AIR
LINE PILOTS ASSOCIATION, INTERNATIONAL
535
Hemdon Parkway P.O.Box 1169, Hemdon VA 22070-9805 (703) 689-2270
Petition
for
Reconsideration
of
Probable Cause
Business
Express,
N811BE,
SN UB-49
Block
Island, Rhode Island
December
28, 1991
Accident
No. NYC-92-FA-053
June
25,
1997
Submitted by
James M. Walters
&
Steven D. Green,
Airline Pilots Association
1997
Table
of Contents
(The following were not reproduced for this posting)
Diagrams: Wreckage
Documentation and Systems Layouts, #1 -
#7 Photographs: Wreckage
Photographs, #1 -
#54
Attachments: Attachments
A - H
Compact
Disc:
Inside
of back cover
List of Attachments
Attachment A: |
AIG - Business Express, Inc., SN UB-49 Beech 1900C, Failure Investigation, Packer Engineering, (metallurgy report)
|
Attachment B: |
Radar Data Study, Data List, Plots, N811BE, Beech 1900C, Block Island, Rhode Island, Associated Data Resource, (wreckage plots and drift reports)
|
Attachment C: |
Letters, (2), from Mr. Joseph G. Dondero, Vice President Maintenance, Pennsylvania Airlines, and spokesman for the Beech 1900 Operators Committee, to Mr. Dave Jacobson, Manager of Airline Support, Beech Aircraft Corp., dated September 3, 1991 and October 16, 1991.
FAA Internal Memorandum, Principal Aviation Inspector (Airworthiness) NE-FSDO-03 to Manager, Aircraft Certification Directorate ACE- 100.
Airworthiness Directive 91-12-02
NPRM Airworthiness Directive 93-CE-4 1-AD, sp. AD 92-06-09.
|
|
Attachment D: |
The Beech Aircraft Corporation Model 1900 Airliner Engine Truss: A Study in Reliability Analysis and Aviation Safety, Dr Ron Stearman P.E., M. Buschow and K Kane
C.V. Dr. Ron Stearman, P.E.
|
|
Attachment E: |
Aircraft Damage Detection from Acoustic and Noise Impressed Signals Found by a Cockpit Voice Recorder , Dr. Ron Stearman P.E., G. Schulze, and S. Rohre, Institute of Noise Control Eng. |
|
|
Attachment F: |
Materials Engineering Investigation Conclusions, Chronological Event Graph, Dr. Richard H. McSwain, P.E., McSwain Engineering, Inc. C.V. Dr. Richard McSwain, P.E
|
Attachment G: |
Expert Report, Donald Hamill (engine teardown) C.V. Don Hamill
|
Attachment H: |
Cockpit Voice Recorder Tape Erasure Gap Study and Signal Level Inventory and Study, Glen Schulze, Data Acquisition Sys.
C.V. Glen Schulze |
page ii
Introduction
On
December 28, 1991, a Beech Aircraft Corp. 1900C, operated by Business Express
Inc. (BEX) crashed while executing a VOR approach to Block Island airport,
Rhode Island. The flight was operated under the provisions of 14 Code of
Federal Regulations (CFR) Part 91, and under visual flight rules (VER). As a
training flight, only company personnel were on board. There were three
fatalities, including the instructor pilot and two trainees. The National
Transportation Safety Board (NTSB) determined the probable cause of this
accident to be:
“...the
instructor pilot’s loss of altitude awareness and possible spatial
disorientation, which resulted in the loss of control of the airplane at an
altitude too low for recovery; and company management’s lack of
involvement in and oversight of its Beechcraft 1900 flight training program.
Contributing to the accident was the instructor pilot’s exercise of poor
judgment in establishing a flight situation and airplane configuration
conducive to spatial disorientation that afforded the pilots little or no
margin for error.”
The
Air Line Pilots Association (ALPA) takes issue with the probable cause adopted
by the Board and with numerous statements contained in the Board’s report
[1] .
In addition, ALPA takes exception to seven of the eleven Findings in the
Board’s final report, and believes these findings are erroneous. Those
findings, and the errors they contain, in summary, are:
Board
finding #2.
“There
was no evidence of airframe or powerplant failures prior to impact with the
water.”
Errors
contained in Board finding #2.
There
was (and still is) extensive and irrefutable evidence that, in fact, there was
structural failure of several key components of the accident aircraft while
maneuvering
at approxnuately 2000’ altitude
.
Those would include the right engine truss tube assembly, which caused a whirl
flutter event in the right engine and propeller and subsequent wing and
empennage failures. AU occurred prior to impact with the water.
Board
finding #3.
“There
were no airplane system malfunctions or failures before impact with the water,
except when electrical power to the captain-trainee’s attitude indicator
was deliberately removed.”
Page 1
Errors
contained in Board finding #3.
As
stated above, ALPA has documented evidence supporting the scenario of a
catastrophic inflight breakup. Obviously, during and after this breakup, there
were disruptions and failures of all aircraft systems.
Board
finding #5.
“The
IP [Instructor Pilot] disabled the captain-trainee’s attitude indicator,
and about 6 minutes later he simulated a failure of the right engine by
retarding the power lever to the flight idle position, which in effect,
introduced multiple emergencies contrary to the provisions of the
company’s BE 1900 operating manual.”
Errors
contained in Board finding #5.
At
no time were any company operations specifications or operating policies and
procedures violated by the instructor pilot or trainees. The IF simulated the
failure of the right engine by reducing the power to
zero
thrust only
,
not to flight idle. The IP had valid reasons for simulating those particular
system failures during this training fight.
Board
finding #6.
‘The
IP used poor judgment by encouraging the captain-trainee to fly with his
attitude indicator disabled and uncovered, followed about 6 minutes later by a
simulated failure of the right engine under simulated instrument conditions on
a dark night.”
Errors
contained in Board finding #6.
The
IF demonstrated his technical and professional competence during all training
maneuvers, as documented on the CVR.
Board
finding #7.
The
IP failed to recognize in a timely manner that the captain-trainee was
spatially disoriented when the captain-trainee asked the IP to take control of
the airplane (‘Your aircraft?”); instead the IP attempted to coach
the captain-trainee into a recovery from an unusual attitude.
Errors
contained in Board finding #7.
There
is no evidence that the IP was ever “spatially disoriented” during
the flight. There is substantial evidence that he was continually aware of the
attitude
and altitude of the airplane at all times. Although bank angle may have been
excessive during the last few seconds of controlled flight, there is no
evidence that the trainee was ever in an “unusual attitude”. The
comment “Your airplane?” may in fact be a query as to who should be
or actually was flying the aircraft, not a request for the IP to take over
control.
Board
finding #8.
The
attempted recovery from an unusual attitude was not successful, apparently
because the IP lost awareness of the airplane’s altitude and rates of
descent and may have become spatially disoriented at an altitude too low for
recovery.”
Errors
in Board finding #8.
The
IP never lost awareness of the airplane’s altitude or rate of descent and
never became spatially disoriented. Therefore, because no “unusual
attitude” was encountered, there was never a recovery attempt. The last
few minutes of radar returns, including the very last return, show the aircraft
in level flight at 1900 feet. In fact, due to whirl mode flutter and
pre-existing structural faults, the right engine departed the aircraft in level
flight, striking and removing the right horizontal stabilizer. This caused the
instantaneous failure of the outboard portions of both wings, and the total
loss of control of the aircraft.
Board
finding #9.
“The
airplane probably crashed into the ocean in a near-inverted attitude with the
outboard section of the left wing striking the water first and with the
longitudinal axis at a substantial angle with respect to the surface of the
ocean.”
Errors
in Board finding #9.
The
exact attitude of the aircraft at time of impact with the ocean is unknown.
However, a major portion of the right wing, the right engine & nacelle, the
right horizontal stabilizer and the outboard portion of the left wing were no
longer attached to the airframe upon water impact, having separated at
approximately 1900 feet MSL. It is believed, based on documentation done by the
Board as well as that done by ALPA, that the remainder of the aircraft impacted
the water nose and left wing root first, in a nearly vertical attitude.
Because
it has been over five years since this accident, ALPA and others have had the
“luxury” of time to investigate, document and analyze all the
factual materials and information related to this inflight breakup. This
evidence is obvious and extensive. Much of it has been discovered subsequent to
the publishing of the final report by the NTSB. However, ALPA is disappointed
that
the Board investigator assigned to this accident chose to conduct only a very
cursory investigation. During the field phase, many very important facts went
undiscovered, and evidence that did not fit neatly into the final scenario was
simply overlooked.
New
evidence that is presented includes:
- Detailed
history of the well known problem of B 1900 engine truss tube assembly cracking
and separation,
- Wreckage
documentation of the landing gear system, the hydraulic system, the elevator
trim system and the right engine,
- Structural
documentation of the right nacelle, the lower spar cap and the empennage nose
cone,
- CVR
spectroanalysis by two independent laboratories regarding several portions of
the CVR tape,
- Operational
considerations, including the use and analysis of the altitude alerter system.
ALPA’s
supporting documentation for all assertions is contained in the following
petition. The major subject areas are:
Wreckage
Documentation
CVR
Human
Performance
Summary
Findings
As
required by the Board, new evidence is presented and the errors in the
Board’s original report are detailed in ALPA’s Petition for
Reconsideration of Probable Cause, Beechcraft 1900C, N811BE.
Wreckage
Documentation
General
The
NTSB summary report of this accident, NTSB/AAR-93/0 1/SUM adopted April 23,
1993, devotes only two pages to the documentation of the wreckage ofN811BE.
Detailed studies of those components of the aircraft that were recovered, and
are still in storage today, have been completed subsequent to the publishing of
the official report
[2] .
The findings are as follows:
Right
Wing
The
right wing failed near WS 124, at the outboard side of the nacelle (see Diagram
#1 for locations of wing structural failures, see photos #1, #2 and #3). The
lower main spar cap failed at three locations, with the structural member of
the lower spar cap separating along the lateral bonded surface:
The
after portion of this member failed at WS 149 (failure “F4”,
diagram #1).
The
forward portion failed further out, at about WS
158
(failure
“F5”,
diagram
#1).
The
skin covering the spar cap failed even further out, at WS 211 (failure
“F6”, diagram
#1,
detail of photo #2).
The
upper main spar cap extends inboard from the main fracture site approximately
28 inches. This would place the failure at WS 96, inside the nacelle (failure
“F7”, diagram #1, see photo #4 and detail in photo #2). The aft
spar failed at the WS 120 connector. However, the upper cap showed a fracture
along the lateral axis, behind the web, which apparently initiated at a rivet
hole. The main spar web is compression buckled outward to the location of the
principal lower cap separation, that is, WS 149 (“F4”, diagram #1,
see photo #6). The bottom panel is also compression buckled to approximately
the same location.
The
Packer Engineering report
[3] (attachment
A) identifies all of these fractures as tensile overload or tensile tearing
fractures. In the case of”F4” and “F7”, diagram #1,
rubbed areas are visible.
The
control surfaces attached to this wing section are the outboard flap and the
aileron. The root of the outboard flap corresponds with the fracture line at WS
124, and this component shows considerable compression buckling (see photo #6).
Both the flap and the aileron have been shifted laterally outboard, towards the
tip of the wing. This is confirmed by examination of the control surface
hinges. All show an outboard displacement.
Photo
#5
shows
the fracture of the forward lamination strip of the lower main spar cap. This
segment
is twisted approximately 20 degrees out of plane with the aft segment of the
cap. This twist corresponds to a leading edge lift of the wing.
Visual
examination of the wing shows it to be straight, with no obvious evidence of
set. There is no damage whatsoever to the leading edge or wingtip (see photos
#7 and #8). The navigation light plastic cover on the right wing is intact. The
bulb was removed by the NTSB for microscopic filament examination and was found
to be stretched but not broken. The wing is straight, and the skin shows no
damage. The stall fence is intact, indicating that the wing could not have
struck the water with any speed. The right wing is, in summary, remarkably free
of damage! (See photo #9).
As
documented in the NTSB report, the right wing was found floating on the.
surface of the water over 4 miles from the main wreckage area. It is
significant to note that the location of the wing when found was opposite the
prevailing direction of surface drift (from the Southeast to the Northwest) but
corresponding to the general wind direction (280 degrees at 12 knots) present
at the time of the accident (see Coast Guard Logs contained in the “Radar
Data Study”
[4]
and
graphic layout, attachment B).
A
reasonable conclusion, then, based on all of the preceding evidence, would be
that the right wing failed in flight and separated from the rest of the
airframe prior to the aircraft entering the water.
Furthermore,
it is clear that the right wing failed in downward bending, based principally
on the buckling of the main spar web and bottom panel skin in the area where
the lower spar cap was separated. It is important to note that the spar
structures show no indication of fore/aft bending; the forces applied at the
time the spars failed appear to have been strictly vertical. Given this
vertical load, the fracture at “F7”, diagram #1 is interesting
because it is located at the end of a 28 inch long segment of upper spar cap
(see detail in photos #2, #4 and #9). This cap section runs directly under some
major bulkheads in the nacelle as well as under the firewall assembly and the
cowling (see diagram #2). For it to lift out of the top of the wing as the
outer panel failed downward, yet remain true to the wing plane, would require
that the nacelle and cowling assemblies be absent. Were they in place, this
length of unreinforced spar cap would have encountered the nacelle and cowling,
thus causing major damage. No such damage is evident.
Consequently,
the nacelle structure departed the upper surface of the wing prior to the
wing’s failure in downward bending.
The
previously referenced Packer Engineering metallurgy report (attachment A)
identified several elongated rivet holes in the wing-to-nacelle skin
immediately aft of the upper spar cap. These holes are elongated upward and
aft. This indicates that the portion of the nacelle structure attached to this
piece of skin was lifted up and moved aft relative to the wing. All or most of
the nacelle would have separated with it.
So
it appears that the right engine/nacelle structure was torn upward and aft
relative to the wing, and that the right wing subsequently failed in the
downward direction.
Left
Wing
The
left wing failed in at least two locations. The outboard section, which extends
from the tip inboard to the leading edge failure at WS 221 (see photos #10 and
#11), includes the main spar failure between WS 216 and WS 221, the trailing
edge failure at WS 206, and the complete aileron (photo #12). While the wing
failed outboard of the inboard end of the aileron, the aileron itself remained
intact and separated with the wing section as a complete piece. This section
shows only moderate damage. The leading edge skin between WS 221 and WS 269 is
compressed upward and torn from the spar cap (photo #13). However, this area of
damage does not show evidence of a high-energy water impact; the exposed spar
cap is not at all peeled or deformed. The tip is mangled. The three hinges are
shifted aileron outboard. The root of the aileron is compression buckled as is
the outboard end of the outboard flap (see photos #14 and
#15).
The
inboard end of the aileron shows obvious deflection and deformation in the
downward direction (see photo #16).
The
left wing inboard section extends from the WS 210 inboard to and including the
main landing gear box with the gear still attached (see photo #17). The leading
edge has disintegrated back to the main spar line (photo #18). A section of the
main spar web is present, extending from about WS 93 outboard to approximately
WS 165, and considerable deformation is present. Both spar caps have separated
from this web section. Outboard of this web section, extending to the outboard
fracture at WS 210, no spar structure at all is evident.
The
left wing main spar failed between WS 216 at the lower cap (“F2”,
diagram #1) and WS 221 at the upper cap (“Fl”, diagram #1, photo
#18). The lower spar cap fracture at WS 216 mates perfectly with the left end
of the recovered lower cap section (see “Packer” photo, attachment
A).
The upper cap is fractured with no mate recovered. The upper cap fracture site
is about five inches outboard of the lower cap fracture site, with the
adjacent, uncapped web bowed forward.
It
is interesting that each end of the surviving 35 foot section of the lower main
spar cap mates with a surviving wing structure. In the area of the
“F4” and
“F5”
site
in diagram #1, the spar cap was pulled from the bottom of the wing back to the
failure line just outboard of the nacelle. At the “F2” site, the
lower cap failure corresponds to the failure line of the left wing outer paneL
Both wing failures were in downward bending.
Furthermore,
it is interesting to note the difference in the condition of the left wing
outer panel and the inner panel (see detail, photo #9). The leading edge of the
inner wing section is totally destroyed, with a few sections completely
missing. The outer panel, however, is damaged but essentially intact. It does
not appear to exhibit any high-energy “hydraulic” damage, that is
damage incurred by the action of the panel forcibly striking the surface of the
water. Indeed, the upward crush of the leading edge between stations 221 and
269 opens up a perfect scoop along the spar line. Yet the spar cap and skin in
this area show no signs of peeling, rolling, or other
hydraulic
damage (see detail of photo #13). So it is highly unlikely that this section
could have entered the water at a high speed, yet highly probable that the
inner wing section did.
At
Fl of diagram #1, the left wing failed in down bending. The aileron stayed with
the outer panel and is complete, which might not be expected in an aft bending
separation. There is considerable compression of the trailing inboard tip of
the aileron, as well as the leading inboard corner of the wing panel (details
of photo #10 and photo #11). These may have resulted from collision during or
after separation. Most interesting, however, is the section of skin overlapping
the lower spar cap flange just inboard of “F2”, diagram #1. The
rivet holes are dimpled but not elongated, showing tension pullout. The skin
section is not rolled back, however, and although wrinkled, still lies along
the plane of the spar cap. There has been no fore or aft movement during the
separation between the lower spar cap and the wing skin in this area, only a
very high energy tension “pullout” failure.
In
contrast, the inner panel is severely damaged (see photo #9 and photo #18
detail). The leading edge is gone. The upper spar cap is gone. The lower spar
cap is part of the 35 foot section recovered intact but separated from the web.
A section of spar web remains but is bent fore and. aft. The nacelle is gone.
This wing section was probably still associated with the airframe after the
major breakup and later, upon impact with the water. The section of skin flange
overlapping the lower spar cap and adjacent to the outer panel skin is visible.
Here, however, the flange is rolled up and back, and the rivet holes show
tearing and elongation, all of which are indicative of “hydraulic”
or water impact damage.
In
summary, then, the outboard section of the left wing failed in downward bending
prior to the airframe’s impact with the water. The inboard section of the
wing remained attached to the fuselage throughout the breakup and subsequent
impact with the water.
The
empennage
The
empennage separated from the fuselage at the junction of the vertical fin and
the fuselage. The dorsal was not attached; skin damage was present up to canted
SS
11.5,
where
the dorsal would have attached (see photos #19 and #20).
The
right horizontal stabilizer was detached from the aircraft, and was never
recovered (see diagram #3). Approximately 9.5 inches of the forward spar
structure, however, remains attached to the vertical stabilizer. This structure
is bent aft nearly 90 degrees against the vertical fin (“F13 diagram #3,
photo #21). Approximately 22 inches of the aft spar structure remains attached,
and is bent aft 10 to 15 degrees from the normal sweep of the horizontal
stabilizer (“F14”, diagram #3, photo #21).
This
bending of the remaining spar stubs indicates an aft bending failure of the
right horizontal stabilizer. There is no evidence of any vertical axis
deformation of these stubs, and the failure appears to have been completely
longitudinal.
The
outboard portion of the left stabilizer was recovered, extending from
approximately HSS 40
to
the tip. The inboard fracture site area shows compression buckling at the
trailing edge (“Fl5”, diagram #3). The two hinges present show
severe displacement elevator outboard (photo #22).
The
left horizontal stabilizer includes three significant features. The first is
that it was physically sawed, by a member of the wreckage recovery team, at
approximately HSS 55 (“C 11”, diagram #2). Secondly, the leading
edge is characterized by two distinct impact marks. A large circumference
depression is centered just inboard of the station 35 rib. This impact caused
the leading edge to roll downward and left. A second depression is located at
HSS 47, which is a knife-edge indentation approximately six inches long and one
inch in width. It is rounded at the bottom, creating a very straight trough.
Associated with this trough are several compression buckles in the skin on both
the upper and lower surfaces of the stabilizer. Beginning at HSS 20 and
extending outboard to this trough, the stabilizer forward spar is bent back
approximately eight to nine degrees beyond the nominal (see photos #23, #24,
#25, #26 and #27).
The
saw line comprises approximately 9 inches of the trough mark. It is now
impossible to determine to what extent the stabilizer was damaged prior to the
cut. However, the leading edge skin between the strike mark and the severance
line is pulled away from the rivets. Several rivet holes are torn in the aft
direction. At the adjacent leading edge site of the outboard section, the skin
is in place with no separation. Further, the photographs taken at Quonset
immediately after recovery show the outer panel of the stabilizer bent
substantially downward, not upward as stated in the NTSB accident report (see
photos #28 and #29).
Additionally,
photo #30 shows a segment of the surface deicing bleed air line mounted in the
leading edge of the left horizontal stabilizer. The bleed air line is cut
through at the span wise location coincident with the cut in the left
horizontal stabilizer. Three additional, smaller cuts are also on this bleed
air line.
The
vertical stabilizer forward spar is twisted 30 degrees clockwise looking from
the top down. Virtually all of this twist occurs prior to SS 11.5 (see photos
#19 and #31). This damage appears to be associated with an impact on the right
side of the spar just below the SS 11.5.
The
forward spar appears to be generally straight above SS 11.5, but the vertical
stabilizer aft spar is considerably deformed along its entire length. While the
left side of this spar structure is virtually straight, the right cap has a
wave set along its length (photos #20 and #32). The web areas that are visible
show this wave to be the result of a severe clockwise twist looking from the
top down.
The
empennage “T-tail” assembly nose cone has a lightweight,
aerodynamic fiberglass shell which protrudes forward at the very top of the
tail (diagram #2). It is located immediately forward of and adjacent to both
horizontal stabilizers, and is physically secured to the top of the vertical
stabilizer. This shell was intact, attached in its normal position, and with
minimal damage at the time of recovery and subsequent retrieval from the ocean
(see photos #23, #31, and #32). Had the previously documented damage to the
horizontal and vertical stabilizers occurred during impact with the water,
particularly with the aircraft impact attitude assumed by the NTSB, this highly
frangible nose cone would have been destroyed. The nose cone’s total lack
of damage indicates that all other “T-tail” components received
their damage in flight,
prior
to, and not because of, water impact.
The
complete severance of the right horizontal stabilizer represents a fairly
high-energy event occurring at the trailing structure. The energy stored by the
airframe during water impact would have been expended by the time the tail
arrived at the surface, indicating that the damage to the stabilizer occurred
prior to the aircraft striking the surface of the water.
In
order to produce the observed wave set across the entire length of the vertical
stabilizer spar, the two forces involved must be applied at opposite ends of
the spar length. The natural candidates for these forces would be the force
(impact) that separated the right stabilizer (top of the spar), and the
airframe resistance to yaw (at the bottom). Because the forward stabilizer spar
stub fractured and is no longer rigidly attached to the aircraft structure, the
twist that is seen in the aft spar would not be evident in the forward spar.
The aft stubs are still attached to the vertical stabilizer, and would have to
have absorbed the entire twisting force, thus “setting” the aft
spar in the position observed.
While
loss of the right stabilizer was clean and complete, the loss of the left
stabilizer must be considered in terms of its failure as a lifting structure.
Had it remained intact, it is possible that it may have supported the airframe
nose aerodynamically (SA Brasilia at Gadsden). However, if the damage to the
leading edge disrupted the airflow over the surface of the airfoil, as the
existing damage would have, then the stabilizer’s functionality as an
airfoil is reduced to some degree.
The
aerodynamic effects of a distortion of the vertical stabilizer are
unpredictable, but easily disastrous. In conjunction with the loss of the right
stabilizer and damage to the left stabilizer, the airplane at that point was
uncontrollable.
Given
all of this structural evidence in the empennage, therefore, the following must
be concluded; that the right stabilizer was struck by something with enough
energy to completely separate it from the tail. Considering the wave set that
was imparted to the vertical stabilizer, this impact had to have occurred prior
to the empennage separating from the rest of the airframe. Had it occurred
after empennage separation, it would simply have implied an angular
acceleration to the entire tail, and just spun it around.
Remembering,
then, that both wings failed in downward bending in flight, it is most logical
to conclude that the right horizontal stabilizer was forcibly struck and
subsequently separated from the aircraft in flight. This caused a dramatic
reduction in necessary negative lift at the tail of the aircraft, and therefore
a severe nose down pitch tendency. Additionally, the partial loss of airflow
over the surface of the left stabilizer, in conjunction with the distortion
imparted to the vertical stabilizer and the severe nose down trim imparted to
the elevator trim system (see documentation and discussion under
“Elevator Trim System” page 13 of this report), rendered the
aircraft uncontrollable. The extreme nose down pitch tendency caused an
instantaneous catastrophic failure of the outboard wing panels in downward
bending.
Lower
Main
Spar
Cap
A
major portion of the lower main spar cap was recovered. At time of recovery
from the ocean, it was one long continuous piece. For ease of storage, however,
this piece was cut in two (see detail in photo #9). The right hand piece is
approximately 19.5 feet in length; the left hand piece is approximately 16 feet
in length. The right end begins at the “F3” site, diagram #1. The
left end of the lower cap section begins at the “F6” site, diagram
#1. The fracture at the “Fl” site and the fracture at the
“F6” site appear to be brittle type fractures with little plastic
deformation visible around the fracture site. A third fracture is located
approximately 9 feet from the left end (“F9”, diagram #1, photos
#33, #34 and #3S). This fracture occurred in the downward direction, as in wing
down bending. This fracture would have been located close to WS 108,
approximately in the center of the left nacelle structure. Notice the darker
area on the cap in photo # 33 — this indicates that portion of the spar
cap directly under the nacelle. This fracture shows considerable deformation
and lipping on either side of the fracture line. Indeed, the spar cap at this
location is held together only by the intact segment of the skin.
Also
of significance is a distinct scrape mark on the lower surface of the right
lower spar cap, at approximately WS 12S (photos #36 and #37).
Right
Main
Landing
Gear
The
right main landing gear wheel well and box support structure separated from the
rest of the aircraft (see photo #38). The right main gear upper strut support
assembly freely rotates around its pivot point, making it impossible to
determine the gear upper strut position at time of water impact.
The
right main landing gear oleo strut is fractured just above the wheel truck
(detail of photo #38, photos #39 and #40). The oleo is partially collapsed
above the fracture side, on the forward side of the strut. A triangular scrape
mark indicates a sliding impact. The fracture has occurred at the point at
which the internal structure of the strut offers the least resistance. This
fracture was the direct result of the lower main spar cap separating from the
aircraft structure in a severe and rapid downward direction. The relative
locations of the right main landing gear oleo strut, while in the “up and
locked”, retracted position, and the adjacent position of the lower spar
cap, are shown in diagram #2. The distinct witness mark on the lower spar cap
(photos #36 and #37) is at the point at which it severed the right main landing
gear strut. Note that the mark has a width corresponding exactly with the width
of the strut. Note also that photo #36 which shows the unpainted area of the
spar cap that passes through the nacelle/wheel well area. The scrape mark and
deflection are exactly centered in that area.
The
right main landing gear wheels, tires, brake assembly and lower oleo strut
separated as a complete unit (see detail of photo #9 and photo #42).
In
photo #43, the right main gear assembly has been placed in its relative
position to the outboard panel of the right wing. The spar cap area (red in
photo) lines up exactly with the fracture in the landing gear oleo strut. See
also diagram #2.
Therefore,
the right main landing gear was in the up and hydraulically
“locked” position (see diagram #3) at the time the lower spar cap,
right nacelle and engine, and a major portion of the right wing separated
violently from the aircraft.
Nose
Landing
Gear
The
nose landing gear drag brace is fractured (photos #44 and #4S). Lips on both
ends of the fracture face and uniform bending of the entire segment indicate a
column buckling fracture. The degree of deformation on either side of the
fracture line is considerable (photo #46).
The
forces required for this failure mode are exclusively from forward to aft,
causing compression overload failures of the drag brace (photo #47) and the
mechanical down lock (see diagram
#4).
Therefore,
at the time of impact with the water, the nose landing gear was down and
mechanically locked. This was also the conclusion of the NTSB 10 of the
report.
Of
obvious
significance, then, is the fact that two separate and distinct events occur
relative to the position of the landing gear; the main landing gear was up and
locked (hydraulically) during the inflight breakup (first event), yet the nose
landing gear was down and locked at the time of impact with the water (second
event).
The
main and nose landing gears are held in the up (locked) position with normal
hydraulic pressure through a motor/pump system physically located in the left
wing root, immediately
forward of the main spar.
[5]
During an in-flight breakup of the type encountered here, hydraulic pressure
would be available to initially keep the gear retracted. However, as electrical
power to the 28v motor/pump became unavailable due to numerous electrical
overloads, shorts and failures, the only source of normal hydraulic pressure
would be lost. Additionally, ruptures and failures of numerous hydraulic lines
located in the wheel well and wing spar areas would cause the rapid loss of
remaining pressure. That total loss of normal hydraulic pressure would then
allow all gear to free fall to the down and mechanically locked position.
The
right main gear, however, would by this time have been cut in half, and the
response of the short stub of shock strut still attached to the airframe would
be unpredictable. The left main gear position would be influenced by the amount
and type of deformation caused in the wheel well area by the separation of the
outboard portion of the left wing, and portions of the lower spar cap. Indeed,
there are indications that the left main gear oleo strut may have been bisected
by the rapid “pullout” of the lower spar cap, exactly as happened
to the right main gear (see photo #17, and compare with photo #43). Therefore,
whether the left main gear would fall completely out of the wheel well, and
what position it may have had at impact, is unknown.
However,
at some point in the uncontrolled descent and prior to impact with the water,
we would expect the nose gear to free fall out of its well to a “drag leg
overcenter” or down and locked position. This is exactly what happened,
and has been thoroughly documented.
Elevator
Trim
Elevator
trim tabs are installed on both elevators. The tabs can be manually controlled
by the pilot through drum-cable systems using jackscrew actuators. Moving the
cockpit elevator tab control wheel forward results in tab deflection up causing
elevator movement
down,
and aircraft nose down. The amount of wheel and tab movement is shown by
reference to a geared position indicator located immediately adjacent to the
elevator trim wheel on the cockpit pedestal (see both pages of diagram
#5).
The
elevator trim tab control cables travel under the left side of the fuselage
floor, through pulleys in the fuselage and vertical stabilizer to the tabs
located in each elevator (see diagram
#5).
These
cables are fitted with turnbuckles in various locations. One such location is
on top of and to either side of the vertical stabilizer, accessible through
inspection covers on the top surfaces of the horizontal stabilizers (see
detail, diagram #2). These cables and their respective turnbuckles have an
operating range (movement along the normal left and right axis of the aircraft,
and along the fore/aft axis of the cable) of approximately 8 inches.
Photo
#48 shows the pedestal area. Notice the white index mark visible immediately to
the right of the trim wheel. Photo #49 is taken from directly above, and is
only to document the position of the “nose down” reference beside
the trim wheel. The position indicator is geared directly to the elevator trim
control wheel. As is evident in this photo, the indicator has been forcibly
driven forward, or in the “nose down” trim direction to a point
well beyond the normal operating range of the mechanism, but it’s
integral drive gearing with the trim control wheel is still intact.
Photos
#50,
#51
and
#52
document
the elevator trim control cables as found on the accident aircraft. The cable
shown in photos
#50
and
#51
is
the “crossover” cable and turnbuckle, located out in the
stabilizer. But note particularly the position of the “nose down”
or ‘forward” turnbuckle (see diagram
#5,
photo
#52).
Because
of pulling action exerted on the right side of the crossover” cable, the
“nose down” cable has been pulled in the opposite direction, from
it’s normal location inside the right horizontal stabilizer to a position
inside the vertical stabilizer, approximately 24 inches away! This would
correspond to an elevator trim position far forward of full nose down, which
was the post-accident position of the indicator on the cockpit pedestal (photo
#49). The cable has had its travel stopped only by the turnbuckle, which is
totally inflexible, trying to move around the pulley located in the vertical
stabilizer! Once the “nose down” cable had been completely fouled
in the pulley, the “crossover” cable was prevented from being
pulled any further, and failed outboard of the vertical stabilizer in tensile
overload.
As
stated earlier, ALPA believes that the damage observed to the nose and left
wing root area of the aircraft was caused by initial water impact. Because of
normal crash “kinematics”, then, impact damage would occur to the
cockpit area prior to the rest of the aircraft, and certainly before the tail,
which would have entered the water at a later time than the nose. Therefore, it
is evident that the cockpit trim control wheel and indicator have been
“back driven” by the cable in the tail. Obviously, the force that
would be required to cause movement of the cable along the entire length of the
aircraft
would
have to have originated at the tail
.
A
logical conclusion
,
then,
is that the event causing the “back driving” of the trim system
occurred at altitude, prior to the inflight breakup of the aircraft, and long
before water impact.
Therefore,
the sequence of events would have been:
An
event at the tail of the aircraft causing extremely rapid movement of the
elevator
pitch
trim system towards the “nose down” position,
Resultant
movement of the elevator trim control wheel and trim indicator in the
cockpit
towards the “nose down” position, only stopping well beyond the
full nose down position,
At
some later time, cockpit impact with the water.
Right
Engine
Truss
A
significant portion of the right engine mounting truss arrived at Pratt &
Whitney with the right engine gas generator. This assembly consisted of the
hoop structure with the right side V mount still attached (photo #S3).
Examination of this structure revealed that of the eight truss tubes attached
to the hoop, five had failed at or very near the hoop junctions. A sixth had
failed at the hoop junction as well as near the mounting boss. The remaining
two, both forward ends of the right side V structure, had not failed.
The
aft mounting boss of the right side V structure was present. The truss tube
junctions at this boss were intact. The mounting boss itself was closely
examined, and the bolt hole itself shows no obvious signs of distress, although
there is a scratch visible on the aft side of the boss extending
circumferentially about 180 degrees around the bolt hole (photo
#54).
Additionally,
the V structure is bowed noticeably in the vertical plane. This bow brings the
aft mounting boss at the tail of the V inboard relative to the hoop structure
(photo
#53).
The
engine truss structures of the B 1900 have a long history of cracking, often to
the point of separation. In 1989 alone, Business Express filed over 70 Service
Difficulty Reports with the FAA concerning truss cracks. Other operators have
experienced similar difficulties. In September and October of 1991, the
“Beech 1900 Operators Committee” was so concerned with this problem
that they twice requested action on the part of Beechcraft to remedy this
“very serious situation.. and known failure condition” (see
attachment C).
But
Beechcraft was already aware of the seriousness of the problem, as evidenced in
their own summary of the situation. “Numerous instances of fatigue cracks
in the Model 1900/1900C engine truss including several instances of complete
truss tube separation indicates the potential for degradation of the
truss’ load carrying capability.”
[6] Beechcraft
issued Service Bulletin (SB) 2196 on September 1, 1987, proposing inspections
and repair or replacement of cracked truss tubes. Two and a half years later, a
mandatory service bulletin (2255) was issued in an attempt to remedy ongoing
structural engine mount failures (see chronological sequence of regulatory
events, attachment F).
A
detailed and comprehensive reliability study
[7] of
this problem, completed in June, 1995, is included in its entirety as
attachment D.
This
report documents 10 years of extensive in service difficulties with
all
models of B 1900 engine support trusses. Most of the problems have to do with
cracking truss tubes, including those resulting in complete separation of the
tubes. As early as 1989 the FAA was been aware of the problem, as evidenced by
internal memorandums (see attachment D, page 16). In that document, the FAA
representative states that because of the engine truss tube structural
problems, “the airworthiness of the aircraft will remain in question.
This inspector believes this is a very serious safety problem”
[8] .
Airworthiness Directive (AD) 9 1-12-02 requiring initial and repetitive
inspections of the engine trusses and installation of reinforcing doublers was
issued May 1S, 1991. A second AD was proposed in November of 1991, and became
law three months after the crash of N811BE (both Airworthiness Directives are
included in attachment C). Since that time there have been at least three
additional AD amendments or revisions and four mandatory Service Bulletin
revisions, all pertaining to engine truss tube cracking (see table, attachment
F).
237
events of truss tube cracking, separations and failures are documented between
1985 and 1997’
[9]
which generated Service Difficulty Reports (DR) (see attachment D, page 17). To
date six different truss types have been designed in an attempt to correct the
ongoing problem. They vary in tube size, design and thickness, but the
increasing mass of the support system has actually decreased the reliability
and in-service life expectancy of the truss (see attachment D, pages 18-38).
It
is interesting to note that all of the six truss tube failures on the accident
aircraft right engine truss occurred at locations that are cited by Beech in
four Service Bulletins and two Airworthiness Directives as areas requiring
increased inspection.
Please see
Compact Disc enclosed (inside
backcover)
for
additional engine truss
information.
"Whirl
Mode Flutter"
The
following is excerpted from the previously referenced study by Dr. R..
Stearman, attachment D, pages 9- 11.
“Whirl
flutter is the onset of unstable and destructive oscillations usually involving
a lifting surface and propeller disk in an airstream. Whirl flutter is a
precession-type instability that is the result of the coupling of gyroscopic
and aerodynamic forces acting on the propeller. These forces can cause the
pitch and yaw degrees of freedom to couple, yielding a whirl mode. Gyroscopic
forces can couple the pitch and yaw modes of vibration exhibited in a rotating
propeller mounted on a flexible support or structure. This coupling results in
one of the following two modes:
forward
whirl mode, where the direction of the cyclic precession is the same as the
propeller rotation, and
backward
whirl mode, where the precession is in the opposite direction of the
propeller
rotation.”.
The report continues with a detailed discussion of the whirl
theory, but concludes that section with “Thus whirl flutter occurs when
the aerodynamic forces provide the coupling necessary to induce an unstable
whirl motion. Gyroscopic forces tend to destabilize the backward mode further.
This phenomenon explains why whirl flutter invariably occurs in the backward
mode.
The
report concludes (see attachment D, pages
49-52):
“Several
findings in this investigation suggest that the occurrence of a highly
divergent whirl flutter instability led to the destruction of Beech 1900 UIB-49
during the pilot training mission. Some of the more significant findings
leading to this conclusion are presented in the following listing:”
- ‘The
statistical reliability study carried out in this investigation indicates a
serious fatigue cracking phenomenon within the engine mount truss elements that
is getting worse as attempts are made to improve the design. Past experience
has shown that damaged engine mount trusses are a known contributor leading to
propeller whirl flutter instability of turboprop aircraft.”
- ‘The
engine truss to firewall mounting bolts on UIB-49 could have encountered a
foreseeable and significant pre-stress at the time of the truss installation or
replacement due to the manufacturing quality control problem illustrated in
Figure 3-14. The extent of this bolt and engine truss pre-stress build up will
depend upon how the out of tolerance dimensions add up, but will generally, for
a given misalignment, be the greatest for the stiffer or newer truss designs.
This pre-stress condition will accelerate both fatigue and overload failures.
Pre-stressing due to misalignment is not an uncommon occurrence, according to
Beech records and a third party informant. This informant suggested that
Conquest Airlines of Austin, Texas found it necessary to force fit every
replacement engine truss that was installed on the Model 1900C aircraft”.
- “Further
inspection of the right engine truss revealed that it had damage and tube
separation in areas where service difficulty reports indicate fatigue cracking
regularly occurred...”
Additionally,
both MSC/Nastran 6810
[10]
and Brock’s
[11] flutter
analyses were conducted for this aircraft, engine and propeller configuration.
A detailed description of the two studies is included in the report,
“Aircraft Damage Detection from Acoustic and Noise Impressed Signals
Found by
a
Cockpit Voice Recorder”
[12] by
Stearman et al, attachment E.
As
documented, when truss cracking occurs in two or more tubes, the whirl flutter
speed lies well within the aircraft flight envelope. At 1550 RPM (flight idle),
the whirl flutter speed with two broken tubes is precisely in the airspeed
range (180k - 190k) in which the accident aircraft was operating.
An
independent study of the engine mounting systems including truss tubes,
attaching bolts, nacelle bulkheads, etc. of N811BE was completed subsequent to
the accident.
[13] This
summary document is included in its entirety as attachment F. This study
describes the failure mechanism and separation sequence of the right engine
mounting truss. As it concludes:
‘The
right engine mount failed in flight, allowing the engine and engine mount
structure to separate from the wing, followed by an in-flight right wing
separation. This opinion is based on visual examination of the right wing mount
structure and the right wing condition.”
‘The
failures in the right engine mount truss structure are consistent with
propeller whirl mode type loading and the significant Beechcraft 1900 engine
mount service failure history has given ample warning of potential engine mount
failure and separation as in this accident.”
Please
see enclosed Compact Disc inside back cover for description and demonstration
of “whirl mode flutter”
Engines
The
official NTSB accident report states that “Both engines displayed scoring
from internal rotating parts indicating that the engines were developing power
when the airplane struck the water. The scoring was not extensive, and no
estimate could be made about the amount of power the engines were developing.
However, since one of the last comments on the CVR was
‘power
to idle’, followed by sounds of the landing gear warning horn, it appears
that the power lever on the left engine was reduced to match the power from the
right engine, which was at a flight idle power setting to simulate its
failure.” These statements are in error.
In
fact, to simulate the engine failure, the right engine would
not
have been at flight idle,
but
at a reduced
power
setting to simulate zero thrust
.
Therefore, a number of situations could cause the gear warning horn to sound,
including retarding
either
the left
or
the
right power lever with the gear not down and locked. It is ALPA’s belief,
based on discussions later in this report, that the cause of the gear warning
horn was, in fact, due to the retarding of the right throttle from the
“zero thrust” position to the flight idle position. Furthermore,
the actual condition of the
engines
as found during a very detailed examination done in June of
1995
[14]
(see the Hamill report, attachment G) indicates the following:
* “The
left hand engine was developing power at impact as evidenced by the wear
marks
on the rotational parts. This wear and especially heat discoloration on the
impeller
and shroud are indicative of power above low idle even with contact with
water
which has less impact force than solid matter such as the ground or trees.”
* “The
right hand engine was not developing any power at all (emphasis added). The
condition of the rotating parts, especially the impeller and shroud, show no
signs of wear that can be attributed to the impact force. The impeller and
shroud have such a small clearance between them that even low idle power and
impact with water would show more than is seen in this instance. I believe that
the rotation of the right hand engine was somewhere below low idle at the time
of impact. This condition would exist if fuel were not being delivered to the
engine for a short period of time prior to impact.”
The
summary section of the Pratt & Whitney engine teardown report
[15]
states that;
“The
left hand engine and right hand engine gas generator section displayed
rotational signatures to the engine internal components characteristic of the
engines developing power at impact. The minimal impact deformation of the
engine cases limit the severity of the rotational signatures and precludes
definitive assessment of the power levels and impact.”
Additional
review of the above referenced report shows that the left engine centrifugal
impeller and impeller shroud were “circuniferentially rubbed, with
frictional heat discoloration and material transfer, due to axial contact with
each other”. The right engine impeller and shroud were only
“lightly circumferentially rubbed”, with no heat discoloration or
material transfer at all.
The
damage patterns on the gas generator sections show only a positive indication
of rotation, whereas the difference in impeller damage between the two engines
indicates a difference in power. It is ALPA’s conclusion, then, that the
evidence indicates that the left engine was developing power at time of impact,
whereas the right engine was
not
developing power, but only rotating.
The
NTSB goes on to say, ‘The right engine was ... recovered, but during the
transfer from the water to the salvage barge, the forward part of the right
engine including propeller hub, reduction gears and exhaust casing separated
and sank back into the ocean. These latter components were not
recovered.” Why the NTSB investigator chose not to recover those portions
of the right engine, when their exact location was known, while the recovery
team and equipment were m place and in relatively shallow water, is not known.
CVR
Analysis
The
NTSB provided a transcript of the CVR tape to the investigative team, but at no
time was a
spectroanalysis
study of the recording done by the Safety Board. The official report, page 20,
states:
‘The
Safety Board cannot conclusively exclude the possibility of an event that
caused premature termination of CVR operation before the plane struck the water
because the CVR events could not be precisely coordinated with the position and
altitude as recorded by ATC radar, nor was there any FDR information available
to establish the airplane’s actual performance during it’s final
descent... .The Board believes that any event that would have caused
termination of the CVR must have been sudden and probably catastrophic, which
leads to the conclusion that the event was a high speed collision with the
surface of the ocean," ALPA
agrees with the Board that the event causing the CVR to terminate must have
been sudden and catastrophic. But events other than water impact could easily
cause cessation of recording.
Immediately
prior to termination of the CVR, as documented by the official transcript and
all subsequent CVR studies, there were no recorded acoustic sounds of impact,
inflight breakup or vocalization of the pilots indicating awareness of
impending disaster. As we will discuss shortly however, there were at least two
non-acoustic structural “events” recorded on the CVR, that were
detected only by detailed spectroanalysis..
G
Limiting Switch
The
B&D recorder installed on the accident aircraft had a factory installed, 5g
limiting switch, which automatically terminates recording with the sensing of a
high “g” load upon the switch. This switch prevents inadvertent
continual recording by the CVR after an incident or accident, which could erase
important data. Any acceleration to the switch of 5g of more, whether imparted
to the airframe during the accident sequence or directly to the empennage area
(where the CVR is housed) would cause termination of the recording.
CVR
”Events”
Several
spectroanalysis studies have been conducted on the accident CVR tape subsequent
to the Board’s own investigation. The first, the Stearman report
(attachment D), labels two very serious events that are detected at the end of
the tape, ‘spike 1’ and ‘spike 2’. To quote
,
“the maximum peaks of the events are approximately 0.263 seconds apart
... The first major spike is preceded by a highly divergent signal of the type
present during whirl flutter” (emphasis added). The first spike is
immediately preceded by a strong acoustical signature, indicative of a very
violent, “explosive” event.
The record head shutoff transient occurs at the first spike, while the erase
head shutoff transient is evident during the second spike.
A
power
spectrum analysis of the first spike indicates a frequency of approximately 36
Hz. In tests done by Peter Zwillenberg for Beech Aircraft Corporation in
Wichita, Kansas
[16] ,
the wing torsion asymmetric vibration mode for the engine in vertical and
lateral translation is 37.6 Hz.
ALPA
believes
the similarity of the frequency on the CVR and the asymmetric vibration test
data to be very significant. If the engine were to “translate”, or
tear itself out of the wing, we would expect to see frequencies in the range of
36-37 Hz.
That
same report details the extensive time-series and auto power spectrum analysis
that was also conducted. The extensive testing procedures are described in
attachment D. To quote only the summary sections,
.“
. .we suspect that the drastically reduced acoustic power signal associated
with the four-blade passage frequency of 104 Hz to 112 Hz
(1550
to
1700 rpm) at the end of the tape indicates the loss of about one-half of the
source of the signal sound power level - that is, the probable loss of one
engine. Regardless of the resolution setting of the analyzer program, (4 or 8
Hz), the results indicated a loss of one-half of the sound power signal. The
impact of the engine into the tail would most likely impose an acceleration on
the fuselage tail cone structure exceeding the
5 G (five times the force of gravity) threshold required by the CVR for automatic
shutoff If the CVR was automatically shut off by an impact with the water, it
is likely that there would be noticeable sound of impact (based on the accepted
impact scenario) and that both engines would have continued to produce power
with the accompanying typical sound power level until impact.”
Finally:
“The
CVR tape provided some additional strong evidence that a catastrophic event
occurred within three to four seconds after the right engine was placed into a
flight idle condition. During this operation, the thrust on the right engine
would drop to zero and could probably set up a negative thrust or drag
configuration on the right propeller disk plane. In [a report for NASA,] ’
[17] Reed
demonstrates the stabilizing influence that positive thrust has on the whirl
flutter phenomenon. Conversely, a system which might be stable during cruise of
normal operating thrust conditions could become unstable when the thrust drops
to zero or becomes a negative thrust or drag condition. It is interesting to
note that between three to four seconds after a right engine flight idle
condition was induced, [the highly divergent acoustical signal is] seen to
occur on the
CVR
tape.
This would be characteristic of the aerodynamic lag time for a zero or negative
thrusting condition to be induced on the right engine. . . Finally, some of the
most convincing evidence on the
CVR
tape
is the loss of 50 % of the engine-propeller sound power level after the two
catastrophic events...This clearly indicates that half of the engine-propeller
noise source is missing, implying that [an] engine has departed the aircraft.
One final factor that is evident from the CVR tape is the fact that the
page 20
catastrophic
event was so sudden that no one in the cockpit had time to vocally respond. No
expletives or other critical comments were noted on the tape. This further
suggests a flutter event which is usually an explosive type of
phenomenon” (emphasis added).
CVR
“Flutter
”
A
second, very detailed spectroanalysis study was conducted by Glen Schuize
[18] early
in 1996 (attachment H). This study isolated and identified a distinct and
unique ‘flutter” during the fifth and final beep of the landing
gear warning horn. To quote,
“Figure
30 displays the time-series of a tape track recorded with acoustical voice
signals superimposed with the
5
landing
gear warning horn signal beeps found right at the end of tape. Amplitude
erosion can be seen of the 5th and last horn beep that is not found on the
preceding beeps. An attentive listening to these last
5
beeps
also revealed a brief but definite audible flutter detected only on the 5th
beep.”
“Figure
31 was obtained at the 5th beep of this horn signal 19 minutes earlier in the
tape. No amplitude erosion or audible flutter was found indicating the horn
signal source did not suffer fatigue with this length of operational time.”
In
Stearman’s second report, attachment E, he summarizes:
“Just
seconds before the EOT (end of tape), a significant FM modulation was detected
both audibly and with the aid of a spectrum analyzer. The modulation was at the
propeller fundamental rotation frequency and was due to the dynamic mass
unbalance generated by a rotating propeller as it tore loose from its mounting
system”.
Please
see the Contact Disk enclosed inside back cover for digital CVR recordings of
the “flutter”
CVR
“Silent” Tracks
One
of the four CVR tracks on this aircraft was unused, or “silent”,
with no microphone attached.
This
track was found to be rich in non-speech sounds. This process is called
electroacoustic transduction, and can include “triboelectric”,
"magneto-electric” and “piezoelectric” effects. As Stearman
states (attachment E):
“Close
inspection of the amplified (30 dB) times series from the CVR silent track also
revealed a very periodic set of transient components occurring at a frequency
of 0.86 Hz. Furthermore, this frequency correlated with an independent
structural dynamic and flutter analysis of the engine mount damage, which is
evident from Figure 3. This transient frequency was found throughout the 32
minutes of this
CVR
tape
indicating the
condition
that generated this phenomenon was an ongoing long-term condition rather than a
rapid onset event. These transient components are thought to be typical of
impact signatures that would occur from a broken tube end impacting on the tube
joint where the fracture occurred. These transients became more frequent about
15 seconds before the airframe disintegrated and then diminished to nearly zero
at the end of the flight where the CVR suddenly stopped (emphasis added). [The
diagram]
End
of tape time series,
shows
the presence of these 0.86 transients which were demonstrated by independent
structural and flutter analyses to be quite close to the frequency experienced
by a damaged engine mount.”
Please
see Compact Disc enclosed inside back cover for digital demonstration of the
“triboelectric” effect of electroacoustical transduction.
To
summarize then, the CVR data which was never researched by the NTSB, hence not
available for inclusion in the official report of this accident:
- There
are two shutoff spikes recorded on the CVR, approximately .263 seconds apart,
at the end of the recording. There is a highly divergent, whirl flutter type
signal immediately prior to the first violent event, detectable both
acoustically and analytically.
- The
frequency of vibration of the acoustical event prior to the first spike
corresponds to the frequency previously found to be the wing torsion asymmetric
vibration mode for the engine in vertical and lateral translation. In other
words, the structural frequency detected at the CVR is what would be present
had the engine been in the process of separating from the wing.
- Half
of the
acoustic
signal associated with the four-blade passage was lost at the end of the tape.
Simply put, the CVR detected only one engine/propeller combination producing
power after the first violent event recorded on the tape. At that point in
time, it can be implied that one engine departed the aircraft.
- The
CVR terminated with a
5
G (or greater) load, with no sounds of aircraft impact with the water, due to
the loads imparted to the empennage by the right engine as it struck the right
horizontal stabilizer.
- An
intermittent amplitude modulation was found on the “silent” track,
which occurred over the entire thirty two minutes of the tape. Its frequency
correlated with the predicted frequency of a damaged engine mount vibration
just prior to and at the onset of the whirl flutter event. Thus, the CVR acted
as a latent transducer to not only confirm the whirl flutter event, but also to
warn of the existing engine mount damage at least 30 minutes prior to the
catastrophic event
Sequence
of
Events,
Based on CVR
Analysis
and Wreckage
Documentation
* Power
is reduced on right engine from a “zero thrust” setting to flight
idle. This produces the sound of the first gear warning horn “beep”.
* Whirl
flutter is induced in the right engine/propeller due to cracked or failed truss
tubes in the right engine mount truss tube assembly. Whirl mode flutter is
acoustically evident in the fifth gear warning “beep”. Pre-existing
truss tube failures are electroaccousticaily evident in CVR transient signals,
throughout 32 minutes of the tape.
* The
right engine and nacelle separate violently from right wing. First significant
“spike” on CVR is preceded by a violent acoustical event. Frequency
of vibration of this event is exactly that expected if engine were to separate
from the wing in flight (see diagram #6).
* The
right engine/propeller translates aft and strikes right and left stabilizer.
This causes at least a 5g force upon the tail and the termination of the CVR
recording.
* The
right engine completely removes right stabilizer from aircraft. Right propeller
impacts left stabilizer, causing the characteristic leading edge damage and
bleed air line cuts. Fiberglass nose cone at very top of T-tail is not damaged.
* As
the right horizontal stabilizer is torn from the aircraft, the right elevator
“nose down” trim cable is cut, and the “crossover”
cable is pulled violently and stretched. The “nose down” cable is
then pulled violently in the opposite direction until mechanically fouled in
the pulley. The cockpit trim wheel moves well beyond the nose down trim stop.
* As
the right horizontal stabilizer is torn from the aircraft, the tail twists
rapidly clockwise (as viewed from above). This twist is “set” into
the vertical stabilizer.
* With
the removal of the right horizontal stabilizer, the damage to the left
horizontal stabilizer and the extreme nose down trim, the aircraft pitches over
violently.
* Portions
of both wings fail instantly in downward bending. The right wing separates at
approximately WS 124, at the nacelle. The left wing fails at WS 211.
* The
outboard portion of the right wing, although separated from the fuselage,
remains attached (temporarily) to the lower spar cap. This cap then
“zippers” out of the bottom of the wing skin, completely bisecting
the right main landing gear shock strut.
* The
aircraft is now completely uncontrollable. All normal electrical and hydraulic
power is lost. The landing gear is no longer hydraulically held in the up
position. The nose gear free falls during descent, into the mechanically locked
down position.
* The
aircraft impacts the surface of the ocean, probably nose and left wing root
first, in a nearly vertical attitude.
Human
Performance
Training
Issues
A
detailed review of the CVR reveals that throughout the entire 32 minutes prior
to cessation of the recording, the flight is conducted with the highest degree
of professionalism. In that time, there are numerous discussions of proper
basic flying technique, instrument approach procedures and emergency
“drills”, but not even one spurious or casual remark that did not
directly pertain to the training environment.
it
is important to remember that this is not a “check” ride for the
student, Mr. Lurie. Any evaluation being made by the instructor, Mr. Murphy, at
this point in the training scenario are only to identify those obstacles to
satisfactory performance that impede the student. Through demonstration,
communication and repetition (specific drills), the instructor attempts to
correct
the problem, not “pass” or “fail” the student.
Additionally, there are times when repeating and successfully completing
maneuvers that had been difficult for the student in the past instills
confidence in the trainee.
As
stated in the FAA’s “Aviation Instructor’s Handbook"
[19]
"..Those things most often repeated are best remembered. It is the basis of
practice and drill”, and “Every time practice occurs, learning
continues. The instructor must provide opportunities for students to practice
or repeat and must see that this process is directed toward a goal.”
Captain
Murphy is an extremely thorough but demanding instructor. Throughout the
flight, he maintains total control of the situation, while allowing the student
room to make those minor errors which are so essential for effective training.
At no time does the instructor ever let the training scenario go beyond the
reasonable limits of safety.
It
is evident that the instructor detects basic weaknesses in the performance of
the student, particularly in the areas of “partial panel”
instrument flying, single engine work and instrument approach procedures. There
are numerous discussions regarding these particular maneuvers recorded on the
CVR, and in fact, instructor Murphy has the student repeat these procedures
several times in order to strengthen the trainee’s abilities. Murphy does
load the student up with a lot of work, but on all occasions he gives the
trainee ample time to stabilize one situation (i.e., partial panel) before
introducing another (engine failure). Lurie is a “captain”
candidate, and as such, has to be able to handle difficult situations such as
those presented to him by his instructor.
One
FAA inspector, upon being interviewed after this accident, stated that if an
applicant for an ATP or type rating (which Lurie was) had lost his attitude
indicator during flight, he would have expected the applicant to be able to fly
partial panel without referring to the other attitude
indicator[20] Certainly
Captain Murphy was aware of this.
The
precise reasons the instructor created the scenario he did are unknown.
However, it would appear that it is an attempt, through repetition and
exercise, to correct basic student weaknesses. It may also be that because of
Captain Murphy’s knowledge of this particular aircraft’s systems,
he was attempting to simulate failures that, while seemingly unrelated, may in
fact be realistic because of the basic design of the aircraft’s systems.
In any case, this particular scenario did not technically introduce
“multiple emergencies” as defined by the FAA (see Board report,
page []), but did emphasize those areas of student weakness that were of
concern to instructor Murphy. Those same areas would ultimately need to be
demonstrated to an inspector during Lurie’ s upcoming check ride with the
FAA.
Altitude
Alert
Chime
The
altitude alert “chime” or tone is recorded on the CVR 8 times in
the final 32 minutes. This chime is activated
either
by
the aircraft climbing through or descending away from an assigned (and
“set”) altitude by 300 feet, or by resetting the altitude in the
alerter control panel. For example, if the aircraft is level at 3,000 feet, and
2,000 feet is then selected in the control panel, the altitude alert chime
would sound as the alerter is dialed through the 2700 foot setting towards the
2000 foot setting.
We
can assume that the pilot flying has the responsibility to set the controller,
as evidenced by Murphy’s comment on the CVR at 2120:37 “ah,
what’d you put in there?”, and Lurie’s remark, “I
didn’t, that’s just it, huh”. It is obvious in looking at the
entire conversation that both comments refer to the altitude selected in the
altitude controller.
On
at least three occasions, the alert chime sounds due to the setting of the
controller by one of the pilots. At 2117:39 “...ah, the magic number as I
said, is two thousand”. The aircraft at this time is at 2,500 feet. At
2117:46, the chime sounds. While reviewing the approach plate, Lurie sets the
2,000 foot initial altitude into the controller, causing the chime to sound.
Later,
during a different approach, there was a similar situation. Again, at 2141:16,
the comment “two thousand after I cross the VOR”. At 2141:44, the
altitude alert chime sounds. The aircraft is still at 2,500 feet. At 2142:14
the comment “out of two and a half for two thousand”. The altitude
alerter is set prior to, and in anticipation of the descent out of 2,500 feet,
which is initiated 30 seconds after the alerter is set.
Both
of these events occur during briefings regarding proper altitude awareness
during descent from one intermediate approach altitude to another. Both occur
approximately 30 seconds prior to planned aircraft descent.
The
last time the altitude alert chime sounds due to the flying pilot changing the
setting is the last chime recorded on the CVR, at 2146:34. Student Lurie has
had a tendency to descend out of the
page 26
initial
approach altitude early, as indicated in the prior two approaches flown that
evening. Both times he started descent to MDA before completely established on
the inbound final approach course of the non precision approach, and both times
instructor Murphy “coached” him in the proper approach technique.
At
2145:30, student Lurie asks, “what altitude I’m still good down to
now”. He is anticipating his descent to the MDA and the resetting of the
altitude controller. Instructor Murphy replies “ah, two thousand
still”. In other words, no descent yet, because the aircraft is still on
the initial approach segment. At 2146:27 Lurie again asks, “what altitude
am I good down to?” still anticipating the descent. Obviously the student
feels that he is close to intercepting the final approach course at this time,
probably within the same general 30 second time frame he has used on the
previous two approaches. Murphy responds “ah, once you’re
established inbound, right, you’re good down to what?” This
reinforces the procedure of waiting until interception of the course to
initiate descent, but also forces Lurie to think more about the correct
approach procedure, and perhaps allowing less concentration on the task at
hand, flying the aircraft. In response to the question, the student pauses,
then reaches up and selects the new altitude in the controller. This diverted
thought/action process distracts him somewhat, and probably results in an
unwanted aircraft roll due to asymmetric engine thrust. Murphy immediately
responds with some “coaching” to keep the aircraft’s flight
regime within safe parameters, and responds, “get the bank” .
The
assumption that the final altitude alert chime is due to the aircraft departing
it’s “set” altitude of 2,000 feet in an uncontrolled descent
is in error. At no time prior to inflight breakup does the aircraft depart
1,900 feet. This is well documented by the 1,900 foot mode C altitude returns
that continue up to the last sweep of the CDR data.
It
must be remembered that the only instrument in the cockpit that is inoperative
is the left side (student’s) attitude indicator. Both altimeters,
vertical speed indicators and airspeed indicators etc. are functional. Because
of the greater number and quicker reacting “cues” available for
altitude (i.e., altimeters, vertical speed and airspeed indicators), it is
inherently easier to maintain pitch attitude, therefore
altitude,
than it would be to maintain roll attitude with an inoperative attitude
indicator. Therefore, we would expect the student to have more trouble
maintaining “wings level” than maintaining altitude.
And
we see this with the instructor’s comment, “get the bank”.
Murphy has carefully monitored aircraft altitude throughout the flight - there
is no reason to suspect he is not aware of the aircraft altitude at this point,
and he makes no comments regarding any altitude deviations near the end of the
tape.
Situational
Awareness
The
probable cause of this accident, as determined by the NTSB was partially
“...the instructor pilot’s loss of altitude awareness and possible
spatial disorientation, which resulted in the loss of control of the airplane
at an altitude too low for recovery...
page 27
There
are 32 minutes of recorded conversation on the CVR prior to the cessation of
the tape. In that time, instructor Murphy continually comments on the
airplane’s altitude, the student’s altitude awareness and the
proper aircraft attitude. There are 10 different discussions of descent to MDA
prior to final approach course interception, totaling several minutes. There
are over 15 different comments and discussions that reveal the
instructor’s total command of the “situational awareness” of
the flight. ‘You’re eight miles from the beacon...”
(2118:17), “Just be one step ahead of the plane. Know you’ve got to
get to the VOR (procedural discussion)...” (2121:04), and “..want
to be aggressive to get on that course outbound now.” (2142:47), are a
few examples.
There
are over 20 different instances of Murphy’s coaching the student
regarding altitude and attitude, including “...already at eight hundred
feet.” (2115:03), “watch altitude..” (2128:26),
“don’t climb” (2114:55), “watch what you’re
doing...” (2124:31), “what speed do you want?” (2121:30),
“fly the airplane first...” (2114:36), “watch your altitude,
altitude, altitude...” (2130:56), “now think what you’re
doing, one fifty ...“ (213 1:13). And finally, at the end of the
recording, “Stop one thing at a time...” and “get the
bank” (2146:44).
All
of these indicate an acute awareness on the part of the instructor, of the
attitude and altitude of the aircraft. Obviously in order to state “get
the bank”, one of the final comments on the CVR, Murphy has to have
referenced his own operating attitude indicator, processed the information,
determined that the trainee can provide proper control inputs to keep the
aircraft on profile, and has relayed that information to the student.
At
no time during the flight, and specifically in the last few minutes, is there
any stress evident in instructor Murphy’s voice. None of the
spectroanalysis done indicates any fluctuation or deviation from normal voice
patterns, which would be present if there was any indications of
“alarm” in his comments. Captain Murphy maintained aircraft
attitude and altitude control throughout the flight by reference to his own
instruments. There is absolutely no evidence that would indicate that Capt.
Murphy would fail to notice any deviation from desired altitude. The
termination of the CVR and the breakup of the aircraft was due to an
instantaneous, catastrophic event at altitude to which the pilots had no
warning and from which they could not recover.
Summary
Findings
#1.
The flightcrew was qualified and current in accordance with FARs and company
policies.
#2.
All training was conducted in a professional manner. The instructor pilot (IP)
was concentrating on those areas of weakness for the captain-trainee.
specifically “‘partial panel” flying and single engine
approaches.
#3.
The Beechcraft 1900 has a well-known history of engine mount truss tube
cracking, separation and failure. The original truss assembly has undergone
extensive design and manufacturing changes in an attempt to alleviate some of
these problems.
#4.
Through electroacoustic transduction, the CVR continually records the transient
structural frequencies normally associated with the pre-existing condition of
two failed engine truss tubes.
#5.
Very near the end of the flight, an excessive bank angle probably resulted 4ue
to trainee distraction and asymmetric thrust. However, because airspeed was
well below configuration maneuvering speed, pilot control input could not have
resulted in overstress to the airframe.
#6.
Power was reduced on the right engine from a zero thrust setting to the flight
idle setting. A whirl mode flutter was induced to the right engine and
propeller assembly, due to pre-existing right engine truss tube failures.
#7.
This whirl mode flutter is evident in the CVR spectroanalysis, and is followed
immediately by a violent acoustical event and instant cessation of the CVR
recording.
#8.
The whirl mode flutter caused a catastrophic failure within the truss mount
system. This failure allowed the right engine and nacelle to depart the right
wing, evident as a strong acoustical event recorded immediately prior to the
first ““spike” on the CVR.
#9.
The right engine struck and removed the right horizontal stabilizer. The
propeller probably damaged the left horizontal stabilizer. The aircraft then
pitched over violently and instantly. The CVR stopped recording when the 5g
limiting switch was tripped by the force of the engine striking the empennage.
#10.
Both wings failed upon pitchover, and the aircraft was no longer controllable.
After descending 1900 feet, the fuselage of the aircraft impacted the surface
of the ocean in a nearly vertical attitude.
NOTE:
The
complete set of diagrams, photos and reports are not reproduced at this web
site, except one diagramdepicting the tail damage (Figure 2) is attached to this file. Also,.
a
paper describing Stearman's work with the voice recorder, which includes audio
and visual illustrations can be accessed at
http://www.acoustics.org/press/133rd/2psa1.html
Footnotes:
[1]
National
Transportation Safety Board, “Loss of Control, Business Express, Inc.,
Beechcraft 1900C N811BE Near Block Island Rhode Island, December 28,
1991”, PB93-910405, NTSB/AAR-93/O1/SUM, April 27, 1993
[2]
ALPA
has included approximately 60 photographs as part of this exhibit. An additional
350
photographs
are available for review upon request.
[3]
Bowers, David F., Ph.D., Packer Engineering, Inc., “MG - Business
Express, Inc., SN UB-49 Beech 1900C”, March 1, 1993
[4]
Cauble,
Robert F., Associated Data Resource, “Boston ARTCC & Ocean Terminal
Radar Control, Recorded Radar Data Report, Radar Data Study, Data List,
Plots”, 1994
[5]
“Beechcraft
1900 Series Maintenance Manual”, P/N 1
-0021-7a26,
Beech Aircraft Corporation, 1988
[6]
"Staff Study, Model 1900/1900C engine Truss Fatigue Cracking". Beech Aircraft
Corporation, Oct 19, 1990
[7]
Stearman,
Dr. Ronald, P.E., Buschow, Monte, Kane, Kevin, “The Beech Aircraft
Corporation Model 1900 Airliner Engine Truss: A Study in Reliability Analysis
& Aviation Safety”, June 3, 1995
[8]
Complete
memorandum and cover letter are included in attachment C.
[9]
Complete
SDR list compiled by Air Data Research May 20, 1997 available upon request
[10]
MSC/Nastran -“ Handbook for Aeroelastic Analysis” Vol. 3, Version
68 Mac Neal-Schwindler, 1995, USA. This is a commercially available finite
element modeling code. This study used the whirl flutter program resident in
the aeroelasticity option.
[11]
Brock, B., “Personal Communication Concerning the Modified XC 142 Whirl
Flutter Software Code”, 1966. A refinement of this system was used, using
a basic whirl flutter software code developed by Voight Aeronautics
[12]
Stearman, Dr. Ronald, P.E. , Schulze, Glen H., Rohre, Stuart M.,
“Aircraft Damage Detection from Acoustic and Noise Impressed Signals
Found by a Cockpit Voice Recorder”, 1997, Institute of Noise Control
Engineering.
[13]
McSwain
Richard H., Ph.D., P.E., Investigation Conclusions, Materials Engineering
Investigation. The original reports includes approx. 410 photographs and
supporting microscopic examination documentation, not included in this petition
but available upon request.
[14]
Hammill Donald F., “Expert Report”, 1995
[15]
Pratt & Whitney Canada, Service Investigative Report, “Business
Express Beech 1900C NS11BE, Block Island, Rhode Island”, Report #TL-852,
August, 1992
[16]
Zwillenberg,
Peter, Report 1900JE361 D, page 80, Beech Aircraft Corporation, September 30
1983, as quoted in Attachment D, page 47
[17]
Reed, W. III, “A Review of Propeller-Rotor Whirl Flutter”, NASA
TR R-264, 1967
[18]
Schulze,
Glen H., “Cockpit Voice Recorder Tape Erasure Gap Study & Signal
Level Inventory and Study”, Data Acquisition Systems, Littleton, CO.
February 26, 1996
[19]
"Aviation Instructor's Handbook", US Department of Transportation, Federal
Aviation Administration, 1977, page 3
[20]
NTSB report of subject accident, NTSB/AAR-93/O 1/SUM, April 27, 1993, page 18
|