around EPR, the pitch attitude positioned to
15° at a rate of 37 second, while respecting stick shaker, and configuration maintained. If
the stick shaker does not activate and the aircraft is descending, the pitch attitude is
increased in 2° increments until either the descent is arrested or intermittent stick shaker
activation. If at 15° pitch attitude the stick shaker is operating, the pitch is decreased until
intermittent activation. Intermittent stick shaker activation is always considered the upper
limit of pitch. To maintain consistency in the data, go-around thrust was set at 2.00 EPR.
The variable pitch escape maneuver, described by Dogan and Kabamba (2000)
and Miele et al. (1987), was adapted for this study. Figure 7 is a graphical representation
of the variable pitch maneuver as briefed to the flight crews. Optimization was not
target altitude, which is maintained until energy is available for initiating a climb.
123
Variable Pitch Escape Maneuver
1 Dive Phase
Microburst recognized
Thrust set to maximum
Pitch attitude decreased to 0°
ROD 2,000-2,500 fpm
2. Level Phase
Altitude Maintained
Thrust Maintained
Respect stick shaker
3. Climb Phase
Airspeed VREF +10
Microburst end
Normal climbout
Figure 7. Variable pitch escape maneuver training aid. Note. Microburst and aircraft
adopted from Influence of wind shear on the aerodynamic characteristics of airplanes (p.
17), by D. D. Vicroy, 1988, NASA technical paper 2827, Washington D.C.: NASA.
The dive flight path angle used by Dogan and Kabamba (2000) was employed for
this study. This differed somewhat from the optimized dive presented in Miele et al.
(1987), which was dependent on a feed back loop. Maintaining an optimized dive would
require reprogramming the flight director, and was deemed out of scope for this research.
The simplified dive guidance was used instead. At the first indication of a microburst, the
pitch angle (0) was decreased to 0°, generating a flight path angle (y) of about -7°. The
pitch attitude change was accomplished at a rate of 37 second. The 0° pitch attitude was
The target altitude is a function of the initial altitude (Equation 6), being 40% of
the initial altitude rounded up 100 feet and bound by the lower limit of 200 feet above
ground level. The initial altitude was set at 800 feet AGL and was only several seconds
prior to the point corresponding to wind shear annunciation by the EGPWS, Mark V.
For ease in flight technique, the target altitude was rounded to the nearest 100-
foot level. The target altitude used was thus 400 feet AGL, not 420 feet as computed
through Equation 6. Target altitude was maintained through pitch control. Similar to the
constant pitch maneuver, the upper limit of pitch was the activation of intermittent stick-
shaker. The flight path trajectory changed to the climb phase as the airspeed stabilized
past VREF + 10, with the aircraft out of the microburst environment.
The climb phase was left to the discretion of the PF; some chose to fly at high
speed and low level while others performed high-speed pitch up maneuvers. No data
were analyzed in this phase, as the microburst effects were no longer present and the
through 1,400 feet AGL. This was a limit imposed by the test equipment, in practice,
however, it did allow enough time for all escape trials to exit the microburst wind shear
with the aircraft in a stabilized climb for all test maneuvers. The recording window
opened when the aircraft initially descended through 1,400 feet AGL and continued
thereafter for 100 seconds, after which time the simulator froze in data and motion
output; the simulator was then reset to the initial conditions for the next trial.
Data
The subject of examination was the safety, in terms of altitude, of the variable
pitch escape guidance trajectory. Comparison of the safety of the individual escape
maneuvers was achieved by noting the minimum altitude of each maneuver while
Data are reported in the form of English Standard Units. Altitude is measured
from the center of gravity (CG) location of the airframe and represented in feet and
Real time data recorded by the simulator system computer were attained. A sub-
routine was written to capture the altitude data at a rate of 5Hz during the recording
window of 100 seconds. This program was then tested and operated by the simulator
engineer during the experiment trials. Simulator 737 # 4 is run from a VAX computer
using a VMS language. Graphical data output to the printer included height above
ground, indicated airspeed, average EPR, stick force in pounds, body attitude in degrees,
proceeding to the next phase, a paper form was filled out (Figure 6) for each pilot flying.
approach parameters were noted to confirm consistency in settings for each pilot subject.
A running tally was kept of each trial phase and coordinated with the simulator plots,
The simulators at UAL TK are equipped with video cameras installed for contract
crews to use if part of their training curriculum. Simulator training is not currently
videotaped at UAL, per agreement with the pilot's union. As this was an independent
study, the cameras were run to facilitate and backup record keeping. After data reduction,
any and all individual or identifying characteristics were removed from the data sets.
Procedures
microburst, and the data collection devices. They were also accordingly debriefed and
The simulator was set for a landing gross weight of 90,000 pounds and a
configuration with gear extended and flaps set at 15 for all maneuvers. A positioned on
the extended centerline, 9nm from the runway threshold above 2000 ft AGL, served as
the initiation point. This position closely corresponds to the international recommended
practice in validating simulators for wind shear training (ref: Royal Aeronautical Society,
1996, p. 101) and is similar to the FAA Piloted Flight Simulator Study of Low-level
Sim Data
Date Sim# Type Engines
Configuration
Weight CG Flap Gear
Performance
Vref EPR limit Ft-lbs Thrust Overboost
Environment
Shear # Surface Wind Temperature Pressure (Hg)
Navigation
ILS Altitude at OM DME of OM DH
Crew
Pilot Flying Pilot Not Flying Sim Admin
Maneuver Tally
Landing CPM TVPM VPM
Notes
maneuverability throughout the speed range anticipated. At flap 15, the flaps extend
mainly aft and only slightly down, increasing drag very little (Boeing, 1985, p. 04.60.02).
Flap 15 is the setting used in a go-around maneuver until the aircraft accelerates to flap
speed schedule. In practice, the flaps remained at 15 and a flap overspeed condition was
not penalized. The airspeed never exceeded 10% of the upper limit of flap speed.
Reference airspeed for this configuration and weight was computed to be 132
kias. Approaches are traditionally flown at VREF + 5 knots, and the target approach speed
was set at 137 kias for every trial run. Each approach was hand flown, and it was the
runway 26 at Denver International Airport (KDEN). Each approach was initiated 9 miles
from the runway threshold on the ILS course (Figure 9). The aircraft was in landing
configuration and trimmed for 161 kias with all checklists complete. The approach began
when the pilot subject advised ready. Normal airline procedures and callouts were given
by the PNF during approach. Additionally, two non-standard callouts were annunciated
by the PNF; an 800 foot call to advise the PF to initiate the selected maneuver, and a 400
foot call (target altitude) when the variable pitch escape maneuver was flown.
of 49.9 statute miles, and surface wind 190° at 8 knots. Turbulence was set to zero for all
approaches, as this was considered noise introduced into the system and would further
\
- 4 - * 3 o-' MM HIStt—*
\ I & H 4 7 DVV|
D14.2 UOY
6236
Gnc sp&ed Krs 70 ' 9C ' rcc r_>£ ' M0 i ?<>2 _____
3 CG*' 376] 486, 540 648, 756 8o4]
rA?!~ 15900' j 8 0 0 0 ' ; 340°
GPA$P to MAP 5 6 FTla"" 3 4 4 ' 3 2212 4-gj 2 2 i 2 06 J t I !tr r>rfg
STRAIGHT IN LANDiNG RW> 26 CIPCIE TO LAND
ILS OtOC{GSouf)
DA* S$Q&'{200', MDA» $7W*$4
fUU JT5I ar Cl ou-t?8AlUfAtSatft Rail ml AJLS oat
*^50sr1 1K2
Q N o * authorIze-d d y m g s^mu^faii^cws operas on^>
CHANGES Cornrtvnica* WWi*-*-**--
Figure 9. ILS runway 26 Denver International Airport. Afote. From United States Airway
Manual by Jeppesen Sanderson, Inc., 1998, Englewood, CO: Jeppesen Sanderson.
During the escape maneuver phase, the microburst model was positioned at a
constant point along the approach path. Data acquisition began as the aircraft descended
through 1,400 feet AGL, and the microburst winds began as the aircraft descended
through 1,200 feet AGL. The pilot subjects were instructed to continue flying as normal
an approach as possible until reaching 800 feet AGL, at which point they performed
either the constant pitch escape maneuver or the variable pitch escape maneuver.
The constant pitch escape maneuver was familiar to all the pilot subjects and was
the first sequence in the trial runs. Upon reaching 800 feet AGL the pilot subjects
commanded go-around thrust and pitched initially for 15° on the attitude deviation
indicator (ADI). Stick shaker was respected with pitch and the aircraft continued on its
trajectory until the recording window closed, 100 seconds after opening, and the
simulator stopped.
The next maneuver in sequence was the variable pitch escape training maneuver.
This was conducted in the absence of the microburst to demonstrate the procedure and to
train the participating pilots. The aircraft was pitched to 0 = 0° from 800 to 400ft AGL, at
which point altitude was maintained. A reduction in thrust mimicked the variable pitch
nature of holding altitude with decreasing airspeed. At stick shaker speed (Vss) the thrust
was returned and the aircraft allowed to climb. After the demonstration and a practice
After the training exercise, the pilot subjects flew the variable pitch escape
maneuver through the same microburst as used during the constant pitch escape
maneuver. At 1,200 feet AGL the microburst wind shear began, and at 800 feet AGL the
pilots commanded go-around thrust while decreasing pitch to 0° on the ADI, accelerating
131
to the target altitude of 400 feet AGL. The pilot subjects then attempted to maintain this
altitude with pitch control. The wind shear ended 21,000 feet after it began, and the
Thrust was limited to 2.00 EPR immediately after recognition of the microburst.
This setting, 10 % less then the normal take off go-around (TOGA) EPR for the JT8D-17
engine at that pressure and temperature, was used to enhance the effects of the microburst
and limit the aircraft from powering out at a relatively high altitude prior to data
collection. In previous trials it was found that the aircraft was not entering the microburst
After each run, the simulator was positioned back to the initiation point; the pilot
subjects prepared for the next run on the flight deck, and the simulator supervisor set the
required parameters at the simulator console and computer. The flying was segmented to
minimize fatigue and the crossover of maneuvers. Each pilot subject flew about 5
maneuvers before trading position with the accompanying crewmember. The variable and
constant pitch maneuvers were separated into these 5 run sequences when possible, so
that constant pitch trials for the day were completed prior to the introduction of the
The total escape maneuvers flown by the pilot subjects during the trial phase of
the study amounted to 35. The constant pitch maneuver was performed 15 times, while
the variable pitch maneuver accounted for the remaining 20 trials. Further time was not
available to add to the constant pitch maneuver tally. The disparity in cell frequency
shear to a landing on the runway, and 3 practice variable pitch maneuvers were
performed, for a total of 42 approaches in the simulator during two days of trials. The
approaches through the microburst with a landing on the runway provided the pilot
subject an opportunity to feel the effects of the microburst on the handling qualities and
performance of the aircraft. The practice maneuvers were necessary because this was a
new technique for the pilots, and their understanding of the procedure was important to
perform the maneuver. These 7 approaches were not statistically evaluated, and the 3
The minimum number of approaches planned per maneuver was 15. This was
determined as a compromise from data supplied in Hinton 7s study (1989) and the desire
to retain symmetry of data. Increasing the trial runs to 18 was considered untenable in the
time allotted. The opportunity to perform additional maneuvers was deemed more
important than cell frequency, this accounts for the disparity in trial runs between the
Altitude data, recorded by the simulator computer at a rate of 5Hz, generated 500
individual readouts for each run. The individual readouts were extracted from the host
computer via a subroutine and deposited into files. The files for each run were labeled
consecutively and transferred to the mainframe computer at United TK, they were then
relocated to a server and emailed to the researcher for reduction and statistical analysis.
Data were also captured on graphical output (Figures 15 through 53). The graphs were
printed from the simulator computer at the conclusion of the day's trials and include
recovery altitude between maneuvers answers the hypothesis of which escape maneuver
provides for greater safety. A difference in recovery altitude will also determine the
homogeneity of the sample group within the escape maneuver. Two techniques in
variables.
performed by a 3x2 analysis of variance (ANOVA) with pilot subject and maneuver type
The difference in recovery altitude between pilots for the same maneuver is
evaluated via a single factor ANOVA, with pilots as factor and altitude as criterion. This
analysis is performed to determine any outliers in data, and validate that a difference in
recovery altitude is due to the escape maneuver rather than the individual.
role in the safety of flight for conventional aircraft are the altitude and airspeed. It was of
interest to determine if a relationship between these factors existed for the individual
escape maneuvers. The relationship between altitude and airspeed is examined with a
The data for recovery altitude and airspeed for each maneuver are displayed in
numerical form in Chapter IV, the graphs of the trial runs are provided in Appendix A
RESULTS
The data were collected on two separate days and involved simulated flights of
two hours duration each day. In the time apportioned, three pilots flew a total of 42
landing on the runway and 3 variable pitch training maneuvers with no shear present, did
not contribute to the data output. These maneuvers were provided for training purposes.
The approaches providing data for analysis were 15 constant pitch maneuvers through the
microburst wind shear and 20 variable pitch maneuvers through the same wind shear. The
Data acquisition began as the flight descended through 1,400 ft AGL and
continued for 100 seconds. All trial runs were completed within the data time frame. The
microburst, UAL-7 "severe on approach" (Table 3), began when the aircraft initially
descended through 1,200 feet AGL and it continued horizontally for 21,000 feet
thereafter. The pilots proceeded into the microburst, on approach, until 800 feet AGL, as
determined by the barometric altimeter. At this point, the PNF called "800 feet" and the
PF performed either the constant pitch maneuver, or the variable pitch maneuver, as
Altitude data were collected at 5Hz from the simulator host computer via a
subroutine providing numerical output to seven decimal places. Graphical data were
which are provided in the Appendices as Figures 15 through 53. The recovery altitude
135
derived from the computer subroutine, and the airspeed interpreted from the graphical
output, is provided in Table 4 for each pilot subject, maneuver, and trial run.
Prior to the escape maneuvers, each pilot flew an approach through the microburst
wind shear to a landing on the runway to familiarize themselves with the handling and
performance of the aircraft in a microburst wind shear. The pilot subjects then flew the
constant pitch maneuver through the shear; each run increasing the individual's tally for
the particular maneuver. The pilot subjects then profiled a training maneuver, with no
microburst wind shear present. Data for the training runs were not recorded. The subjects
finally performed the variable pitch maneuver through the microburst wind shear. At the
end of day two, time remained for additional approaches, and though this caused a larger
The data in the appendices are presented with landing approaches in Appendix A,
followed by Appendix B with the constant pitch maneuver (Figures 19 through 33), and
Appendix C housing the variable pitch maneuver data (Figures 34 through 53). The data
number after the data collection to provide a level of confidentiality to the individual. The
maneuvers flown by the pilots are only those that represent the experimental procedures,
the maneuver column does not include any training maneuvers or familiarization flights.
The run number is reported in sequence by maneuver for the individual. Altitude was
recorded in feet AGL to seven decimal places, but in this table is given to a hundredth of
a foot. The next column, airspeed, is reported to the whole number only. The last column,
Figure #, indicates the graph in the appendix that corresponds to the data presented.
Table 4
Altitude and Airspeed of Simulator Trials for Constant and Variable Pitch Maneuvers
lowest altitude, above ground level, recorded by the simulator computer, of the aircraft
during the microburst escape maneuver. The descriptive statistics in Table 5 are
delineated by pilot subject and maneuver, while the totals are provided in the text of the
discussion section. The reported values are given to four decimal places, but were
computed to seven, the last digit being rounded per conventional accounting methods.
The mean (M), standard deviation (SD), and total number of trials (_V) are provided for
statistics, ANOVA, regression analysis, and post hoc procedures follow in Tables 6
through 11. A discussion of the descriptive statistics, along with the measures of central
Table 5
Pilot Maneuver M SD N
ANOVA was employed. The ANOVA compares group means. The independent
variables, or factors, were the pilots and maneuver type: With three pilot subjects and two
different escape maneuvers a 3x2 ANOVA was formed. The data indicate no significant
difference between pilot subjects and no interaction effect between maneuver and pilot. A
significant difference does occur between maneuvers. The probability that the null
Table 6 reports the source of the inferential statistic followed by the degrees of
freedom (df) and the F-factor (F). Mean square error is reported in parenthesis at the
Table 6
Source df F
Between subjects
Maneuver 1 104.753*
Pilots 2 1.467
Error 29 (7088.258)
*p < .001
139
Determining a relationship between the lowest airspeed encountered, during the
constant pitch microburst escape maneuver, and the recovery altitude was examined
through the use of a regression analysis, Table 7. All 15 constant pitch runs were
examined and no significant relationship was observed. The variable is presented first,
followed by the regression coefficient (_9), the standard error of the regression coefficient
(SE B), and then the beta value (p). The multiple correlation (R 2 ) is given at the bottom
of the table.
Table 7
Variable B SEB J3
Similar to the constant pitch maneuver, a regression analysis was performed for
the variable pitch maneuver. Five more trials were performed in the variable pitch
maneuver, bringing the total runs to 20. Table 8, the regression analysis of the variable
pitch maneuver is consistent in layout to Table 7, the regression analysis of the constant
pitch maneuver. The data in Table 8 indicate that a relationship does exist between the
lowest airspeed and recovery altitude, and it is significant at the p < .05 level. The
multiple correlation, here 19.9, is the percent of variance in the dependent variable
Variable B SE B p
Note. R2 = . 199
*p < .05
Evaluating the difference in recovery altitude between pilots for each maneuver
was accomplished via a one-factor ANOVA. For the constant pitch maneuver, Table 9,
the factor, or independent variable is the pilot subject. Three pilot subjects participated,
so the degrees of freedom (df) is two, while the F-factor (F) is reported as 0.513. This
value does not meet the level of significance established, and therefore the null
subjects for the constant pitch maneuver. The error is reported in the lowest row with
Table 9
Analysis of Variance for Altitude in Constant Pitch Maneuver with Pilots as Factor
Source df F
Pilots 2 0.513
Error 12 (11999.282)
reported as a one-factor ANOVA in Table 10. The degrees of freedom (df) is two for the
pilot group and the F-factor is significant at the .05 level. There was a significant
difference in recovery altitude for one or all the pilots in the variable pitch maneuver.
Error is reported in the last row with mean square error in parenthesis.
Table 10
Analysis of Variance for Altitude in Variable Pitch Maneuver with Pilots as Factor
Source df F
Pilots 2 5.436*
Error 17 (3621.653)
*p < .05.
Determining where the difference lies within the pilot subject group was
minimize Type I errors and not likely to reject the null hypothesis. The values, as
provided in Table 11, indicate a significant difference in recovery altitude exists only
between pilot subject 2 and pilot subject 3 at the p < .05 level. No other statistical
In Table 11, the pilot subject to which comparison is made is presented first (/),
followed by those subjects to which compared (/'). The mean (M) is provided along with
Post Hoc Comparison between Pilots for Altitude in Variable Pitch Maneuver
*p < .05.
The data collected represent graphical and numeric output from the simulator
computer. Though provided to seven decimal places, the recovery altitude is only
reported to four decimal places in the statistical analysis, and is discussed to two decimal
places in the text. This level of precision represents about an eighth of an inch in altitude
for an aircraft that is 100 feet long. The altitude was determined from the CG location of
Airspeed was derived from the original graphical plots and is not accurate below a
one-knot distinction. As an ancillary value, this provides adequate accuracy for the
calculations imposed. Stall speed was generated by the computer, and a stalled condition
is provided in the graphical plots of angle of attack versus time. The reported value of
stall in the discussion section, Chapter V, was derived from graphs furnished by Boeing.
143
CHAPTER V
DISCUSSION
In every approach flown through the microburst, the effects of the wind field on
the aircraft are apparent. A characteristic trace of the altitude plot is the portrayal of a
4
W' starting around the 50-second time hash, a result of the vortex modeled in the
microburst. The wind had an affect on the altitude, the airspeed, and the handling of the
approach, a constant pitch strategy, and a variable pitch strategy. With the landing
approach, each pilot was in a position to put the aircraft in the touchdown zone at the
conclusion of the run, and confidence was gained that the shear was navigable. This
introductory run provided the pilot subject a point of reference to compare with non-wind
The constant pitch escape maneuver, familiar to all the pilot subjects, followed the
landing approaches. The pilot subjects were instructed to fly as normal an approach as
possible until the commencement of the escape maneuver. At 800 feet AGL the
maneuver was initiated-the aircraft was about a third of the way into the microburst at
this point, with a headwind of 23.5 knots and a downdraft of 705 fpm.
Attempting to maintain glide slope to 800 feet AGL, the pilot subjects had the
thrust levers at idle and used pitch for flight path control. When the aircraft arrived at 800
feet AGL the engines were spooled up to 2.00 EPR, concurrent with an increase in pitch
attitude to 15°, or beyond as needed. The time required between idle power and go-
144
around power was about six seconds. A review of the simulator plots in Appendix B
shows the response in pitch attitude slightly lagging EPR. With thrust applied, and a
positive pitch, the aircraft continued to descend below the initiation point.
The lowest altitude attained in the constant pitch maneuver was generally in the
second valley of the aforementioned W. Just 2 of the 15 trials exhibited the lowest
altitude in the first valley (Figure 20 & 24). For the majority of trials, the low altitude
point follows the highest peak in angle of attack and the lowest level in kinetic energy.
This second valley in the altitude trace is initiated by a downburst of 2,520 feet per
minute occurring 14,800 feet beyond the start point of the microburst (Table 3). Though
this second downburst is not as great as the preceding downburst, the aircraft is in a lower
As the aircraft reaches it's low point, and begins climbing, it is aided by a vertical
wind change from a downburst to an updraft of 1, 080 feet per minute. Combined with
the thrust and positive pitch of the aircraft, this updraft allows for climb rates exceeding
Descriptive Statistics
Mean recovery altitude attained for the constant pitch maneuver was 479.52 feet
with a standard deviation of 105.66 feet. The highest recovery altitude for the constant
pitch maneuver was 633.96 feet, while the lowest altitude was 264.48 feet. No outliers or
extreme values were observed in the altitude data. Individual mean recovery altitudes and
standard deviations are provided in Table 5. The data do not contain any zero values, as
pitch maneuver, shows a similarity among groups. The heavy line in the box is the
median altitude attained by that pilot, while the box itself represents the interquartile
range, that is from the 25th percentile at the bottom of the box to the 75th percentile at the
top of the box. Whiskers protruding from the bottom and top of the box are the observed
700'
600-
1 I
CD 5 0 0 ' I 1 1 ——"—•~ — —
<.
J
a> ' 1
3
CD
| 400- j ^ ^ I
< I I
300«
200,
1 2 3
Pilot
Figure 10. Boxplot of pilot versus altitude for the constant pitch maneuver.
In the graph, the boxes are roughly located at similar altitudes for all pilot
subjects, indicating little difference among groups. The median line shows skewness in
the distribution of altitude for pilot subjects 1 and 3. The difference of the median from
the mean for pilot 1 is 29.50, for pilot 2 the difference is 1.46, and for pilot 3 the
146
difference is -30.54. When the data from the pilots were combined, the constant pitch
maneuver had a median of 498.55. When compared with the mean altitude of 479.52, the
An investigation into the skewness and kurtosis shows that the data do not violate
the assumption of normality. For this test, the skewness (-0.390) was divided by the
standard error (0.580) to give a value (-0.672) within an acceptable range (-2 < -0.672 <
2). The heuristic for kurtosis is similar, and the computed values for the Fisher kurtosis
were -0.338 with a standard error of 1.121, giving a value of-0.302 which falls within the
Analysis of Variance
moderate departures from normality occur. Equal variance of the dependent variable
across the independent variables is an additional postulate of the ANOVA. In this study,
homogeneity of variance was established through Levene's test, which clearly established
(p = .251) that the error variance of the dependent variable was equal across groups.
Scheduling restrictions dictated the trial runs of the participating pilots, causing an
unequal sample size. In performing the ANOVA, this unequal sample size was relegated
by a Type III sum of squares. As there were no missing cells in the data, merely unequal
trials, the Type III sum of squares provided a linearly unbiased estimate of the marginal
in the dependent variable recovery altitude when compared among groups for the
surmised that trading the greater amount of potential energy would relieve the burden on
was computed (Table 7) for airspeed predicting altitude in the constant pitch maneuver. A
scatter plot (Figure 11) of the 15 datum points for recovery altitude versus minimum
airspeed in the constant pitch maneuver shows no apparent relationship. The nearly
horizontal line midway in the graph is the plot of the regression equation.
700 •[
• •
600 J
I •
$ 500 | . " ,
^* •
CD
| 400 4
< 1 •
"
300 4
•
200 J B u t § ( (
98 100 102 104 106 108 110 112
Figure 11. Regression plot of constant pitch maneuver: Airspeed predicting altitude.
The resulting equation from the analysis shows that altitude and airspeed do not
In Equation 8, altitude is in feet AGL and airspeed in knots indicated. A one-knot change
in airspeed will change the altitude by 0.46 feet. Though the slope is relatively flat, it is
recovery altitude induced by airspeed, in this analysis, R2 = 3.24 x 10"4. Basically, none
of the variance in the recovery altitude is attributed to airspeed. It is thus apparent that
recovery altitude and minimum airspeed are not intertwined in the constant pitch
maneuver.
The escape maneuver was initiated at 800 feet AGL, but the aircraft continued to
descend in the microburst to a mean altitude of 479.52 feet AGL. The pilot subjects did
not demonstrate a significant difference between themselves in recovery altitude for the
constant pitch maneuver. Though some of the recoveries occurred at a lower altitude, and
some at a higher altitude, there was no relationship between recovery altitude and
The variable pitch maneuver was a new concept to the pilot subjects, and some
were vocal in their skepticism. Nonetheless, the maneuvers were flown to the best of the
crew's ability. Similar to the constant pitch maneuver, the PF attempted to adhere to the
glideslope until 800 feet AGL, at which point the pitch was lowered to zero on the ADI,
and the aircraft accelerated to 400 feet AGL. Thrust was simultaneously increased to 2.00
Leveling at 400 feet AGL proved to be difficult, and the roll vortices in the
microburst are apparent in the altitude plots presented in Appendix C (Figures 34 through
53) for the variable pitch maneuver. Most of the low altitude conditions, 14 out of 20,
occurred in the first segment of the 4 W\ as opposed to the constant pitch maneuver,
which had low altitudes predominantly 13 out of 15 in the second segment of the 'W'.
Climb-out from the low altitude condition commenced when the energy level had
increased and the speed was acceptable to the PF. Some of the pilot subjects chose to
maintain the 400-foot level and let the airspeed build. The climb phase was 20-30
seconds after recovery altitude and did not affect the data.
Though the lowest altitude recorded occurred during the variable pitch maneuver,
there appeared to be greater altitude control overall, airspeed was generally higher, and
Descriptive Statistics
Altitude control was more definitive in the variable pitch maneuver, and pilot
standard deviations were generally less. The total standard deviation of recovery altitude
for the variable pitch maneuver was only 72.89 feet with a mean altitude of 184.73 feet.
The highest recovery altitude was 283.16 feet, while the lowest was 11.37 feet. This low
value does not constitute a statistic outlier; all the datum points were valid with no zero
entries occurring in the 20 trial runs. Individual mean recovery altitudes and standard
slightly staggered, suggesting there might be a difference among the groups. There is
overlapping between highest and lowest values for each pilot subject, but the medians do
show some disparity. The median for pilot 1 was 196.13, the median for pilot 2 was
132.48, and pilot 3 had a median of 243.76. The difference in medians between pilot 2
and pilot 3 was 111.28, while the difference in means was 118.77. When all pilot subjects
were combined, the total median for recovery altitude was 197.99 feet, while the mean
was a bit less at 184.73 feet. This indicates a negatively skewed distribution; where the
extreme scores are at the minimum altitudes, while most of the recoveries were at
400-j
300-
-j
o
<
£ 200-
CD _
<
100'
°1
1 2 3
Pilot
Figure 12. Boxplot of pilot versus altitude for the variable pitch maneuver.
151
There is negative skewness in the distribution of altitude for pilot subjects, but
overall the data do meet the assumption of normality. Skewness for the variable pitch
maneuver is -0.840, and when divided by the standard error of 0.512, gives a -1.641
value, which falls inside the normality guidelines of ±2. Testing of the kurtosis gives a
more conservative value. The Fisher kurtosis (0.394) divided by the standard error
(0.992) results in 0.387, close to the mid-point of the ±2 limit for normality. The positive
value of the kurtosis suggests that most of the recovery altitudes centered around each
Analysis of Variance
was examined through a one-factor ANOVA. The precondition of normality was met
through the skewness and kurtosis tests, while homogeneity of variance was computed
using Levene's test. The error variance of recovery altitude was similar across pilot
test of homogeneity of variance is robust when examining departures from normality, and
were ensured to provide a sound statistical basis. Accounting for unequal trials, a Type III
sum of squares was used, as it was in the constant pitch maneuver ANOVA, to estimate
The one-factor ANOVA for the variable pitch maneuver (Table 10) does show a
significant difference in recovery altitude between pilot subjects at the p < .05 level.
Post Hoc Analysis
A Tukey-Kramer post hoc analysis (Table 11) was performed after the one-factor
ANOVA for the variable pitch maneuver rejected the null hypothesis. The Tukey-Kramer
variance is a requirement, but unequal sample size is allowed and controlled using a
harmonic mean.
The difference in recovery altitude for the pilot subjects, as reported by the
ANOVA, occurred between pilot 2 and pilot 3. The mean difference was significant at
the p < .05 level. Other differences in recovery altitude between pilot subjects were not
One possible explanation for the disparity between pilot 2 and pilot 3 is the
amount of practice given to the participants. Prior to data collection, pilots 1 and 3 had
the opportunity of flying the simulator through the various maneuvers while data
collection anomalies were rectified. This increased training time was not afforded pilot 2,
who was scheduled for a later session. There were no significant differences in recovery
altitude for the constant pitch maneuver, which had been familiar to all the pilot subjects,
lending credence to the hypothesis that the amount of practice time increased recovery
difference, can be seen in Table 11 and Figure 12. The departure in means was 78.27 feet
between pilots 1 and 2, compared to the significant difference of 118.77 feet between
pilots 2 and 3. It is therefore plausible that the additional practice time was beneficial, to
The ethos of the variable pitch maneuver is the trade of potential energy for
kinetic energy to minimize the time the aircraft spends in the microburst. A lower
recovery altitude should translate into a higher minimum airspeed for the maneuver. A
scatter plot (Figure 13) of altitude versus airspeed depicts the spread of the datum points,
while the heavy line is a projection of the regression equation, showing the relationship
300 -
_ 200
CD
<
CD
CD
CD
•a
3
100 4
Figure 13. Regression plot of variable pitch maneuver: Airspeed predicting altitude.
The relationship between altitude and airspeed is not as strong as theory would
seem to indicate. Percent of variance in the recovery altitude, accounted for by airspeed,
was about 20% (R2 = .199). The P weight was significant at the p < .05 level and was
154
computed as -.446. In this analysis, p is the amount that the standard deviation of the
recovery altitude changes with a one-standard deviation change in airspeed. The standard
Data derived from Table 8 give the equation for airspeed predicting altitude.
Again, airspeed is in knots indicated, and altitude in feet AGL. A one-knot decrease in
airspeed represents an increase in recovery altitude of 6.48 feet. Using Equation 8, the
the upper limit of pitch was set at the stick shaker speed, this would also correspond to
the minimum usable airspeed. The stick shaker speed for the 737-200 advanced with flap
15, gear down, wings level, and at 90,000 pounds is 112 kias (Vs = 103 kias). At stick
shaker speed, the computed maximum recovery altitude is 206.72 feet AGL.
In practice, almost half of the trials, 9 out of 20, had recovery altitudes above the
computed maximum, while only 4 of 20 runs had indicated airspeeds below calculated
stick shaker speed. Equation 9 might be representative of a trend, but it is not an accurate
All trials in the variable pitch maneuver descended below the target altitude, of
400 feet AGL, as a consequence of the microburst wind. The mean altitude attained was
184.73 feet with a standard deviation of 72.89 feet. Within the pilot group, there was a
statistical difference in recovery altitude between pilot 2 and pilot 3. This was possibly a
The purpose of this study was to compare the safety of the current microburst
escape procedure with an alternate maneuver. Safety was quantified as the maximization
of the minimum altitude attained by the aircraft during the escape. Altitude data were
supplied by the simulator computer to seven decimal places at a rate of 5 Hz, the lowest
value from this output became the recovery altitude and provided the data for the
statistical analysis.
The constant pitch maneuver was performed 15 times during the data acquisition
phase. Pilot subject 4 dismissed himself prior to the start of the experiment, and so pilot 1
performed the trials slotted for number 4. This created unequal datum cells, though steps
have been employed in the statistics to mitigate the errors this may impose. In the
variable pitch maneuver, 20 trials were performed. Again, unequal cells surfaced, and
pilot 1 has three more trials than pilot 3, and four more trials than pilot 2; totals are
At 800 feet AGL the pilot subjects initiated the constant pitch maneuver;
regardless of the climb attitude, the aircraft continued to descend an average 320.48 feet.
The mean minimum airspeed in the recovery was 105 knots indicated-this is below stick
shaker speed, but above stall speed. Climb-out was at a high deck angle.
In the variable pitch maneuver, the pitch attitude was zeroed at 800 feet AGL, and
the aircraft dove to the target altitude of 400 feet AGL. It was intended that the target
altitude be maintained, but this was not possible. The microburst influenced the flight
path and the aircraft descended to an average altitude of 184.73 feet before recovering to
Figure 14. The standard deviation for the variable pitch maneuver was 72.89 feet, while
the constant pitch maneuver standard deviation was 105.66 feet. The closer confine of
altitude in the variable pitch maneuver might be a result of greater airspeed, which aids in
aircraft control. The minimum airspeed was higher in the variable pitch maneuver by an
average of 10 knots.
the combined boxplots (Figure 14). In some cases, the recovery altitudes overlapped for
maneuver type, but the interquartile range of recovery altitude for the variable pitch
700-
600
v 1
500
—i
- CD
1
5 400
<D
»^
% 300
<
2001
1001
°l
CPS VPS
Maneuver
Figure 14. Boxplot of combined altitude versus maneuver type.
3x2 Analysis of Variance
performed via a 3 x 2 ANOVA. The pilot subjects and maneuvers were the factors, and
recovery altitude the criterion. Homogeneity of variance was assumed with a non-
significant p value (.074) reported through Levene's test of equality of error variances.
in a two-way ANOVA the interaction effects are free from this influence. Mitigating the
problem of unequal sample size for the main effects was the use of a Type III sum of
squares, which provides a good linearly unbiased estimate of the marginal means.
interaction between pilot subject and maneuver is not significant. That is, the pilot
subjects do not modify the effect of the maneuver on recovery altitude. The main effect
for pilot subjects on recovery altitude is also not significant. Pilot subjects, by
Significant main effects for maneuver type were established in the 3 x 2 ANOVA.
There is a difference in recovery altitude for maneuver type at the p < .001 level. The
constant pitch maneuver had a total mean recovery altitude of 479.52 feet with a standard
deviation of 105.66 feet. The variable pitch maneuver generated a total mean recovery
altitude of 184.73 feet with a standard deviation of 72.89 feet. There was a little overlap
between the low recovery altitudes of the constant pitch maneuver and the high recovery
altitudes of the variable pitch maneuver, but this was not extensive enough to bring the
CONCLUSIONS
than the constant pitch maneuver through the same microburst. The research hypothesis
must therefore be rejected: The constant pitch maneuver, and not the variable pitch
maneuver, exhibits the greater factor of safety as determined through the maximization of
The mean altitude for the constant pitch maneuver was 79.52 feet higher than the
target altitude of 400 feet AGL adopted for the variable pitch maneuver. The target
altitude came from a previous study (Miele, Wang, Tzeng, & Melvin, 1987) and was not
increased during the course of the research. Hence, the variable pitch maneuver started at
an altitude lower than the mean recovery altitude of the constant pitch maneuver. Even
with the increased airspeed, the aircraft sank an average 215.27 feet below the target
altitude. To generate the same mean altitude, the variable pitch maneuver would require a
target altitude 100 feet below the initiation altitude, not enough of a height difference to
As might be expected, the variable pitch maneuver did have a higher average
minimum airspeed than the constant pitch maneuver. The difference of 10 knots, coupled
with the reduction of time spent at stick shaker angle of attack, probably contributed to
the better altitude control demonstrated in the variable pitch maneuver. However, this
control, and airspeed increase, was gained at the expense of recovery altitude, the
climb attitude, economized the recovery altitude and so outperformed the alternate escape
maneuver. Though there was less altitude control, greater altitude loss from initiation
altitude, and slower airspeeds, it must be concluded, by the original definition, that the
constant pitch maneuver demonstrated a greater factor of safety than the variable pitch
maneuver.
In some flight parameters the constant pitch maneuver may be less than ideal,
however, no crashes occurred in the trials, and a higher altitude was maintained. In regard
to airspeed, there was altitude to recover from a stall if necessary, and 6 of the trials were
successfully completed with minimum speeds below the one g stall speed of 103 kias.
Pilot subjects felt more comfortable with the prompt establishment of a climb
attitude, as in the constant pitch maneuver, rather than initiating a recovery by pitching
toward the ground, as in the variable pitch maneuver. As disclosed by the subjects, the
Airline pilots have trained to proficiency in the constant pitch maneuver and they
seem comfortable with the philosophy and performance of immediately initiating a climb
attitude. The WSTA established the curriculum and the escape maneuver, which is
endorsed by the major manufacturers and employed by the airlines of the United States.
There is, for the time being, minimal motivation to change maneuver strategy.
Avoidance remains the safest maneuver of all. Until a more robust strategy is
found, the constant pitch maneuver will be employed as a last ditch effort for
RECOMMENDATIONS
recovery altitude. The research presented herein is contrary to many of these studies. Not
disparaging the experience and knowledge of the previous researchers, it is felt that
The flight simulator used for the research is a training device. As such the
environment programmed into the flight simulator is optimized for training, and this
includes the microburst. Any flight descending through 1,200 feet AGL would
experience the same microburst wind shear, regardless of altitude. In the atmosphere, a
microburst has a varying wind with height; higher vertical winds with less horizontal
wind shear at altitude and lower vertical wind, but greater horizontal wind shear at low-
levels. One of the benefits of the variable pitch maneuver that could not be tested is
escaping this high vertical wind at altitude and using the increase in airspeed to mitigate
the horizontal wind shear. A more realistic microburst model, with varying wind, would
increase the validity of transferring the data to the real world environment.
The microburst required positive control of the aircraft; airspeed, pitch attitude,
roll, yaw, and displacement continually changed in the microburst wind. During the
variable pitch maneuver, the pilots found that upon arriving at the target altitude, it was
easier to try and maintain a pitch attitude of about 11° on the ADI, rather than holding, or
trying to hold, altitude. A flight director steering command programmed for the variable
pitch maneuver may provide for even tighter altitude tolerances. Any future maneuver
should consider reprogramming the flight director for pilots to follow. It is felt that this
161
would greatly aid in reducing variance between and among pilots while providing a more
A program sub-routine can be written to capture the difference between the pitch
attitude commanded by the flight director and the actual airplane attitude. This might be a
better indication of pilot performance than recovery altitude. The use of recovery altitude
into account airspeed and other aircraft parameters would increase relevancy.
simulators, however, the phugoid presents several problems. McCarthy and Norviel
(1982) report the long period mode phugoid frequency exhibits an overdamped response
in standard training flight simulators (p. 29). Frost, Turkel and McCarthy (1982) concur
with this finding that low-frequency response is overly damped in the simulator (p. 10).
The phugoid influences the altitude, airspeed, and controllability of the aircraft as
it transitions through a microburst. The nature of the constant pitch maneuver excites this
mode, thus decreasing aircraft performance. If not accurately modeled, the constant pitch
The simulator evaluation handbook (Royal Aeronautical Society, 1995) allows the
same tolerances as the FAA. The wavelength of the phugoid in the simulator should be
±10 % of the flight test value. Time to half amplitude is equally controlled at ±10%,
while damping ratio is ±0.02. Variance between flight test data and simulator data is
acceptable in evaluating the phugoid. "The purpose of this test is not to obtain a perfect
match of all plotted parameters for the entire length of the manoeuvre [sic]" (f 2C.33).
The test is performed in cruise flight and should include 3 full cycles of the phugoid, or
Prior to conducting a simulator study, the phugoid mode should be examined for
provide data only as reliable as the input. Knowledge about the mechanics of microburst
wind has grown substantially since the accident of Eastern 66, however, any simulation is
a reflection of what should happen, and not necessarily of what does happen. Good
decisions are based on good information, and there might come a time when actual
To meet the potential requirement of real world data, Psiaki and Park (1989),
advise that escape maneuvers be tested in remotely piloted vehicles (RPV). "The danger
of flight testing can be avoided by using relatively cheap RPVs. Flying them in
thunderstorms under the automatic control of some of the suggested guidance schemes
should provide a wealth of data by which they can be evaluated" (p. 1138).
employ a more realistic microburst, one with varying wind at altitude, a guidance system
for the pilots to follow, without substantial training, and an accurate phugoid mode
oscillation that is not overly damped. Real world data, obtained by RPVs will enhance the
American Airlines. (1990). DC-9 Operating Manual. Rancho Cordova, CA: Aviation
Business Services.
Author.
Atkins, N. T., & Wakimoto, R. M. (1991). Wet microburst activity over the southeastern
Barnard, R. H., & Philpott, D. R. (1995). Aircraft flight: A description of the physical
Boeing Company Windshear Task Force. (1985, January-March). Windshear: Flight path
Bohun, R. (1671). A discourse concerning the origine and properties of wind: With an
W. Hall.
Brady, C. (2003). The Boeing 737 Technical Site. Retrieved August 14, 2003, from
http://www.b737.org.uk
Bray, R. S. (1986). Aircraft performance and control in downburst wind shear. In:
Aerospace Technology Conference and Exposition, Long Beach CA, October 13-
Byers, H. R., & Braham, R. R. (1949). The thunderstorm. Washington, DC: United States
Weather Bureau.
Caracena, F., Holle, R. L., & Doswell, C. A., Ill (1989). Microbursts. A handbook for
Caracena, F., Maddox, R. A., Purdom, J. F. W., Weaver, J. F., & Greene, R. N. (1983).
Cashman, J. E., Kelly, B. D., & Nield, B. N. (2000, October). Operational use of angle of
Cotton, W. R., & Anthes, R. A. (1989). Storm and cloud dynamics. New York: Academic
Press, Inc.
Dodge, J., Arnold, J., Wilson, G., Evans, J., & Fujita, T. (1986). The cooperative
Dogan, A., & Kabamba, P.T. (2000). Escaping Microburst with Turbulence: Altitude,
Etkin, B., (1972). Dynamics of atmospheric flight. New York: Wiley & Sons, Inc.
wind shear alerting and flight guidance systems. (AC 120-41). Washington, DC:
Printing Office.
Federal Aviation Administration. (1999). Aviation weather services (Rev. ed.). (AC 00-
53(8), 1-11.
Frost, W., Turkel, B. S., & McCarthy, J. (1982, January). Simulation of phugoid
excitation due to hazardous wind shear. AIAA 20th Aerospace Sciences Meeting,
Fujita, T. T. (1976). Spearhead echo and downburst near the approach end of a John F.
Kennedy airport runway, New York City (SMRP res. 137) Chicago, IL: The
Kenner, Louisiana on July 9, 1982 (SMRP res. 199). Chicago, IL: The University
Fujita, T. T. (1983b). Andrews AFB microburst (SMRP res. 205). Chicago, IL: The
Fujita, T. T. (1985). The downburst. Microburst and macroburst (SMRP res. 210).
Fujita, T. T. (1986). DFW microburst: On August 2, 1985 (SMRP res. 217). Chicago, IL:
Gartner, W. (1977). Piloted flight simulator study of low-level wind shear: Phase 1 (FAA-
Hallowell, R. G., Wolfson, M. M., Delanoy, R. L., Forman, B. E., Pawlak, M. L., Smith,
performance, capabilities and limitations. In: Workshop on Wind Shear and Wind
Higgins, P. R., & Baker, E. L. (1986). Insight gained from analysis of windshear related
Long Beach CA, October 13-16, 1986 (pp. 13-38). Warrendale, PA: SAE.
Higgins, P. R., & Patterson, D. H. (1979, January). More about wind shear hazards.
Higgins, P. R., & Roosme, A. (1977, January). Hazards of landing approaches and
Hinton, D.A. (1988). Flight management strategies for escape from microburst
wind shear encounters. NASA technical paper 2886. Washington DC: NASA.
Montreal: Author.
Ireland, R. L., & Simmon, D.A. (1986). Evolution of windshear training at United
Airlines. In H. Rosen (Ed.), AIAA Los Angeles Section: Vol. 32. Wind Shear
Jacob, F. (1982). The possible and the actual. New York: Pantheon Books.
Job, M. (1994). Air Disaster (vol. 1). Weston Creek, Australia: Aerospace Publications.
University Press.
McCarthy, J. (1983). Windshear and aviation safety: The JAWS project. Eastern Flight
Line, 8, 4. 17-22.
Workshop on Wind Shear and Wind Shear Alert Systems (pp. 1-13). Boston:
McCarthy, J., Blick, E. F., & Bensch, R.R. (1979). Jet performance in thunderstorm wind
McCarthy, J., & Norviel, V. (1982). Numerical and flight simulator test of the flight
In: Second International Conference on the Aviation Weather System (pp. 247-
McCarthy, J., Wilson, J. W., & Fujita, T. T. (1982). The joint airport weather studies
Melo, de D. A., & Hansman, R. J., Jr. (1990). Analysis of aircraft performance during
AIAA.
Technology Conference and Exposition, Long Beach CA, October 13-16, 1986
Miele, A., Wang, T., & Melvin, W. W. (1988). Penetration landing guidance trajectories
Miele, A., Wang, T., Tzeng, C. Y., & Melvin, W. W. (1987). Optimization and guidance
Moldrem, L. J. (1996). Tiger Tales (Rev. ed.). Clearwater, FL: Flying M Press.
Mulgund, S.S., & Stengel, R.F. (1992a). Target pitch angle for the microburst escape
Mulgund, S.S., & Stengel, R.F. (1992b). Optimal recovery from microburst wind shear.
do: Taming the microburst windshear. Retrieved April 18, 2002, from
http://www.dfrc.nasa.gov/History/Publications/SP-4216/ch5.html
National Research Council. (1983). Low-altitude wind shear and its hazard to aviation.
National Transportation Safety Board. (1968). Aircraft accident report. Braniff Airways,
Inc., BAC 1-11, N1553, Near Falls City, Nebraska, August 6, 1966 (SA-393, File
National Transportation Safety Board. (1974a). Aircraft accident report. Iberia Lineas
National Transportation Safety Board. (1974b). Aircraft accident report. Pan American
World Airways, Inc., Boeing 707-32IB, N454PA, Pago Pago, American Samoa,
National Transportation Safety Board. (1976). Aircraft accident report. Eastern Air
Airport, Jamaica, New York. June 24, 1975 (Report No. NTSB-AAR-76-8).
report. Pan American World Airways, Inc., Boeing 707-32IB, N454PA, Pago
Airlines, Inc., Douglas DC-9, N994VJ Philadelphia, Pennsylvania, June 23, 1976
National Transportation Safety Board. (1983). Aircraft accident report. Pan American
World Airways, Inc., Clipper 759, Boeing 727-235, N4737, New Orleans
National Transportation Safety Board. (1985). Aircraft accident report. United Airlines
Flight 663, Boeing 727-222, N7647U, Denver, Colorado May31, 1984 (Report
National Transportation Safety Board. (1986). Aircraft accident report. Delta Air Lines,
National Transportation Safety Board. (1995). Aircraft accident report. Flight into
product derived from GEOS I-M satellite sounder data. In: Seventh Conference on
Meteorological Society.
NTSB studies wind shear as factor in Eastern crash. (1975, June 30) Aviation Week and
Neter, J., Wasserman, W., & Kutner, M. H. (1990). Applied linear statistical models:
Poellot, M., Borho, A., & Bassingthwaite, S. (1997). Microbursts as an enroute hazard-a
Proctor, F. H., Hinton, D. A., & Bowles, R. L. (2000). A windshear hazard index. In:
Robinson, P. A. (1991). The modeling of turbulence and downbursts for flight simulators.
Rosenfeld, J. (1999). Eye of the storm: Inside the world's deadliest hurricanes,
Royal Aeronautical Society. (1995). Airplane Flight Simulator Evaluation Handbook (2nd
D.C.: NASA.
Smith, M. R., Pryor, S. P., & Prater, E. (2000). The crash of USAir flight 1016, preparing
the weather case for trial. In: Ninth Conference on Aviation, Range, and
Soffer, E. P. (1990). Establishment criteria for integrated wind shear detection systems:
Low-level wind shear alert system (LLWAS), terminal Doppler weather radar
Stinton, D. (1996). Flying qualities and flight testing of the airplane. Washington DC:
AIAA.
Stull, R. B. (2000). Meteorology for scientists and engineers 2nd ed. Pacific Grove, C A.-
Brooks/Cole.
United Air Lines, (1991). 737-200 flight manual performance and procedures, additional
United States Air Force. (1980) Flight test handbook: Flying qualities handbook.
United States Air Force. (1983). Flying Training Air Navigation (AFM 51^0).
Visser, H.G. (1996). A minimax control analysis of lateral escape maneuvers for
Visser, H.G. (1997). Lateral escape guidance strategies for microburst windshear
associated with intense convection: The 20 July storm during the MIST project.
Webb, J. (1990). Fly the wing. Ames, IA: Iowa State University Press.
Sanderson, Inc.
Conference on Severe Local Storms, Alberta, Canada, 22-26 October 1990 (pp.
APPENDIX A
•r
-
t^»
C>
A/ '•' 1
1
I t
~"
UJ
L3 -"S' »
1
<-D
-
/
t'
»•-•<
U4
re -_, J I 'J\iix-\JL J fii i i-
_ :\ \
j. J .t , L_, i v
_- i..... i A .1., _...! 1. ,-L 1 f
- STICK SHAKETR
/
\A, / V
• \ /
rW _J ! l l 1 t
•W^.
-J„K—j.
~i I 1 L L J
lOO
TIME - :EC TIME ~ SEC TIME - SEC
: H
1 A
1
I 1 L_ -1 1 L I . M
ion
; i t
x1
1 J i-i . 1. ....
l 1
\,
.J 1
_L__i i^ • 1 [_-/y.
L
UME - SEC TIME - <^EC TIME - SEC
- STICK SHAKER
« !-••-'
V \
W
^**t 4^ J i 1 1
\ hf
L_y_j i
iUO L L
- -_J I -JL L_ _L I
o TIME -•EX TIME ~ SEC TIME - 3EC
A u
-L __J i i. J J J i i _-LJL J L L. _JbJ J I u
n m
TIME - SEC TIME - SEC TIME - SEC
/ v
STICK SHAKER
VV,VM^ ^
' v " v \\/,VMl\
v
in\
_1 i I i l I L
v -./
L___i _ J
l J L __I__„J L.
too ' ~-
o
TIME - SEC TIME - SEC T I M E - SEC
WEIGHT: wQOOO L P -
C. ©. J>r.. O ":MAC
AIRPORT TEMPS 8 2 lit In.
SURFACE WIND: 8 « -1 •
\ 'V A
Z
r \
i
I.. I I
_J L J _L . _l S I L._— I L. .'
mo
riMC - EC riMr ~ «-_.E«~ TIME - -E>
i
r
i
/
:.... "\A, i
^^MyW
: i \/
i t i i ! . i i * i i
HEIGHT' OOQOO L P ^
i . tf. 25. n MAi
ATkPOFT TEMPs S^ PEV». «
SURFACE WIND: & C? !£<=»
APPENDIX B
/
-
/
"
t
-
/ \ }
_i I . I L 1.... i ... , i S - 1 1 1 1 . ,J_ i.. .I
STICK SHAKER
s \
A :^.<A
v^ KVr\
m o ••
o
!_. 1 I
t,
I 1 I I I
too
_J ^ L.
/"
_ ___ M} \
«-» -1- S !- fc ^
i
/ \ /
L
s
^ \\ /., ,/t 1 1 -J _. i
r?
7
_1 L J L r , . . - }> .
„:•?._... . . 1 ..J I , J
-
STICK SHAKER
-
« ISO -
—--"""""
r
- ' \ \
w
\
l s /
/
/ ^ :-' V SA.
*V~\j
-J
Jrx
_J i I I -JL. »
: i i. J... 1 t A « 1 1 \
1r
E '
- ^ / - ' ^ V „"A
i i—J-J -/-i i 1 -L
7/- "W
-J L I t . , L- 1 i J I l __L™._JLL„ L_ L___J__J I
- ^
« l^O
I
—-.-'"
1 1. i
TIME ~
M A
i. . . . i
SEC
V
, . 1 1 » .J
TIME - SEC
<v
UJ 5 -
TIME - SEC
WEX&HTs VOOOO L G S
C. G« ^ S . O "iMAC
AXRPOPT TEMPI 3 2 DCG. C
SURFACE WXNIts ~« « dS1?
_ y
r
i
/
/
- t r N
/
}
i i 1.......1........ f 1 1 !. i t i _J I J I S t | \ \ 1
too
T I M E - SEC TIME - SEC TIME - SEC
_.
STICK SHAKER
h
/A : J
L ,s
\
M^ft - ^h
'V
r- _L~_1 !_ I I L_ L~-__L_i .--J
I 1. , . - L - i 1 J 1 J „..!.... J
_J 1 Xjy*
V.VI LJL-JL. _1 I.. I I J
/\
J L L J _ L_ ~J „_x__JU l .JL-__ ..j,.. ,I l i I
\ /
-J, . . I, J L_-™X.
HEXAHT* ^OOOO L B ^
i i^ o n * MAr
AIRPORT TEMf-: 32 DE© «
MJFFnLr MX Nil* O <* 1W9
. 3
-'' A
15
t "'IMr
lO
J
i%
/• \
5 J
L
1
"\ .,/
I [ '• , A ft/»v i i
_ I I
- v
-J L 1 ™J L_™_L- _i l L -J. ._L_„Ji I J
l«JO
~5
STICK SHAKER
-A H\^yAv%\I \ v
u 1
_l 1 1 l_JL_J M
K l_J 3
"V"
X„ L___J I J I 1 *
-J _JL-™X™~JL I 1____JL_-J L 1
-
.pM,
-
- A •H^
1 -_..L. J 1
_JL L X_ JL—. I I J 1 i.... ...L L 1 i
...1
too
TIME ~ Et TTME - SEC TIME - =-EC
h- STICK SHAKER
- \
RMV ^^^^M/^
o 150 A
IA
--s,V A»• \A
V
J L 1 I \ 1_. I I J
' \ /
-J ( L__~J . I ! J I
MEIGHT: "9QOOG L B S
C. O. _."•. 0 ^MAC
HTPFCIF-T TEMP* 32 HE&. C
»- in
tj n
•A " .
-L—JL, L-A-i L _1 i J
_J I I- I I I L J 1 _J I
1 I
I
I _ i _ _ X „ I __ 1_ > 1 1
lOO lOO
TIME - SEC TIME - «-EF TIME - SEC
STICK SHRKER
^^M^^tJ1
V\
V\/ _i J l I J
_J L I _J_ _J J 1 I »
lrj
iU / ' 'V
£ /
<v,
_1 L L
--J L _ I I L J„_. _1 ! L . .1 J J I L J J
_
STICK SHAKER
r
-
-
-N %v25(Wv w - % ,M/
-J l I
vL
\ 1 1 1 I 1 .-___ J
~~, /
"_-'
_
t»
'-- /
._ /
C3 ri^n f—. /
CO
cr
_r
UJ
-
. i .1 i I i i. . i -_- Li_i 1_
TIME -- SEC TIME - SEC TIME ~ SEC
is STICK SHAKER
B K'M
I » » » 1 t L.__l
=A
~~A
J L„ 1 J X 1 I .1. ../J 1 1 L-__X___J
lOO
TIME - SEC TIME ~ ^EC TIME - SEC
L.,-s ,
' \
/
i J ,.t ...i. T t 1 1
<l \
/\
£ 10
m
\ ^
~' ~\
-J i I L_ -J J L-_, -X_, 1 I I
lOO
TIME - ?EC TIME - SEC TIME - SEC
-
- STICK SHAKER
LJi-JX I J
V
K/
x
\ / -J I J I I I L_
i . . . i . i
f
TIME - SEX *- i i i TIME ~ SEC TIME - SEC
m
-
- \
f f {
/ J. U
- ^\ f-
/
/
- V./' / -Vv-^r-^
^~
i i i , \ i _. j
CvCv
~
. A . . 1 1 .1. . .. 1 I i 1 .. 1 ...) ___________SI L I l J »_J
100
TIME - SEC TIME - SEC TIME - SEC
STICK SHAKER ft
.' \
A V-NA-V.^"-
K M
_l I I I —I - 1 - I
-J i [ _ _ .
TIME ~ SEC TIME - SEC TIME - SEC
WEIGHT: <?ODOO L B S
C . ©. 2 5 . O MMAC
AIRPORT TEMPS © 2 DEG- C
SURF'ACE WIND! 8 <? 1 3 9
a:.-;
U\
1
i:* *
•~>
Ul
re
~
n I L __i ,1 1 ! i 1,.,... i t -t i H— . \ f. u a_~~_J_ J J
y\ •> / UJ 5 -
V- v
v
.j i Ji i t < i i L i
a.
UJ
<r
a.
_. O
1. e
1. 3
™
-
i
\
sl^V^A
rW
J
' \
1. 3
_t > I I \ I I 1\ - 1 !. 1 .i 1. ...J
A
v\
/
_X_. ,1 1 . I\^ 1 I I
TIME - SEC TIME ~ SEC TIME - SEC
APPENDIX C
£ 1D
MJ\
w
I
\ r
J L _ _J L_
I '\
.J _J .-_w__L___J™™L I I L I
w^ _L_ -U-'.
L_y
"\
-
j
A/ ^ ^ ^ ^ A
V
1 1 1 _..L. X * 1 1 J
'• J \ J. JL_.il.
A I L J I I .1 I t , I
y — , \
_i i t. \L i i i i i i
lOO
TIME - SEC TIME - SEC TIME ~ SEC
\
L_
« ISO
k
A / *A
1 X L _ . .......1 L... __.t L. ... 1. . . 1... 1 J
-. 3
1. o
en .^.n
/
A
\
-J L__„J „ _ J 1 . _I„__J L I I U __I LJ_wL_ A X___„J i .J
TIME - SEC TIME - SEC TIME - SEC
1
—
rt ISO - -'\ /, /
— s^ y v V\ / /
to
A
\J V^A/'
. 1 ... J 1 t. X,,, ...\ i »., i ,i
1 im
_.
X
__
U.J
-
„
-_ \
1 !.. 1 . ._ J L 1 f i 1 1
lOO
A
_J___ t
x
LA. I .
j
12 L.
TTM£. ~ ^EC TIME - SEC TIME - SEC
1 5 k STICK SHAKER
J
\
V
AA / p^^ "~yf
|V/t, *
JU, \ I. , \ I 1 L
,J I I 1 t , 1 I I I f - S O L-
O
TIME - SEC TIME - SEC TIME - °EF
- ~_
J
f
A
1
TIMr stc
i
L i
1
,\
i„
\
L A~
TIME
I1
„J__ i
- SEC
1_ ._J~ _L_ J
lOO
TIME - ^EC
'
-
/. /
/ JL- ~t ~J ftX
^ VV,
nfw 1
L — / ftWjy
i 1
HFI©HT s o n i i n i Ll?b
i . <-. I ' A O \riHC
AIRPORT TrMp; er UER. C
sUPF H L C WIND: ° (=» i85>
' *
A ^JtlL
•\Vv gH
J L I I L J J L -J- ! _ J __L____i__ L. L_ I
ino
TIME - "^E«~ TIME - SE* TIME - -EC
-
1 \ STICK SHAKER
f
^A / m
--
„ A. L .. i i J -i ...-.-. i .. ,i
*^K ui 5 -
A ™J I I L
\L I\
WEIAHTe <=»UOUU L t «-
L. G- . *A A *.MAi
HlRfOPT TLMf s . D £ M .(
>HPFACE WINDS F . 1 C -
r
'INI NT ND HI Mf- t I Mill n ! OP r * . ~ 1*4 I t
.,ft.A
M
A
IA
M
_L I I V\{\«/ Ifl) L_
.Au .
-J L_ L_ 1 L L J L I L_ L__2J I I I L
STICK SHAKEP
l,U M,
II
v\
V i# _j x i t J ii ._
A 1 __™-J J L
( . \ .. *.*-.. i J M 1 H L
\/L /
*J \ i\ WHtXiti
W ]A
A I l_J I L_
TIME - SEC TIME ~ SEC TIME - SEC
~"
- \
\
/
£ 10
,--W
vVl
-
\
1
\ / - " /
V
r
V.AJ • ' A
a
_J_ t -.1 1,. .1 1 1 J i i iL J LA. I L I I I
/
f
V ~^4*M{ a
\A lyW
l o c , I X J I I I I I I L I
O lOO
TIME - SEC TIME - SEC TIME - SEC
\ /
J 1 L -I I I J __J 1 X_Ji I LJ » ,. A _ t I I
100
TIME - ^EC TIME - SEC
TIME - SEC
> i i i
V
i i_
^4ji)/j
WEIGHTS «OOOD L B S
C. ©. ? " ) . n I-.MAC
AIRPORT TEMPt SJ2 DEA. C
SURFRCE WINDS © g ia*
V
-J 1 1 1 J J- I 1 J • "V.. JL .V.t .1 J...
STICK SHAKER
lOO > L- A ! L. ! J ~ 3 «-
O O
TIME - SEC TIME - SEC TIME: - SEP
WEIGHTS « n o n n t R<->
CO. 25 O ^MAC
AIRPORT TEMP* A £ DF©. r
SURFACE MINDs & ^ 1«»*
£ io VVT
,^-^A
-A—J--/-J _l I 1—*vrp
V/"
.1 I 1 I 1 S A I ______ --, L_
O
TIME - SEC TIME - SEC TirtE - SEC
W«W
*$w I I J
_J J L___ J 1-_J I -J i I
TIME - SEC TIME ~ SEC TIME - SEC
- * - 3 __ 2G
_, \
1?
— -\
--
5 ~.y,^Ax.
1 1 _ i. i
TIME ~ SEC
i.... i i » i -J 1 ...X XX A
TIME - SEC
L_ J L
TIME - SEC
« iso
MWVy*
- x>_~~x
_ \ * / \
— -v. •
_- 5 -
\ (\(4
• '\J"J
A /
- ' " \ 1
\ --
—
-
X/ V
I... .1 „_J ... , . i ._! .. 1 1 1 I '"""1 ^ i - _1 A —I I
TIME ~ SEC TIME ~ SEC TIME - SEC
'V
A I I L
TRAINING? W I N D « H E . i i K l : r
TtV I L ^ I UF- Gr r 3?' « 4 FF^
^-
u_
V_i
l_i
O "S.O
-
ca
<L
t—
_c
u&
i-«
UJ
_T
\, -.
c=> ISO Vn /
-JL-J I J L
-j—> 1 1 — 1 — 1 1
10
£
V •
A L, L____JL.
TIME - SEC
1 , 1 t I 1
"XJ J
lOO
TIME - SEC
TIME ~ SEC
__v« '^xW
inn I I t —J I I I I I I I
O lOO
TIME - SEC TIME - SEC TIME - SEC
WEIGHTS 90000 L O S
I . G. 2S. U ^MAC
AIRPORT TEMPS 62 DEG- C
SLIRFACE WINDS B <_ 189
IA /
X 1
~ s.
K/
~ - /
/
- V- . /
A
t f ..—1 1 1.. 1 I .. _. J -J I Xl I . __LI l_ l_
TIME - SEC TIME - SEC TIME - SEC
-VtyVV
«^iite
TIME - SEC TIME - SEC TIME - SEC
TPMININW NlNDrH_.HF : c
XMIlLnTOR L ~ 3" **4 FF S
UJ 4 O 1D
£
v.- J L _ J
STICK SHAKER
\ -w- *\ V, J
V
/
l c o L J -JL____-1 1 L i 1 L 1 _J
O IOC
TIME - SEC TIME - SEC TIME - SEC
10
£
A
i I —l i i i .J_____J ZL. A I I
TTMF - ^EC TIME - SEC TIF IE - SEC
.,_.,....,. ...j.
I j i i '1 1 ! i i i i ' ' S l
- /
< - }
!
H
i -
ctT K
• \
o
-
s (,
CD - X^
a
< OL .
CD O
-
3 w_ *
ill 1 11 i 1 i • I ' I 1
. ;
H H _
H H
I I
n n
•n i
r y>
x>
•o m
0. n
o n !
i
!
j
O U O J» o
J
r~r T rn—i—i—s—i—i
to