Air Safety Through Investigation - Journal of the International Society of Air Safety Investigators - International Society of Air Safety ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Air Safety Through Investigation JANUARY-MARCH 2021 Journal of the International Society of Air Safety Investigators Do We Need an Annex 13 for Commercial Space Accidents? page 4 Video-Based Flight Data Reconstruction page 8 Why Did the Helicopter Collide with Trees? page 14 Competency- Based Education: Framework for a More Efficient and Safer Aviation Industry page 19 Analyzing Large and Complex Image Collections page 24
CONTENTS Air Safety Through Investigation
Journal of the International Society of Air Safety Investigators
FEATURES Volume 54, Number 1
Publisher Frank Del Gandio
4 Do We Need an Annex 13 for Commercial Space Accidents? Editorial Advisor Richard B. Stone
By Joseph M. Sedor, Chief, Major Investigations, the NTSB—The author suggests that with Editor J. Gary DiNunno
the growth of global commercial and governmental space vehicle launches more mishaps Design Editor Jesica Ferry
may become inevitable and perhaps space flight safety professionals will require interna- Associate Editor Susan Fager
tionally recognized rules and procedures similar to Annex 13.
ISASI Forum (ISSN 1088-8128) is published quar-
8 Video-Based Flight Data Reconstruction terly by the International Society of Air Safety
By Dr. Marcus Bauer, Managing Director, iwiation—The author observes that traditional Investigators. Opinions expressed by authors do
safety investigation data may not always be available. He notes that video footage from wit- not necessarily represent official ISASI position
nesses with their mobile phones or from security cameras is becoming more prevelent and or policy.
can be effectively used to recontruct flight data, aircraft attitude, descent rate, and ground
Editorial Offices: Park Center, 107 East Holly Ave-
speed.
nue, Suite 11, Sterling, VA 20164-5405. Telephone
703-430-9668. Fax 703-430-4970. E-mail address,
14 Why Did the Helicopter Collide with Trees? isasi@erols.com; for editor, jgdassociates@
By Koji Fukuda, Deputy Investigator for Aircraft Accidents, Japan Transport Safety starpower.net. Internet website: www.isasi.org.
Board—The author describes a novel accident investigation approach using drone and ISASI Forum is not responsible for unsolicited
video images, sound recordings, and reenacting the flight path with another helicopter of manuscripts, photographs, or other materials.
the same type. Unsolicited materials will be returned only if
submitted with a self-addressed, stamped enve-
19 Competency-Based Education: A Framework for a lope. ISASI Forum reserves the right to reject,
More Efficient and Safer Aviation Industry delete, summarize, or edit for space con-
By Dr. Flavio A.C. Mendonca, Ph.D., Assistant Professor; Dr. Julius Keller, Ph.D., Assistant siderations any submitted article. To facilitate
Professor; and Dr. Brian Dillman, Ph.D., Associate Professor, Aviation and Transporta- editorial production processes, American Eng-
lish spelling of words is used.
tion Technology, Purdue University—The authors examine how evidence-based training
can help identify, develop, and evaluate the pilot competency requirements to safely operate Copyright © 2021—International Society of Air
in a commercial air transport environment. Safety Investigators, all rights reserved. Publica-
tion in any form is prohibited without permis-
24 Analyzing Large and Complex Image Collections sion. ISASI Forum registered U.S. Patent and
By Floris Gisolf, Dutch Safety Board Investigator and Data Analyst for the Maritime De- T.M. Office. Opinions expressed by authors do
partment; Zeno Geradts, Senior Forensic Scientist at the Netherlands Forensic Institute not necessarily represent official ISASI position
of the Ministry of Security and Justice at the Forensic Digital Biometrics Traces Depart- or policy. Permission to reprint is available upon
ment; and Marcel Worring, Computer Science Professor, University of Amsterdam—The application to the editorial offices.
authors review the Dutch Safety Board’s Flight MH17 investigation and the considerable
Publisher’s Editorial Profile: ISASI Forum is print-
amount of images that had to be sorted, catalogued, and saved or rejected. They consider
ed in the United States and published for profes-
methods and proceudres that could have made the process more efficient and may be useful sional air safety investigators who are members
in future investigations. of the International Society of Air Safety Inves-
tigators. Editorial content emphasizes accident
investigation findings, investigative techniques
DEPARTMENTS and experiences, regulatory issues, industry ac-
cident prevention developments, and ISASI and
2 Contents member involvement and information.
3 President’s View Subscriptions: A subscription to members is pro-
28 News Roundup vided as a portion of dues. Rate for nonmem-
30 ISASI Information bers (domestic and Canada) is US$28; Rate for
32 In Memoriam nonmember international is US$30. Rate for all
libraries and schools is US$24. For subscription
information, call 703-430-9668. Additional or
ABOUT THE COVER replacement ISASI Forum issues: Domestic and
Canada US$4; international member US$4; do-
mestic and Canada nonmember US$6; interna-
NASA announced on Dec. 9, 2020, that 18 astronauts were selected to train for tional nonmember US$8.
the Artemis missions designed to return humans to the lunar surface. As com-
mercial space missions are on the rise in the United States and other nations,
the question for ISASI: Do we need a space annex to provide international
space safety procedures and investigation authority similar to Annex 13 here on
Earth? (See page 4.) Fused photo of a super moon by Gary DiNunno.
INCORPORATED AUGUST 31, 1964
2 • January-March 2021 ISASI ForumPRESIDENT’S VIEW
REHABILITATING PARKED AIRCRAFT FOR FLIGHT
IT’S GOING TO TAKE A LOT OF PEOPLE AND A LOT OF TIME TO GET THINGS GOING AGAIN. AND IF
YOU'VE GOT LESS PEOPLE, IT’S GOING TO TAKE MORE TIME. — STEPH SMITH
I
wish everyone a good and safe new year. As 2021 began, runs in accordance with the aircraft maintenance manual.
we’re still facing individual and air safety professional diffi- “These are done to ensure the engines are still performing as
culties throughout the world—both trying to stay healthy expected and that the long-term storage hasn’t caused any det-
during the pandemic and surviving the resulting economic rimental effects to any of the systems that wouldn’t be obvious
turmoil that’s wrecked national economies and left many just by looking at them,” said Smith. But Scott stressed that
people permanently unemployed or temporarily furloughed. it will take a considerable amount of time to get everything
There’s hope now for a better future as vaccines and improved flying again given the volume of maintenance work that needs
medical treatments move from their development stage to ini- to be done and the number of engineers who’ve been laid off
tial distribution and eventually become widespread. Recovery during the pandemic. “It’s going to take a lot of people and a
for global health and the revival of the aviation industry will lot of time to get things going again,” Smith remarked. “And if
still require years of coordination, cooperation, and careful you’e got less people, it’s going to take more time.”
observation. “With some of the newer aircraft, these tests need to be fol-
In my last “President’s View” and in my video on ISASI’s lowed in the exact order and to the exact second, otherwise it
website, I mentioned remaining vigilant about safety issues can fail the test and set you back a few hours,” she observed.
as mothballed aircraft are returned to service. Last Decem- “Once the maintenance work packs are cleared and certified,
ber, Kathy Scott, a CNN reporter, provided some interesting the engineer can then sign the aircraft off as airworthy. They’re
interviews with a manager of a site in Alice Springs, Austral- the final signature that says, ‘I'm happy that everything has
ia, where unused aircraft are stored, and a certified aircraft been done correctly. I’m now releasing the aircraft to service.’
engineer and the commercial director of an aviation services That final signature is what the captain will see to then sign
the logbook for the aircraft,” noted Smith.
company. Scott observed that more than two-thirds of the
Mike Cone, commercial director of eCube, an aviation
world’s commercial aircraft were grounded in 2020, and some
services company, told Scott that engineers must religiously
31 percent are still in storage. The Asia Pacific Aircraft Storage
follow procedure when reactivating an aircraft. Cone also
(APAS) facility still has twice its usual number of jets, now
observed that only companies approved by national airwor-
about 150, and has increased its capacity to more than 200,
thiness authorites can perform this maintenance. He said
said APAS managing director Tom Vincent, who expects de-
no aircraft stored at eCube facitities in Spain and the United
mand to rise in 2021. Vincent observed that getting these air-
Kingdom have returned to the flightline (at the time of this
craft back on the line requires an extensive program prescribed CNN interview).
by the maintenance manual provided by each manufacturer Cone speculated that the
before a certificate of release to service can be signed. “more marketable” aircraft such
Scott also learned there are a myriad of usual problems that as the Airbus A320 and the
must be resolved in these rehabilitation programs. There’s B-737 will return to passenger
more than just removing engine protectors and tape covering flight and that other aircraft
every hole, port, or probe, according to Licensed B1 aircraft may undergo passenger to cargo
engineer Steph Smith. For example, bugs (wasps have a way of conversions. Vincent suggested
nesting in hard-to-reach places), water, or debris can invade that newer aircraft will leave his
aircraft systems. Smith estimated that getting a widebody facility first as demand returns.
aircraft ready for flight takes about 100 staff-hours, and a nar- Smith commented that the
rowbody aircraft takes about 40—depending on the size of the “greener” more sustainable
aircraft and the length of storage time. aircraft types will survive this Frank Del Gandio
She commented that engineers have to do a series of engine downturn. ISASI President
January-March 2021 ISASI Forum • 3Joseph M. Sedor
DO WE NEED AN ANNEX
13 FOR COMMERCIAL
SPACE ACCIDENTS?
By Joseph M. Sedor, Chief, Major Investigations, the NTSB
T
(Adapted with permission from the author’s
he growth of the U.S. commercial space industry has accelerated over the last
technical paper Do We Need an Annex
13 for Commercial Space Accident several years; FAA-licensed commercial launches have increased from four in
Investigations? presented during ISASI 2010 to 33 in 2018. This increase was partly the result of policy changes after
2019, Sept. 3–5, 2019, in The Hague, the the retirement of the space shuttle that have required NASA to use commer-
Netherlands. The theme for ISASI 2019 cial launch/reentry systems for International Space Station resupply and astronaut
was “Future Safety: Has the Past Become transfer. Therefore, commercial space launches and reentries will likely continue to
Irrelevant?” The full presentation can be increase, which will inevitably result in more mishaps.
found on the ISASI website at www.isasi. Over the last 25 years, the NTSB’s Office of Aviation Safety (OAS) has been devel-
org in the Library tab under Technical oping technical expertise and building relationships with stakeholders involved in
Presentations.—Editor)
this emerging mode of transportation to ensure that the NTSB and stakeholders are
prepared to investigate any commercial space accident or incident. Although there
are similarities between commercial space and aviation investigations, the industry
structure, technologies, national security laws, and international treaties that govern
space operations dictate that the investigations will be considerably different. Ulti-
mately, these differences will also drive how states interact during an international
accident or incident investigation.
The NTSB has been leading or supporting commercial space accident investiga-
tions for more than 25 years and has conducted two major space vehicle investiga-
tions. In 1993, the NTSB investigated the procedural anomaly that occurred during
the launch of an Orbital Sciences Corporation Pegasus expendable launch vehicle.
The investigation found safety issues related to command, control, and communica-
tions responsibility; launch crew fatigue; launch interphone procedures; efficiency
4 • January-March 2021 ISASI Forumof launch constraints; and the lack of Commercial Space Transportation regulations, AST regulations refer to any
common launch documents. In its final Department person aboard a commercial space vehi-
report, the NTSB issued 17 safety recom- The space industry has historically been cle who is not a crewmember as a “space
mendations to the Department of Trans- led by the government and military, flight participant” and not a “passenger.”
portation, NASA, and Orbital Sciences. so the commercial sector is relatively In addition, these space flight partici-
In 2014, the NTSB investigated the young, especially when compared to pants are required to acknowledge the
accident of the SpaceShipTwo reusa- the aviation industry. The U.S. commer- risks by signing an “informed consent”
ble suborbital spaceplane that broke cial space industry officially launched that identifies the risks or probable loss
up during a rocket-powered test flight, in 1984 when the Commercial Space during each phase of launch/reentry and
killing the copilot. The NTSB identified Launch Act (Space Act) was signed into the safety record of the launch/reentry
safety issues regarding the lack of human law. The Space Act created the Office of vehicle type (describing the launch/reen-
factors guidance for commercial space try failures, if any). The launch/reentry
Commercial Space Transportation (AST)
operators, the efficacy and timing of the operator must also purchase a specific
in the Department of Transportation,
preapplication consultation process, lim- amount of liability insurance, deter-
and in 1995, AST was moved within
ited interactions during the experimen- mined by AST for each launch, to cover
the FAA. The mission of the FAA AST is
tal permit evaluation process, deficien- any third-party loss (injury, death, prop-
to ensure protection of the public and
cies in the evaluation of hazard analyses, erty damage, etc.), which can be up to
property; protect national security and
and the need to improve the lessons $500,000,000. Since this liability amount
foreign policy interests of the United
learned database. The NTSB issued a might be insufficient for an exceedingly
States; and to encourage, facilitate,
total of 10 recommendations to the FAA’s “bad day,” the Space Act also requires
and promote U.S. commercial space
Office of Commercial Space Transpor- the federal government to indemnify
transportation. As part of its oversight
tation and the Commercial Spaceflight launch/reentry companies for claims
responsibilities, the FAA AST issues
Federation. that exceed their required insurance
licenses and experimental permits for
In addition to these NTSB-led investi- coverage, which could be up to $3 billion
commercial launches and reentries of
gations, NTSB investigators have assist- (in 2016 dollars).
orbital and suborbital rockets. However,
ed in multiple spacecraft accident inves- unlike the FAA’s aviation regulatory goal
tigations. Throughout the seven-month of providing the safest system in the NTSB Relationship with Commercial
Columbia space shuttle investigation in world, the U.S. Congress has charged the Space
2003, more than 40 NTSB investigators AST to primarily focus on protection of NTSB involvement with the commercial
assisted both the Columbia Accident the public—and not “mission assured- space industry is similar to the aviation
Investigation Board and NASA with bal- ness.” industry. The NTSB investigates any
listic analysis, debris recovery, wreckage This is not to say that the U.S. Congress launch/reentry accidents and certain
examination, and vehicle reconstruction. does not want a safe commercial space incidents, and that authority is derived
Several NTSB investigators also assisted industry. The Space Act, which was most from the NTSB’s general authority
NASA in 2004 with the investigation of recently amended in 2015, states that the under 49 Code of Federal Regulations
the Genesis sample-return capsule that FAA AST should “encourage, facilitate, (CFR) 1131(a)(1)(F), which states that
crashed into the Utah desert. NTSB and promote the continuous improve- the NTSB shall investigate “any other
investigators documented the accident ment of the safety of launch vehicles accident related to the transportation
scene, organized the wreckage recov- designed to carry humans.” However, of individuals or property when [the
ery, and examined the vehicle’s wiring Congress also does not want to discour- accident is] catastrophic.” Although this
harness for evidence of micrometeorite age industry development since human statement is not as clear as 49 CFR 1131
impact damage. space flight is still inherently risky. So (a)(1)(A), which states that the NTSB
More recently, NTSB investigators the Space Act includes a provision for a shall investigate all “aircraft accidents,”
have observed or taken part in several “learning period,” which limits any regu- the NTSB’s interpretation of this statute
operator-led mishap investigations, in- lation “restricting or prohibiting design is reasonable and is accepted by the
cluding the October 2014 Orbital Science features or operating practices” unless commercial space industry.
(ATK) Antares engine failure shortly resulting from an accident that caused a The process to clearly specify the
after liftoff; the June 2015 launch failure serious injury or fatality to a person on NTSB’s authority to investigate commer-
of the SpaceX CRS-7 mission; the Sep- board, or from a serious incident that cial space accidents has begun; however,
tember 2016 pad explosion of the SpaceX almost caused a serious injury or fatality. statutory changes are seldom fast. In
Falcon 9 with the Amos-6 communica- This learning period, also known as the addition to the NTSB statutory authority,
tions satellite; and the April 2019 SpaceX human space flight regulation morato- the NTSB also entered into a memoran-
Dragon explosion that occurred during rium, is currently in effect until October dum of agreement (MOA) with the FAA
a ground test. This “on-the-job training” 2023. in 2000 to ensure both agencies under-
has provided NTSB investigators with This is quite different from aviation stand when the NTSB would initiate an
significant and critical experience in the regulations that have developed over investigation of a nonmanned commer-
commercial space industry, which has the years to cover almost all aspects cial launch accident. The MOA defines
helped the OAS to prepare to lead the in- of an aircraft design and operational an accident that the NTSB would inves-
vestigations of future commercial space rules to ensure and improve the safety tigate as a mishap when any portion of
accidents and incidents. of passengers and crew. Unlike aviation a commercial space vehicle or payload
January-March 2021 ISASI Forum • 5impacts outside the impact limit lines; a rather than just accidents and incidents. vehicle and/or range is grounded until at
fatality or serious injury to a person not The NTSB will likely issue its own defi- least the initial causes are determined.
associated with the launch activities; or nitions along these lines in the next few Therefore, the OAS’s investigative pro-
damage greater than $25,000 to property years. cess will likely have to be accelerated,
not associated with the launch activities. As indicated by these definitions, although even targeting 8–12 months
This agreement was initiated during a it is clear that it takes a more serious for a final report may be too long. To ad-
time when human space flight and com- mishap to constitute a commercial space dress this issue, NTSB staff is evaluating
mercial reentry operations were not an- accident; an event involving an aircraft the possibility of modified procedures to
ticipated for many years; thus, although is classified as an accident if there is sub- release preliminary analytical findings
the MOA is still in effect, it is out of date stantial damage to the aircraft of serious (prior to the final report) to allow the
and does not address those operations. injury to a person. Loss of the space resumption of launches or reentries.
Work on updating this outdated MOA vehicle alone does not automatically Probably the most significant differ-
is ongoing. In addition to this MOA, the necessitate an NTSB investigation. How- ence between an aviation and space
NTSB has an MOA with the FAA and the ever, now that the industry is moving investigation will be in the transparency
U.S. Air Force that defines the relation- toward human spaceflight, should a mis- of the investigation. Normally, the NTSB
ship between all three agencies during hap occur, the NTSB would investigate it releases factual data through press con-
space investigations and describes the if there is a fatality or serious injury. ferences (while on scene), press releases
participation and information exchange As previously discussed, the NTSB has (post on scene), and then via the pub-
procedures. been engaged with the space industry lic docket, which contains all relevant
Although NTSB regulations do not for the last 25 years through our com- factual information NTSB investigators
contain specific definitions related to mercial space program within the OAS. collected during the investigation. The
commercial space mishaps, 14 CFR 401.5 The program has matured along with NTSB even has the authority to release
contains definitions for commercial the industry to ensure that the NTSB is confidential commercial information
space launch and reentry accidents or prepared to investigate any future space (proprietary data)—which is normally
incidents that assist the OAS in deter- vehicle accident or incident. The goals of protected—to support the conclusions,
mining whether to initiate an investi- the NTSB’s commercial space program safety recommendations, or the probable
gation. A launch accident is when there are to develop the specialized investiga- cause of an accident.
is a fatality or serious injury to a space tive processes and procedures necessary However, the space industry has
flight participant or crewmember; a to investigate this emergent mode of regulations (export control policies) in
fatality or serious injury to any person transportation, build critical relation- place to safeguard critical defense-re-
who is not associated with the flight; ships with industry stakeholders, and lated technologies in order to protect
impact of launch vehicle, its payload, ensure that NTSB investigators have the U.S. national security and foreign policy
or any component outside the impact specialized knowledge necessary to lead objectives: the International Traffic in
limit lines ( for expendable) or outside a commercial space investigations. Arms Regulations (ITAR) and the Export
designated landing site ( for reusable); or Administration Regulations (EAR). The
damage to third-party property greater NTSB is not exempt from these regula-
than $25,000. A reentry accident occurs Aviation and Commercial Space
tions and cannot release data that is sub-
when the reentry vehicle, its payload, or Industries Differ ject to ITAR or EAR, even if it is directly
any component impacts outside a des- Although there are some similarities related to the probable cause. Thus these
ignated reentry site; a fatality or serious with aviation, there are distinct and regulations would impact the NTSB’s
injury to a space flight participant or unique aspects of commercial space ability to release information on space
crewmember or a person not associated investigations that necessitate they be accidents.
with the reentry; or damage to third-par- investigated as its own distinct mode As a result, the OAS has developed an
ty property greater than $25,000. of transportation. As with all modes of internal procedure to work with the U.S.
In addition, for those mishaps that do transportation, the NTSB will use the Department of State and Department
not rise to the level of an accident, the party system to investigate commercial of Commerce to review all factual data/
regulation also contains definitions for space accidents/incidents. The overall reports and the final report before public
launch and reentry incidents to better structure will be similar to major avia- release. This extensive review process
define what mishaps the NTSB would tion investigations but will likely have will inevitably delay the release of factual
likely investigate. A launch/reentry fewer parties to the investigation since information until the end of the inves-
incident is an unplanned event during most of the rocket and vehicle compo- tigation rather than being released as
the launch or reentry that would involve nents will be produced by the launch soon as possible (typically about six
a malfunction of a flight safety system or operator. In addition, the OAS’s goal for months into the investigation for major
safety-critical system or a failure of the completion of major aviation accident investigations). The ITAR/EAR review
license’s or permittee’s safety organiza- reports is to issue a final report within will also likely result in extensive redac-
tion, design, or operations. The FAA AST 12–18 months, which is acceptable in the tions to the factual reports contained
has recently proposed to modify these aviation industry since normal opera- in the public docket. Even more signif-
definitions to more closely align with tions generally continue throughout that icantly, on a highly technical accident,
those used by the military and NASA by timeframe. However, for every launch or portions of the final report may also have
having four “classes” of mishap events reentry accident to a space vehicle, the to be redacted, and, in extreme cases,
6 • January-March 2021 ISASI Forumthe NTSB may be required to hold the to the unique aspects of the state conducting the launch (any broad/universal
final board meeting in a closed session the commercial space safety findings would be released by the launching state).
in order to discuss data that is ITAR-/ industry (as discussed in
EAR-restricted. For an agency that this paper). One of the
In Conclusion
normally prides itself on openness and more significant differ-
Given the domestic and international regulatory
transparency, these restrictions on the ences between space and
environment, the national security laws, and the unique
release of data will be a challenge, and aviation is that launch
technical structure of the industry, I do not believe
an appropriate public/media education operators are typically
that—at this time—a formal “commercial space Annex
effort will be necessary to communicate the manufacturer and
13” is necessary. However, informal international
why the restrictions are essential. operator of their vehicle,
cooperation will be vital to ensuring that safety investi-
Any commercial space orbital launch unlike aviation where
gators are prepared to meet the technical and organiza-
or reentry accident could become an there are separate man-
“international” accident depending on tional challenges of investigating commercial space
ufacturers and separate
the trajectory of the launch/reentry. operators around the vehicle accidents. The existing Annex 13 aviation
Space operations are governed by an world. accident investigation authorities and professional
international treaty entitled the “Treaty Accordingly, using organizations—such as ISASI—will serve as excellent
on Principles Governing the Activities Annex 13 terminology, resources in building these relationships. The collabora-
of States in the Exploration and Use of the state that author- tive efforts between state authorities will help the
Outer Space,” commonly referred to as ized the launch would investigative community refine the specific processes
the Outer Space Treaty. This treaty has be the state of registry, and procedures for investigating commercial space
driven much of the U.S. space regulation design, manufacture, accidents, discover or develop training opportunities,
structure. Although the treaty does not and operator, and the gain understanding of various regulatory structures, and
directly address accident investigation, state where the vehicle observe commercial space operations. At some point in
it does address the responsibilities of or components impact- the future, a more formal safety structure may be needed
each state in the event of a mishap. If ed would be the state as the commercial space industry grows and matures
a launch or reentry vehicle impacts in of occurrence Although (point-to-point service, etc.). However, until that time,
another state’s territory, that state has the state of occurrence the spirit of international cooperation, which has been
two responsibilities: render “all possible would have a considera- cultivated by Annex 13, will help ensure that the investi-
assistance” to any persons on board and ble need to understand gative community is prepared to assist this emerging
to “safely and promptly” return persons the facts, conditions, and mode of transportation to improve safety following any
and the vehicle or components. circumstances of the ac- commercial space mishap.
Article VII of the Outer Space Treaty cident, it would be nearly
declares that the state that authorized impossible for the state
the launch or reentry is “internationally of occurrence to conduct
liable” for damage caused by the vehicle a thorough investigation
or components to any persons or proper- of the mishap, since the
ty. This type of liability-focused language expertise resides entirely
likely stemmed from the fact that when with the launch opera- Joseph M. Sedor, chief of major investigations for the NTSB,
the treaty was originally developed in tor—especially since the delivers his technical presentation during ISASI 2019.
1967 all the space operations were one- operator would likely be
of-a-kind rockets operated by state gov- prohibited by law (due
ernments. The Outer Space Treaty does to export regulations) to
encourage international cooperation transfer information.
multiple times throughout the treaty; Likewise, the ITAR and
for example, Article X says that states EAR national security
should afford an opportunity to other regulations would also
states to observe launches and reentries. make it difficult for the
Cooperation, of course, is essential for state of occurrence to
conducting international investigations, send observers to a state
but a state’s specific rights and respon- of the operator-led inves-
sibilities also need to be defined for any tigation. Furthermore,
future international investigative treaty. if there are no distinct
ICAO Annex 13 has provided the basic operators of the accident
structure for international aviation launch or reentry vehicle
investigations for more than 50 years, de- in the state of occur-
tailing the cooperation necessary as well rence, there would be no
as a state’s rights and responsibilities safety reason for them to
during an investigation. However, the conduct an investigation
standards and recommended practices since any lessons learned
in Annex 13 are not directly applicable would likely only affect
January-March 2021 ISASI Forum • 7Video-Based Flight Data
Reconstruction
By Dr. Marcus Bauer, Managing Director, iwiation
(Adapted with permission from the author’s technical paper Video-Based Flight Data
Reconstruction of the Amazon Prime Air B-767 Accident, Trinity Bay, U.S.A., 2019, submtted for
ISASI 2020 in Montréal, Qué., Canada. ISASI 2020 was postponed until 2021 due to COVID-19 re-
strictions. The full technical paper can be found on the ISASI website at www.isasi.org in the Library
Dr. Marcus Bauer
tab under Technical Presentations.—Editor)
Introduction recalculate a flight path and thus the flight
The sole objective of an investigation into history without flight data recorder (FDR)
an aircraft accident or incident conducted data or radar information. The physiology
under the provisions of Annex 13 shall be and psychology of an eyewitness have
the prevention of accidents and incidents. been evaluated; however, the method can
The purpose of this paper is to encour-
age an additional investigation means to
also process video information to approxi-
mate flight data. “Flight
the existing procedures, practices, and
techniques that can be used in aircraft
The methodology has been applied al-
ready and successfully in several investiga- data that is
accident investigations.
The investigation of air accidents is
tions, using video information in examples
such as reconstructed
based on available data and information
to determine the root cause. Flight record-
• AS350 Mid-Air Accident, La Rioja,
Argentina, BEA, 2015, based
er data, radar data, and wreckage anal-
ysis can provide important information.
• EC145 Accident, Hautes-Pyrénées,
France, BEA-É, 2016, and
on video
However, in some air accidents, some if • Gazelle Mid-Air Accident, Carcès, information is
not all of these sources of information
applicable!”
France, BEA-É, 2018.
may not be available to investigators. In
recent years, more and more video footage The iwi method allows to approximate
is available, either from witnesses who flight path as well as aircraft attitude and
recorded the accident with their mobile ground speed. The accuracy of the recon-
phones or videos recorded by security structed data is influenced by multiple
cameras. This information can be used to technical factors.
reconstruct flight data, aircraft attitude, In order to reconstruct the elevation,
descent rate, and ground speed. the following factors must be considered
and computed for the error calculation as
the observed value versus the true value of
Thesis a measurement:
This paper represents the potential and
the implementation of using video infor- Figure 1. Angular deviation in elevation depending on reference object, eyepoint (left), position of
mation to reconstruct the flight history reference object (middle), and observed object location (right).
and the flight path in detail. The consist-
ency of the reconstructed information will
be explained as well as how it has been
validated.
The Methodology
In the frame of my doctoral thesis, the
reconstruction methodology iwi® was
developed in 2009, based on eyewitness
reports in the field of aircraft accident
investigation. The development has been
based on the overview of existing appli-
cations and the existing problematic to
8 • January-March 2021 ISASI Forum• Δd_xyGPS: Video location accuracy
(latitude and longitude) in meters.
• Δd_zGPS: Video location accuracy
(altitude) in meters.
• Δd_xyZPos: Location accuracy of
reference objects (latitude and longi-
tude) in meters.
• Δd_zZPos: Location accuracy of refer-
ence objects (altitude) in meters.
Formula 1
• Ratio of the recorded object resolu-
tion in pixel and object size in meters.
With reference to Figure 1, the formula
considers different errors. To place the
video informationin in 3-D, a reference
object (Obj) as well as the observer (O)
location are required. By knowing the
positions of “O” and “Obj,” the minima and
maxima altitude of the aircraft can be cal- Formula 2
culated by the formula below. The default
error tunnel (grey cone) is the result of an
aproximated model, showing the possible
area where the altitude of the observed
object was located. The error tunnel
dimension is based on the known errors,
such as the camera’s position accuracy, as
well as the video resolution. Another influ-
encing factor of the default tunnel’s size is
the distance between the observer and the
reference point, as a small miss-position-
ing of the observer resulting in a variation
of the observed object’s height. As the var-
iation gets larger, the closer the reference
object is relative to the position (latitude/
longitue) to the observer.
Formula 1 is used with its derivations to
calculate the elevation error based on the
reconstructed distance to the object.
In azimuth, the following factors must Figure 2. Geometry for calculating the error in azimuth, from witness/camera location via
Obj. (reference object) to reconstructed aircraft location (R).
be considered for error calculation:
• Video location accuracy (latitude and
longitude) in meters.
• Location accuracy of reference objects
(latitude and longitude) in meters.
• Ratio of the recorded object resolu-
tion in pixel and object size in meters
Figure 2 illustrates the formula for the
calculation of the observed object’s (R)
position in latitude and longitude. The ob-
server (O) requires (you are here) as well a
reference point (Obj), which indicates the
reference to the observed object’s position.
By knowing the exact position of “O” and
“Obj,” the exact position in longitude and
latitude can be provided by the formula
shown hereinafter.
Formula 2 is used with its derivations to
calculate the azimuth error based on the Figure 3. Aerial view of main debris field looking northwest.
January-March 2021 ISASI Forum • 9reconstructed distance to the object:
The Evaluation Case
On Feb. 23, 2019, the Amazon Prime Air cargo
aircraft that was operated by Atlas Air flew
from Miami, Florida, to Houston, Texas, and
during the arrival phase the Boeing 767-375BCF
entered a rapid descent from 6,200 feet and im-
pacted into a marshy bay area around 40 miles
away from Houston’s George Bush Interconti-
nental Airport (see Figure 3, page 9).
Two security cameras captured the last 5
seconds of the aircraft in a steep, generally
wings-level attitude until impact with the
swamp. For our own interest, the iwi® method-
ology was applied to validate the latest meth-
odology developments, reconstructing flight
Figure 4. Surveillance camera video. Figure 5. Surveillance camera video. path, aircraft attitude, and ground speed based
on the available video information. The results
were shared with the NTSB before the black box
could be recovered.
Flight Data Reconstruction and Data Com-
parison
Two video recordings were available from two
different locations. They were used to create
panoramic images with aircraft positions and
video time stamps. The video distortion was
corrected using the dedicated lens correction
profile (see Figures 4 and 5).
The resolution of the video influences the
Figure 6. Camera location shown in Google Earth. data reconstruction accuracy as well as the
precision of the time stamp information. Video
frames at every second were selected to ensure
maximum accuracy. The security camera re-
cording frequency was approximately one frame
per second.
The GPS location of both cameras was iden-
tified, as shown in Figure 6, using Google Street
View, as well as the location of reference objects
like buildings and trees, as shown in the video.
The location accuracy was defined with +/- 16
feet in latitude/longitude and altitude. The
Figure 7. Lines of sight of witness statement (left) and reconstructed location with reference objects were used to place the image
error (right). information within the 3-D environment. All
data collected was imported into the recon-
struction software called Immersive Witness
Analyzer, which sets all lines of sight consider-
ing reference objects. The software estimates
the reconstructed flight path considering
potential error information using the specified
and decribed formulas.
To estimate the aircraft’s attitude at different
locations along the flight path, a 3-D model of a
B-767 was placed and adjusted with respect to
the video image frame rate.
The left image in Figure 7 shows the lines
Figure 8. Single frame of video showing the B-767 behind trees (left) and with of sight from the perspectives of two security
overlay of 3-D object (right). cameras. The intersection of both lines ap-
10 • January-March 2021 ISASI Forumproximates the position
of one position of the
observed object as shown
in the right image. In this
case, the B-767’s position
(intersection of both lines
of sight) was located,
considering the potential
errors.
The images in Figure 8
show one single frame of
Figure 9. Attitude error +/-10 degrees visualized (blue/cyan), heading (left), pitch (middle), roll (right). the video. The left image
shows the outline of the
aircraft behind trees. To
determine the attitude of
the aicraft, a 3-D model
of a B-767 was placed in
the 3-D software applica-
tion at the reconstructed
position and the attitudes
of the aircraft such as
heading, pitch, and roll
were adjusted until the
objects fitted best to the
outline in the image. The
accuracy for attitude
reconstruction depends
on the resolution of the
frame. However the
potential error can be up
to +/- 10 degrees based on
experience.
To better explain the
reason for the experi-
enced size of the error in
attitude, Figure 9 shows
the visual differences.
The modified values for
+/-10 degrees are visu-
alized in blue and cyan
and overlayed. Based on
Figure 10. Illustration of the reconstructed flight path with error tunnel (white circles) that shows the
line of sight from the first observing camera (light blue) and second observing camera (dark blue). the experience of several
reconstructions and video
data, a good fit could be
determined within the
maximum error of +/-10
degrees for all three axis.
The reconstructed flight
path was compared to the
original FDR data (shown
in Figure 12, page 12) that
was recovered from the
B-767 wreckage. Figure
11 shows in latitude and
longitude the flight path
that could be reconstruct-
ed in black. The original
flight path (grey) is close
to the reconstructed path
Figure 11. Reconstructed flight path (black), FDR flight path (grey), and error tunnel (blue). (black) and within the
January-March 2021 ISASI Forum • 11error tunnel (blue).
To better compare the difference be-
tween both, the extrapolated flight path
from the video method and the recorded
flight path from the FDR, the distance
was calculated in feet and shown along
the time in Figure 13. The solid blue line
describes the difference in latitude and
longitude in feet between the reconstruct-
ed position and the recorded FDR data.
The graph with dotted blue line shows the
maximum possible distance, known as
Figure 12. Recovered black box (left) and ADS-B flight path data (right). error tunnel, based on the formulas. The
achieved accuracy of the reconstructed
flight path was in a range of between 50
and 150 feet, finally within the error tun-
nel (see Figure 13).
Figure 14 shows the reconstructed
altitude (light blue) of the airplane and the
recorded FDR altitude data (light grey).
The reconstructed altitude is within the
error tunnel shown in dark grey, dotted
blue, with the exception of the beginning
from 0.0 to 0.25 seconds, when the aircraft
was shown only partially in the first frame.
This may have resulted in a larger error
Figure 13. Distance in feet between reconstructed and recorded FDR (blue) and error in the early portion of the calculation.
tunnel (dotted blue). Finally, the difference in calculated verses
FDR recorded altitude was between 32.7
and 171.9 feet.
Since the video could be synchro-
nized with the time base, using the video
frame time stamps, a reconstruction of
the ground speed and descent rate was
possible.
Based on the calculated decent rate
(black) in feet/minute, Figure 16 shows
that the aircraft reduced its descent rate
from -39.800 feet/minute to -15.000 feet/
minute. The recorded FDR data showed
that the B-767 reduced its decend rate
from -28.000 feet/minute to -18.240 feet/
minute within 2 seconds. The difference
between the reconstructed data and re-
Figure 14. Reconstructed altitude (light blue), FDR altitude (light grey), error tunnel (dark corded FDR data showed in the beginning
grey, blue dotted), and difference (blue).
quite a large difference up to 10,000 feet/
minute. This was possibly due to the fact
that the aircraft was only partially visible
in the video and that the recording rate of
the security camera was a relatively low
one frame per second, as shown in Figure
15.
Since the video could be synchro-
nized with the time base, using the video
frame time stamps, a reconstruction of
the ground speed and descent rate was
possible.
Based on the calculated decent rate
Figure 15. First frame of video showing aircaft only partially. (black) in feet/minute, Figure 16 shows
12 • January-March 2021 ISASI Forum“Flight
data that is
reconstructed
based
on video
Figure 16. Calculated vertical speed (black), recorded FDR vertical speed (light blue), and
information is
difference value (blue). applicable!”
that the aircraft reduced its descent rate
from -39.800 feet/minute to -15.000 feet/
minute. The recorded FDR data showed
that the B-767 reduced its decend rate
from -28.000 feet/minute to -18.240 feet/
minute within 2 seconds. The difference
between the reconstructed data and re-
corded FDR data showed in the beginning
quite a large difference up to 10,000 feet/
minute. This was possibly due to the fact
that the aircraft was only partially visible
in the video and that the recording rate of
the security camera was a relatively low
one frame per second, as shown in Figure
Figure 17. Calculated ground speed (black), FDR ground speed (light blue), and difference 15.
value (blue). Figures 18 and 19 show the reconstruct-
ed attitude data for heading (dotted lines),
roll (blue), and pitch (dotted lines). The
calculated data is shown as dotted lines.
The heading, pitch, and roll values fit
very well with an offset of less than +/- 10
degrees.
Conclusion
The reconstruction of flight data based on
video information is capable, and the data
is applicable. The accuracy of the recon-
structed data depends on the location
Figure 18. Reconstructed heading attitude (dotted lines) and recorded FDR heading (full accuracy of the video source position
line). and reference objects, as well as on the
frame rate of the video and its resolution.
Further, the attitude data could be recon-
structed very well within the experienced
accuracy.
The more video sources that are aviala-
ble, the better the approximated positions
can be calculated and cross-checked.
In parallel to the increase of digital data,
more and more video footage is available.
The capability of this method can support
to reconstruct data, based on video
information, but also the combination of
recorded data and video information can
Figure 19. Reconstructed pitch/roll data (dotted lines) and recorded FDR data (full lines). provide additional important details.
January-March 2021 ISASI Forum • 13Why Did the Helicopter Collide
With Trees?
Investigating the Causes from
Analysis of Images and Sounds
By Koji Fukuda, Deputy Investigator for Aircraft Accidents,
Japan Transport Safety Board
C
urrently, 75 helicopters are operated Onsite Investigation
by 55 local governments in Japan The three investigators went to the site the
for the purpose of firefighting and next day. The accident site was on a snowy
disaster prevention. For similar mountain, and the roads leading to the site
purposes, there are other services such as were not cleared. The large rescue vehicle
emergency medical service (EMS), police, the had reached near the site. Our rental car had
Coast Guard, and the Self Defense Force. The studless tires, so I thought it was fine, but it
firefighting disaster prevention helicopters was useless. I had to push the car many times
are operated under guidelines made by each on a road like an inclined ice rink that was
local government independently. They are hard to walk.
single-pilot operations even though they are
medium-sized helicopters of about 5 tons. Lesson 1: Select a four-wheel-drive car on snowy
The pilots are not required to have instrument roads. Not enough with studless tires.
flight licenses. And crews are not transferred
among other local government teams. This
case is an example of a firefighting and disas-
Drone Investigation
This JTSB accident investigation was the first
ter prevention helicopter accident in which
time the agency used a drone to collect onsite
all nine people were killed. The helicopter was
data. The drone was used not only to take
not equipped with a recording device such as
photographs from above the crash site, but
a black box, so the investigation required con-
also allowed the investigators to create one
siderable effort to determine what occurred
fine 2-D image (Ortho Mosaic) or 3-D images
and why it happened.
by analyzing the image data taken contin-
There was a mysterious video taken on
uously and the data of the position and the
the aircraft that seems to show that when
altitude of the drone.
approaching trees, the helicopter collided
with the trees without avoidance. To get a
Lesson 2: The effectiveness of the drone for
real sense of the accident, investigators were
accident investigation is not just to obtain aerial
required to conduct a drone investigation and
photos.
to analyze the images and sounds recorded in
the video in various novel ways.
Location Survey (Laser Range Finder and
Summary of the Accident GPS Receiver)
(Adapted with permission from the On Sunday, March 5, 2017, a mid-sized rescue The position and height of the cut trees could
author’s technical paper Why Did helicopter, a Bell 412, operated by the Nagano be measured easily and clearly by using a
the Helicopter Collide with Trees? Fire and Disaster Prevention Aviation Center laser range finder (TruePulse 360, Laser
Approach the Causes from Analysis took off from Matsumoto Airport to go to Technology) and a GPS receiver (MMCX:
of Images and Sounds presented rescue training and collided with trees and Mobile Mapper CX, Magellan Navigation). As
during ISASI 2019, Sept. 3–5, 2019, in described in the report, these tools enabled
crashed into the mountain’s slope while flying
The Hague, the Netherlands. The theme
toward the training site near the summit the investigators to see more details of the
for ISASI 2019 was “Future Safety:
Has the Past Become Irrelevant?” at 13:33. There were nine persons on board accident site than what was available from
The full presentation can be found on the helicopter, including a captain and eight the drone investigation.
the ISASI website at www.isasi.org others, and all of them suffered fatal injuries.
in the Library tab under Technical The helicopter was destroyed, but there was Lesson 3: Easy location survey with laser and
Presentations.—Editor) no outbreak of fire. GPS. (Of course, you need tools and training.)
14 • January-March 2021 ISASI ForumImage Analysis of Video Taken on the with trees when the video recording stopped
Helicopter with the impact of the helicopter crashing
Since the helicopter was not equipped with into the ground.
a recording device such as a black box, it
was difficult to determine fact information. Lesson 4: Analysis of flight tracks from video
However, the camera attached to the rescuer’s taken on the helicopter.
helmet was taking a video of the situation
from the middle of flight until the accident. Detailed Investigation of the Helicopter
The video shows that the weather at the time From the scattering of debris from the acci-
did not interfere with the flight. dent site, it was estimated that the helicopter
The helicopter gradually approached the hit a tree and became uncontrollable and
mountain surface covered with trees at a crashed. The investigation of the details of
constant attitude and speed and crashed into the airframe was conducted in June when
trees. That the helicopter was flying normally the snow melted and the debris was salvaged
and then, without an avoidance maneuver, from the mountain, and no anomalies of the
collided with trees as it gradually approached airframe and engine were found from the
is highly unusual. What happened? Nothing debris.
happened. Why did the pilot fail to avoid? This
was a shocking question that I had to keep
asking until the end of the investigation.
Voice Analysis of Video Taken on the
Estimating the position and altitude of the Helicopter
helicopter over time by analyzing still images An alarm sound and sounds indicating ab-
with the video stopped and then estimating normality of the helicopter were not recorded
the flight path by linking the images was until it collided with trees. A sheet recorded by
possible. We knew that the helicopter took a mechanic sitting in the left seat was found,
off from the airport, headed northeast while which revealed that the crew of the helicopter
climbing above the city, entered the airspace had been performing engine performance
above the mountains, and turned right. It is tests after takeoff.
highly probable that it headed toward Mt. Furthermore, by combining and analyzing
Hachibuse in continuing roughly level flight this one sheet and the in-flight video and
at a speed of about 100 knots. The helicopter’s audio, the in-flight situation became clearer.
altitude above ground level (AGL) became At first, I had no idea what the crewmem-
lower gradually, and the tree-covered moun- bers were talking about, but it became clear
tainside was looming ahead. The helicopter that they were checking the engine, and the
collided with trees while maintaining attitude content of the conversation became more un-
and speed. derstandable. Furthermore, by analyzing the
The helicopter leveled off at about 1,740 engine noise, it was possible to estimate the
meters. It is somewhat likely that this was be- throttle and engine N2 governor operation
cause the helicopter was trying to ensure the status during engine check, which supported
safety altitude of 150 meters or higher from the fact that an engine check was performed.
the destination, the helipad, with an elevation An analysis of the audio recorded by the
of about 1,580 meters. It is highly probable video camera found that a spectrum of ap-
that while maintaining the maximum safety proximately 22 Hz was recorded at a constant
altitude, the helicopter took neither the avoid- frequency from the beginning of the video
ance route at a constant altitude by directly until 4.0 seconds before the audio recording
heading for the helipad nor the avoidance stopped. Assuming that the sound was gener-
route by climbing. Instead, it continued to fly ated by the MRB, this would be equivalent to
toward Mt. Hachibuse at a constant altitude approximately 330 rpms. The 100% number of
after turning right. Its AGL became lower MRB revolutions is 324 rpms.
while flying into and over the mountains An audio spectrum of approximately 3,300
region, and the helicopter approached the Hz and an audio spectrum of approximate-
ground. It is highly probable that the helicop- ly 3,400 Hz were transmitted at a constant
ter flew into an uncontrollable condition as frequency, respectively. However, immediately
it crashed its fuselage and main rotor blade after the voice of “minus two,” in the former
(MRB) into trees over a distance of approxi- case, the frequency increased by approximate-
mately 40 meters. It is highly probable that the ly 200 Hz, while in the latter, the frequency
helicopter turned upside down and collided decreased by approximately 200 Hz. Those
with an approximately 40-degree slope from frequencies returned to original frequencies,
its nose. It is also highly probable that it was and they were constantly transmitted again, Koji Fukuda
four seconds later after the helicopter collided immediately after the voice saying, “I return it.”
January-March 2021 ISASI Forum • 15It is highly probable that there were no When the CP lever was moved up to the
abnormalities in the helicopter’s engines from same position as at the time of climbing, and
the time of takeoff until transition to level the right hand was extended to the ITT trim
flight. As the engine had been operating when switch, almost the same composition as in
the helicopter crashed, it is highly probable the image of video camera (-4 minutes and 5
that the MRB had been rotating at constant seconds) was reproduced.
rpm until the helicopter collided with trees.
It is highly probable that the helicopter was 2) View from the cockpit
conducting engine data checks enroute from The inspection was conducted after parking
takeoff. When the helicopter commenced the same type of helicopter with a magnetic
turning right above the mountains, it is prob- heading of 150 degrees, the same heading as
able that engine checks had been completed. at the time of accident. It was conducted at
It is highly probable that the mechanic was about 13:40 on April 10, 2018; however, the
conducting engine checks, concentrating on pilot’s face was not exposed to direct sun-
the flight instruments, and hardly watched shine. The pilot’s view was not blocked, and
outside. And it is somewhat likely that his it was possible for him to recognize visually
attention was focused on addressing the en- the obstacles lying ahead in keeping the
gine data check records even after the engine piloting posture, when either only glancing at
checks were completed, but this could not be the instruments without moving his head or
specified. facing the instruments and looking at them.
Since the engine check was finished 2 min- However, when the body was bent forward
utes before the collision with trees, it was diffi- and the head was lowered a little, the glare
cult to link the relationship with the collision.
shield blocked the forward view (the horizon).
Therefore, it seems that when the obstacles
Lesson 5: Analysis of human voices requires
lying ahead are approaching, the approaching
understanding what the person is doing.
obstacles may not be recognized visually. With
the posture of the mechanic mentioned above
Lesson 6: We can understand the situation of the
(1), the forward view was blocked because the
engine by analyzing the sound.
position of the head was lowered.
When looking at the pilot’s face with the
Captain’s Helmet Visor visor lowered, the opening state of eyes and
From the fact that the captain’s helmet visor his facial expressions were not recognized.
had an impact mark near the center and ap-
proximately half of its right side was missing, 3) Different views with and without visor
although the visor cover was not broken, it The different views with and without the visor
is probable that the visor received an impact that is attached to the helmet was confirmed.
from the right side while in a lowered state. Without the visor, the contrast between
From the fact that the captain’s visor was outside and inside the helicopter was clear,
raised at the time of takeoff, it is probable that and when looking outside, it seemed that the
the captain lowered his visor while in flight. view was too bright. Immediately looking at
The captain’s right upper arm moved 1 minute the instruments inside the helicopter, it was
and 30 seconds before the collision with trees. not to say that anything could not be seen but
It is somewhat likely that it was because turn- eyes seemed tired. With the visor lowered,
ing right at that time would have the helicop- the outside view was clear and not too bright.
ter face the direction close to the sun, and the
Immediately looking at the instruments inside
captain lowered his helmet visor to ward off
the helicopter, it seemed a little dim, but the
the glare of the sun; however, this could not be
instrument indicators were readable.
specified.
With the visor lowered, the outside view
Lesson 7: Verification will require new discover-
was clear and not too bright, and the instru-
ies or proofs of certainty.
ment indicators were readable. Therefore, it is
probable that the use of the visor had no effect
on flying the helicopter. However, with the ELT
visor lowered, the opening state of eyes and The helicopter was equipped with an emer-
the facial expressions were not recognized gency locator transmitter (ELT) with switches
from outside. (G switches) designed to automatically acti-
vate with impact from six directions. When an
examination was conducted by an agent of the
Verification by Same Type Helicopter manufacturer following the accident, it was
1) Position of the mechanic’s right hand found that the ELT had not activated in this
16 • January-March 2021 ISASI ForumYou can also read
