ISASI Forum April-June 2008.Pmd

APRIL–JUNE 2008

In Memoriam PRINCIPLED PROFESSIONALISM Ronald “Ron” Chippindale (1933-2008) ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ CONTENTS

FEATURES Volume 41, Number 2 Publisher Frank Del Gandio 3 Ron Chippindale ‘Flies West’ Editorial Advisor Richard B. Stone By Frank Del Gandio—This “President’s View” memorial pays tribute Editor Esperison Martinez to one of ISASI’s staunchest advocates. Design Editor William A. Ford Associate Editor Susan Fager Annual Report Editor Ron Schleede 6 Identifying CVR Recorded Sounds and Voices By Yang Lin, Wu Anshan, and Liu Enxiang—Sound identification and voice ISASI Forum (ISSN 1088-8128) is published quarterly by International Society of recognition are aimed at offering more clues in the analysis and classification of Air Safety Investigators. Opinions ex- speech and non-speech CVR signals. pressed by authors do not necessarily rep- resent official ISASI position or policy. Editorial Offices: Park Center, 107 East 11 Reverse Engineering Overcomes Corporate Amnesia Holly Avenue, Suite 11, Sterling, VA 20164- By Peter Coombs—A process of “corporate amnesia” has become common 5405. Telephone (703) 430-9668. Fax (703) 430-

○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ 4970. E-mail address [email protected]; for edi- among manufacturers, brought about by lengthening aircraft service lives and tor, [email protected]. Internet website: shortening career spans of design/development engineers within one employer. www.isasi.org. ISASI Forum is not responsible for unsolicited manuscripts, photographs, or other materials. Unsolicited materials will be 15 Critical Aspects of International Incident Investigations returned only if submitted with a self-ad- dressed, stamped envelope. ISASI Forum By Deborah J. Lawrie, Robert N. van Gelder, and Jan Smeitink—The very fine reserves the right to reject, delete, summa- line between incident and accident clearly emphasizes the importance of having rize, or edit for space considerations any sub- well-trained airline investigators. mitted article. To facilitate editorial production processes, American-English spelling of words will be used. 20 Flight Data—What Every Investigator Should Know Copyright © 2008—International Society By Michael R. Poole and Simon Lie—The correct interpretation of flight data and/or of Air Safety Investigators, all rights re- served. Publication in any form prohibited audio data requires a full understanding of the entire signal path from measurement without permission. ISASI Forum regis- to recorder to investigator. tered U.S. Patent and T.M. Office. Opinions expressed by authors do not necessarily rep- resent official ISASI position or policy. Per- mission to reprint is available upon applica- DEPARTMENTS tion to the editorial offices. 2 Contents Publisher’s Editorial Profile: ISASI Fo- 3 President’s View rum is printed in the United States and pub- 24 ISASI RoundUp lished for professional air safety investigators who are members of the International 30 ISASI Information Society of Air Safety Investigators. Edito- 32 Who’s Who—A brief corporate member rial content emphasizes accident investiga- profile of WestJet tion findings, investigative techniques and experiences, regulatory issues, industry accident prevention developments, and ISASI and member involvement and information. Subscriptions: Subscription to members is provided as a portion of dues. Rate for nonmembers is US$24 and for libraries and schools US$20. For subscription information, call (703) 430-9668. Additional or re- placement ISASI Forum issues: members US$3, nonmembers US$6.

ABOUT THE COVER In Memoriam: Principled Professionalism—Ronald “Ron” Chippindale (1933-2008)

2 • ISASI Forum April–June 2008 PRESIDENT’S VIEW

Ron Chippindale ‘Flies West’ By Frank Del Gandio, ISASI President

“To fly west, my friend, is a flight we all must ○○○○○○○○○○○○○○○○○○○○○○○○ Ron, born in England on March 26, 1933, enlisted in the Royal take for a final check.”—Author unknown New Zealand Air Force (RNZAF) in 1950 and was one of the first two RNZAF trainees selected to train with the Royal Air On February 12 at 7:25 a.m., one of ISASI’s Force at Cranwell in the U.K. When he returned to New staunchest advocates began his “flight west.” Zealand, he brought back his bride, June. Ronald “Ron” Chippindale was doing his daily early morning walk along a footpath near his home in the Aotea subdivision of Porirua, Wellington, New Zealand, when an 18-year-old driver lost control of his car and struck Ron. He died instantaneously. Ron joined our Society in 1971 and his involvement in ISASI was nothing short of phenomenal. He was instrumental in forming the New Zealand Society and became its first president, serving in that post until 2004. He was the NZSASI councillor up to his death. Ron attended every scheduled International Council meeting and was very involved and instrumental in the Council’s policy-setting role. He met his representation responsi- ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ bilities with zealous constructive participation in debating issues and ensuring alternative options were considered. Ron chaired the Organizing Committee for the 1986 ISASI seminar that was successfully held in Rotorua, New Zealand, and again in Auckland in 1996. In 1993, he became one of the first to attain ISASI Fellow status. In 2002, he was granted “Life” membership. In 2004, he was presented with ISASI’s highest honor, the

Ron accepts the ISASI 2004 Jerome F. Lederer Award from on gained international attention and Frank Del Gandio, ISASI president. Rrespect in 1979 as the investigator- Ron flew Bristol freighters with the No. 41 Squadron and in-charge of the investigation into the tragic later Hastings and DC-6 aircraft with the No. 40 Squadron. He Air New Zealand DC-10 accident in which had operations service during the Malayan Emergency and in 257 persons died when the aircraft descended Borneo. into an ice field near Mt. Erebus in Antarctica. He has served in a flight safety officer capacity since 1959, and in 1969 Squadron Leader Chippindale became the New His work on the case has been described as Zealand Defense Headquarters flight safety officer responsible nothing short of brilliant. for the air staff policy on flight safety and the investigations of accidents for the Army, Navy, and Air Force. When he retired Jerome F. Lederer Award, for his outstanding lifetime contribu- from the Air Force in 1974, after 24 years of service, he literally tion in the field of aircraft investigation and prevention, which is moved next door to be an inspector of air accidents with the outlined below. Office of Air Accidents Investigations, a part of the Ministry of Ron’s acceptance speech exemplified the characteristics of his Transport. He became its chief inspector in 1976. demeanor and accident investigative manner: short on banter and In 1990, the Transport Accident Investigation Commission long on meaningfulness. He said, “I have been in awe of those who (TAIC) replaced the Office of Air Accidents Investigations. Ron have been nominated for the coveted Jerry Lederer Award. To was the acting chief executive for 2 years as well as the chief have myself selected for this honor is rather overwhelming.” (See investigator of accidents. As the mandate of the Commission ISASI Forum October–December 2004, page 14). expanded to include marine and rail, Ron adopted aviation investigation methodologies for those modes. (Peter Williams, president of NZSASI, and Steven Lund Ron led or was involved in the investigations of more than contributed to this memorial.) 2,000 aircraft accidents and incidents, which made his name

April–June 2008 ISASI Forum • 3 PRESIDENT’S VIEW Continued . . . Ron’s presence will be sorely missed by his family,

synonymous with aircraft accident investigation. He gained ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ international attention and respect in 1979 as the investigator-in- Peers Honor Memory charge of the investigation into the tragic Air New Zealand DC- 10 accident in which 257 persons died when the aircraft de- At press time, three groups within ISASI have honored scended into an ice field near Mt. Erebus in Antarctica. His work the memory of Ron Chippindale with donations to the on the case has been described as nothing short of brilliant. ISASI Rudolph Kapustin Memorial Scholarship Fund in Many compare the complexity of the investigation to that of the Ron’s name. The Fund sponsors the annual scholarships later TWA Flight 800 investigation. With a very small team, he awarded to college students. The sentiments of the managed an investigation in a very difficult environment, and contributors are well expressed in this comment from the encountered both political and cultural stresses. His main Pacific Northwest Regional Chapter: “All of us who knew finding was challenged through a one-man Royal Commission of Ron Chippendale—and there are many of us—were Inquiry, but the report of the New Zealand Office of Air Acci- deeply saddened by his unexpected and untimely death. dents Investigations stands as the official report of the accident. We mourn along with Ron’s wife June, his family, and his Throughout the ensuing controversy about the report’s findings, many, many friends.” Ron steadfastly expressed himself and stood by his principles on Units making donations include the Pacific Northwest behalf of air safety. Regional Chapter—$500, the Canadian Society of Interna- Ron was known by his fellow investigators as being a very tional Safety Investigators—$500, the European Society of dedicated, thorough, and impartial professional—one who did International Safety Investigators—$500, and the Mid-At- not hesitate to assuredly speak his mind, even in the face of lantic Regional Chapter—$500. At press time, the New staunch opposition. Zealand and Australian Societies were determining the na- Recognizing Ron’s skills, ICAO developed a long-standing ture of their memorial. ◆ relationship with him. In 1986, he worked with the ICAO Technical Cooperation Bureau, assisting in the South African ○○○○○ investigation where a Mozambique Tu-134 aircraft crashed and KAL Flight 007, a B-747, over the Sakhalin Islands, ICAO the president was killed. In 1993, when the Russian Federation assigned him to the team in the reopened investigation. Ron has finally made available the fight recorders from the shoot down of also been an enthusiastic supporter of the ICAO AIG meetings.

Ron (second from right) makes his New Zealand councillor’s report to the ISASI International Council at its September 2004 meeting.

4 • ISASI Forum April–June 2008 friends, ISASI, and the aviation community. ○○○○○○○○○○○○○○○○○○○○○○○○

Following the presentation of the New Zealand Service Medal from Police Minister Annette King (center), Ron accepts congratulations from Police Commissioner Howard Broad. The ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ Medal was presented for his IIC leadership of the accident investigation into the DC-10 that descended into an ice field near Mt. Erebus in Antarctica in 1979.

In 1999, he was an ICAO consultant assisting the secretariat in the conduct of that year’s meetings. He served several times as a consultant assisting in various projects, including the development of the ICAO circular on family assistance and enhance- ment to the ICAO ADREP data system. Ron was a member of the Returned and Services Association for military veterans and a Fellow of the Royal Aeronautical Society. He was also the recipient of the New Zealand 1990 Ron in a 1992 photo at his TAIC chief’s desk. Commemorative Medal for his services as chief inspector of air accidents and in 2007 received the New Zealand Special Service (Erebus) Medal, awarded to those involved in the Erebus accident recovery and investigation. Among the 280 people saying a final farewell at his funeral service was the New Zealand minister for Transport Safety and representatives from the Air Force, the aviation industry, and public service. Ron’s family members gave moving accounts of a loving father and grandfather. Capt. Tim Burfoot, chief investigator of the Transport Accident Investigation Commission, gave an overview of Ron’s accident investigation career. That was followed by eulogies from Paul Mayes, on behalf of ISASI and ASASI, Peter Williams on behalf on NZSASI, and a personal tribute from Milton Wylie, a friend and co-worker of many years. Ron is survived by his wife June, four children, 10 grandchil- dren, and one great granddaughter. His presence will be sorely missed by his family, friends, ISASI, and the aviation community. Ron (second from right) is among the five ISASI Fellows present As the last post and reveille were sounded, a formation of at ISASI 2005. Shown, left to right, are Ron Schleede (2002), three Royal New Zealand Air Force aircraft made a fly past with Caj Frostell (2003), John Rawson (2005), Ron (2004), and John trailing smoke. Purvis (2001). A fitting send off to his “flight west.” ◆

April–June 2008 ISASI Forum • 5 Identifying CVR Recorded Sounds Sound identification and act effectively as a latent signal transducer are simply generated by cockpit controls voice recognition are aimed for both speech and non-speech audio in- and the sound of the aircraft moving formation. Some typical techniques, such as through the air. at offering more clues in sound identification and voice recognition, Analysis background information re- the analysis and classification appear to offer significant clues in the analy- corded in aircraft CVRs has been proposed of speech and non-speech sis and classification of speech and non- as a complement to the analysis of onboard CVR signals. speech CVR signals. FDRs in civil aircraft investigations. One The CVR records audio information on reported case provides a good example of By Yang Lin (AO5012), Center of four channels. Non-speech information the analysis of CVR data playing a key part Aviation Safety Technology (CAST) from the cockpit area microphone (CAM) in an aircraft accident investigation. In 1992, of CAAC, and Wu Anshan and is recorded on Channel 1. CAM records a 19-seater commuter aircraft crashed dur- Liu Enxiang, General Administra- thumps, clicks, and other sounds occurring ing an evening training mission. At that tion of Civil Aviation of China in the cockpit other than speech. Channel 2 time, the U.S. Federal Aviation Agency (CAAC), Air Safety Office and 3 of the CVR record speech audio in- (FAA) did not require the installation of an formation from the captain’s and first FDR on board all small commercial air- (This article was adapted, with permission, officer’s audio selector panels. Channel 4 crafts, and the CVR on board the small jet from the authors’ presentation entitled Sound records the audio information from the that crashed was the only flight record avail- Identification and Speaker Recognition for jumpseat/observer’s radio panel. able to provide clues to the causes of the Aircraft Cockpit Voice Recorder, presented at accident. the ISASI 2007 seminar held in Singapore, Background cockpit sound Fortunately, in this case, the CVR record- Aug. 27-30, 2007, which carried the theme “In- identification ing not only included the voice communica- ternational Cooperation: From Investigation While it may be hard to believe that non- tion, but also structural acoustics as well as Site to ICAO.” The full presentation includ- speech sounds are highly important to the other sounds and noise sources. This al- ing cited references index is on the ISASI investigation of aircraft damage, they are lowed the accident investigation to focus on website at www.isasi.org.—Editor) because the background cockpit sounds can the non-speech sounds taken from the CVR reveal problem areas of the aircraft during tape. A close inspection of the time series raditionally the flight data recorder the time leading up to the accident. Non- from the CVR track revealed a periodic set (FDR) has played the major role in speech data from the CAM can be analyzed of transient components occurring at a fre- Testablishing the causes of most acci- with sound spectrum analysis to detect quency of 0.86 Hz. Comparing this fre- dents or incidents. However, information whirl flutter, as well as possibly distinguish quency with an independent dynamic analy- contained in the cockpit voice recorder the sound of a bomb explosion from the sis of the engine mount damage, the 0.86 (CVR) is also very useful during such in- sound of cabin decompression. Spectrum Hz transient data were demonstrated by vestigations by providing a better under- analysis can also be used to confirm that independent structural and flutter analy- standing of the real situation. The CVR can the clicks and thumps recorded by the CAM ses to be quite close to the frequency expe-

Yang Lin has more than recorder readouts on behalf of govern- Liu Enxiang is the 12 years’ experience in ment authorities and airlines in China deputy director at the the field of aviation safety and other countries. Photo CAAC’s Office of Aviation and is an ISASI member. unavailable Safety. An 18-year veteran Currently she is an air Wu Anshan is a senior of CAAC, he has partici- safety investigator (ASI) ASI with CAAC, having pated in several national and senior engineer at Photo joined the agency in aircraft accident investi- the Civil Aviation Safety Technical unavailable 1970. He investigates the gations as an investigator-in-charge (IIC) Center (CASTC) of Civil Aviation of air traffic control and and as an expert in investigation of avion- China (CAAC), where she has been organizational aspects ics equipment. He graduated from Nan- working since 1994. While at CASTC, she of aviation accidents jing University of Aeronautics and Astro- analyzes and presents recorded data in and incidents and has substantially nautics with a degree in aeronautical support of accident/incident investiga- contributed to more than 20 significant engineering and has commercial aviation tions and also conducts flight data investigations. aircraft maintenance experience.

6 • ISASI Forum April–June 2008 and Voices rienced from a damaged engine mount. in “Method and Article of Manufacture for the process dramatically when audios are Moreover, there was a sudden, loud Content-Based Analysis, Storage, Re- in bad quality. sound at the end of the tape. This 25-milli- trieval, and Segmentation of Audio Infor- A. Kimura adopted a two-pass search second-long event was much louder than the mation.” “Content-Based Identification of strategy. He generated the vector codebook sound in the cabin. Although this short Audio Material Using mpeg-7 Low-Level of audio features first and obtained the dis- length of the sound did not provide adequate Description” takes the spectral flatness tribution of vector code within a period of audio listening time, there was enough sig- measure as feature parameters. C. audio. This distribution is then compared nal time and amplitude to perform wavelet Papaodysseus takes the choice of band rep- to that of a given audio. This comparison is and voice recognition analysis. The conclu- resentative vectors, which is an ordered list regarded as the first rough search. The au- sion drawn after the investigation was that the engine on the starboard wing separated during the flight. Subsequently, the fallen engine struck the tail of the aircraft, dam- aging most of the horizontal surfaces. The loss of the engine also led to the separation of the right wing panel outboard of the engine. As a result, the aircraft pitched down, rolled to the right, and crashed. The results of the accident investigation Figure 1. Framework of described above, and that of Pan Am Flight Background Cockpit Sound 103, which disintegrated over Lockerbie, Identification System. Scotland, in 1989 due to a bomb explosion, motivated CASTC to explore the analysis of aircraft CVR sound sources for use in of indexes of bands with significant energy. dios with similar distribution will continue aircraft accident investigations. In the approaches cited above, the audio sig- to go in for the next fine match. Since the In our system, background cockpit nal is segmented into frames first and a set distribution comparison is processed by a sound identification is used to find the au- of features are extracted frame by frame. block of frames instead of frame by frame, dios identical to the given audio in the back- Some approaches compute a block of fea- it can search audios quickly. The beam- ground cockpit sound database. An audio tures from a big segment of audio. based search approach cuts off branches ID is identified by audio fingerprint in The background cockpit sound feature whose cumulative match scores are higher acoustics. An audio fingerprint is com- extraction approach proposed in this article than the beam width from the best score. posed of a series of audio features. Gener- is based on “Robust Audio Hashing for Con- Plenty of unpromising paths are pruned ally, audio feature design should obey the tent Identification.” It takes into account away during the search process. The search following guidelines, as noted in J. Haitsma that the sign of spectral band energy dif- space is reduced dramatically and high ef- and T. Kalker’s “Robust Audio Hashing for ference (both in time and frequency axis) is ficiency is achieved. Content Identification”: very robust to many kinds of processing. • Discriminability: sensitive enough to dis- By analyzing this approach carefully, we Framework of Background tinguish the different audios. propose an improved approach, one that Cockpit Sound Identification • Robustness: stable enough to various enhances the robustness of the audio fea- System codes, channels, and modest noise. ture significantly. When audio features are Figure 1 shows the framework of the Sys- • Compactness: small enough for easy ready, the below outlined Background Cock- tem. It is composed of three modules: fea- storage. pit Sound Identification System (BCSIS) ture extraction, audio search, and audio • Simplicity: inexpensive to compute and will be able to search identical audios in the database. When audio signals are fed into easy to implement. database quickly and effectively. the System, it extracts audio features first. There are many approaches available to J. Haitsma presented a method that lo- Audio features are compared to the features extract audio features. Mel Frequency cates the potential identical audios by a hash in the audio database. Audio candidates are Cepstral Coefficients (MFCC) are used as table. It is quite efficient. But it needs about generated according to the result of the audio features in “Robust Sound Modeling four times or more space to save data com- match process. for Song Detection in Broadcast Audio,” and pared to other methods, and it slows down The feature extraction module does

April–June 2008 ISASI Forum • 7 Figure 3. Speaker Identification and Speaker Verification System. some preliminary processing, such as down In order to make the audio feature stable, whose scores are higher than the base plus sampling and low band pass filtering. Then a frame length as long as 410 milliseconds the empirical threshold (beam width). it computes the audio features using the is chosen. Frame shift is only 12.8 millisec- algorithm described below. The audio fea- onds. As a result, the frame boundaries of Automatic speaker recognition ture database stores the audio features audio queries in the worst case are 6.4 mil- Automatic speaker recognition automati- computed in advance. The audio search liseconds off from the boundaries used in cally extracts information transmitted in the module compares the features of possible the database that are precomputed. speech signal, which can be classified into identical audios and outputs the best can- identification and verification, and identifies didates. Audio search approach—In the audio fea- a speaker based on his or her voice in the ture similarity measurement, each frame has CVR recording. Speaker identification is Audio feature extraction approach —Fig- one 32-bit audio feature. The similarity of two the process of determining which registered ure 2 shows the audio feature extraction flow features is measured by the Hanning dis- speaker provides a given utterance. chart adopted. Because human hearing is tance, which is the number of different bits. Speaker verification is the process of accept- most sensitive to the frequencies below 2,000 The smaller the Hanning distance, the more ing or rejecting the identity claim of a Hz, high-frequency similar the two features are and vice versa. speaker. Speaker recognition methods can parts lose heavily Bit error rate (BER) defines the similarity also be divided into text-dependent and when audios are of two audio feature serials with the same text-independent methods. The former re- encoded at very length. Let X, Y are two audio feature seri- quires the speaker to say key words or sen- low bit rates. Ac- als, X={x1, x2,…, xN}, Y={y1, y2,…, yN}. tences having the same text for both train- cordingly, in this Where N is the frame number of the fea- ing and recognition trials, whereas the lat- System audio sig- tures. The BER between X and Y is ter does not rely on a specific text being nals are down spoken (Figure 3). sampled to 5,000 Current state-of-the-art systems for text- Hz first. Then sig- independent speaker recognition use fea- nals are segmented tures extracted from very short time seg- into frames and ments of speech and model spectral infor- weighted by a ham- Where, H(.) is the Hanning distance be- mation using Gaussian Mixture Models ming window. Fou- tween X and Y. Obviously, ≤BERd≤, the (GMM). Using a Universal Background rier transforma- lower the BER is, the more similar the two Model (UBM)/GMM-based system is now tion is performed, feature serials are. compulsory to obtain good performance in and spectrum pow- evaluation campaigns such as the U.S. Na- er is obtained. A Beam-based search approach—When tional Institute of Standards & Technology total of 33 over- searching audio candidates in the audio data- (NIST) Speaker Recognition Evaluation lapped frequency base, it would result in very low efficiency if a (SRE). bands are used at whole match comparison is processed at ev- NIST has conducted an annual evalua- an equal logarithm ery possible starting frame. A beam-based tion of speaker verification technology interval. A 32-bit search strategy is presented in this System since 1995. Such an approach, while suc- audio feature is to avoid low efficiency. The main idea of this cessful in matched acoustic conditions, suf- computed for each Figure 2. Feature approach is that it takes the current best score fers significant performance degradation frame. extraction flow chart. as the base and prunes away all branches in the presence of ambient noise. Some

8 • ISASI Forum April–June 2008 Figure 4. Framework of Automatic Speaker Recognition System.

Figure 7. Framework of Automatic Speaker Recognition System.

methods are proposed to compensate for on longer-range stylistic features provide environment is restrained by robust tech- channel variation and intra-speaker varia- significant complementary information to niques. During the speaker modeling pro- tion by normalization techniques such as the conventional system. Another impor- cess, input front-end features characterize the Cepstral Mean Subtraction (CMS), tant issue in the statistical approaches to the speaker. The GMM or SVM modeling feature warping, feature mapping, and speaker recognition is that of score normal- approach is used to train the target speaker joint factor analysis. ization, which covers aspects such as the models, which compose the speaker model Modeling of spectral information by scaling of likelihood scores. database. There are two main aspects of speaker features. Speaker feature extraction approach First, the physiologic structure is different for The speech signal is smoothed in short time. each individual, such as the track length and oral For analyzing the speech signal, the usual frame concept is introduced by shortening cavity structure, so the short-time spectral is the speech segment by 10 ms-30 ms. The different. Second, the uttered habits are different, shift length is the half length of one frame. To compensate for the attenuation of the such as an accent. high frequency, every frame of the speech GMM can be improved or complemented Framework of Automatic Speaker signal uses pre-weighted processing. by the use of other modeling techniques like Recognition System There are two main aspects of speaker Support Vector Machines (SVMs) or by Figure 4 shows the framework of the Sys- features. First, the physiologic structure is transformations of the cepstral space. How- tem. It is composed of three modules: fea- different for each individual, such as the track ever, short-term cepstral modeling fails to ture extraction, speaker modeling, and length and oral cavity structure, so the short- capture longer-range stylistic aspects of a speaker recognition. When audio signals are time spectral is different. Second, the uttered person’s speaking behavior, such as lexical, fed into the System, the speaker features habits are different, such as an accent. It can rhythmic, and intonational patterns. Re- are drawn from the input speech segments. be described as prosody features. In the field cently, it has been shown that systems based Furthermore, the influence of channel and of speech signal processing, the former is

April–June 2008 ISASI Forum • 9 Figure 5. Flow of speaker recognition.

embodied on the structure of frequency. The ponent GMM. The background GMM is speaker identification. The speaker verifi- classical features include cepstral and pitch. trained using data from the NIST 1999 and cation is determined by whether the test And the latter is embodied on the variability 2001 speaker recognition evaluation. The speech segment is uttered from the given of the speech based on the spectral struc- features are mean and variance normalized target speaker or not. ture. The classical features include the delta over the utterance. For channel normaliza- The result of recognition is “YES” or cepstral and delta pitch. tion, the feature mapping is applied using “NO,” and the comparison happens between In speaker recognition, the cepstral is gender- and handset-dependent models one segment and one fixed speaker. The used mostly and could achieve a good per- that are adapted from the background speaker identification is that given the test formance. Besides, it can be extracted more model. Target GMMs are adapted from the speech segment. easily than other features. At present, the background GMM using a Maximum a Pos- The system needs to choose the true Mel Frequency Cepstral Coefficients are teriori (MAP) algorithm adaptation of the speaker from the speaker models database. used successfully in speaker recognition, means of the Gaussian components. The The key function is calculating the log like- which is proven in applications. In feature resulting scores are T-normed. lihood of the input test speech features and extractors of speaker systems, all of the fea- Cepstral SVM system—The Cepstral one target speaker model. Its calculated ture vectors are processed by CMS and the SVM system is based on the cepstral se- method is denoted as follows: ()=−λλ() feature warping method. quence kernel proposed by “The Contribu- SXlog pX() |hyp log pX | UBM Using the delta cepstral information tion of Cepstral and Stylistic Features to Where SX() is the final output score, λ based on a time domain proves that the per- SRI’s 2005 NIST Speaker Recognition pX()| hyp is the probability of the formance of speaker recognition is mostly Evaluation System.” All of them use basic speech segment based on the hypothesis ()λ enhanced. In our system, speech data are features, which are similar to the cepstral model; pX| UBM is the probability parameterized every 25 ms with 15 ms over- GMM system. The only difference is that of the speech segment based on UBM. lap between contiguous frames. For each MFCC features are appended with only The final output score SX() is accord- frame, a feature vector with 52 dimensions delta and double delta features. This results ing as the final answer “YES” or “NO” is calculated: 13 Mel Frequency Perceptual in a 39-dimensional feature vector. This vec- by comparing with the system threshold. Linear Predictive (MFPLP) coefficients, 13 tor undergoes feature-transformation and The flow of speaker recognition is shown delta cepstral, 13 double delta cepstral, and mean-variance normalization using the in Figure 5. 13 triple cepstral. same procedure as explained in the cepstral The CVR can act effectively as a latent GMM system. Each normalized feature signal transducer for both speech and non- Two of the speaker models vector (39 dim) is concatenated with its sec- speech audio information. Sound identifica- Cepstral GMM system—The GMM sys- ond (39x39) and third (39x39x39) order poly- tion and voice recognition are aimed at of- tem uses a 100-3800 Hz bandwidth front end nomial coefficients. Mean and standard de- fering more clues in the analysis and classi- consisting of 24 MEL filters to compute 13 viation of this vector are computed over the fication of speech and non-speech CVR cepstral coefficients (C1-C13) with cepstral conversation side. signals. mean subtraction, and their delta, double From the test we have done, the result delta, and triple-delta coefficients, produc- Speaker recognition approaches shows that two system works quite well for ing a 52-dimensional feature vector. The As mentioned above, the speaker recogni- some cases, and the search speed is reason- feature vectors are modeled by a 2,048-com- tion includes speaker verification and ably fast. ◆

10 • ISASI Forum April–June 2008 Reverse Engineering Overcomes Corporate Amnesia A process of “corporate amnesia” has become common among manufacturers, brought about by lengthening aircraft service lives and shortening career spans of design/ development engineers within one employer.

By Peter Coombs, Senior Inspector of Accidents, Air Accidents Investigation Branch (AAIB), U.K.

(This article was adapted, with to be at its greatest when the permission, from the author’s accident under investigation presentation entitled Use of Re- occurs far from the home ter- verse Engineering Techniques ritory of the type certificate to Generate Data for Investiga- holders. tions, presented at the ISASI The above phenomenon can 2007 seminar held in Singa- be unfortunate in circum- pore, Aug. 27-30, 2007, which stances where the compliance carried the theme “Interna- of the subject aircraft with the tional Cooperation: From In- design requirements, or in vestigation Site to ICAO.” The some respects the adequacy Figure 1 full presentation including and relevance of those design cited references index is on the requirements to the accident ISASI website at www.isasi.org.—Editor) With mature aircraft types, archived de- circumstances, have come into doubt. sign data in the possession of TC holders On a number of recent investigations, ccident investigation has traditionally may not be readily accessible. If it is avail- where structural failures have occurred, a relied on a variety of sources of evi- able, it may be in a form not easily identi- process of “reverse engineering” has been Adence. One of the most important has fied, understood, or manipulated by their carried out by the AAIB, under the super- been analytical data supplied by type cer- structural, aerodynamic, or systems spe- vision of the author, to combat these diffi- tificate (TC) holders or original equipment cialists, because they will probably be more culties. This has been done in order to es- manufacturers (OEMs). Such information used to operating with state-of-the-art de- tablish important parameters that might is particularly important in those complex sign tools. They are often inexperienced in previously have fallen under the province investigations involving structural failure. the use of earlier methods of technical analy- of the type certificate holder, but where in- A number of problems with these sources sis and design data recording systems, rou- adequate data have come from that source. of data have, however, been encountered in tinely utilized in the past in the development recent years. processes of aircraft and their components. Investigation summaries This assumes that the relevant data can The two investigations summarized here Peter Coombs, an actually be located and identified, a situa- have been to aircraft in very different cat- engineering investigator tion that cannot always be guaranteed. egories, suffering very different accident with the U.K. Air Acci- A process of “corporate amnesia” has causes. Similarities in the investigative pro- dents Investigation become common among manufacturers, cess for each were, however, considerable. Branch since 1972, has brought about by lengthening aircraft ser- The first of these events was to a me- investigated accidents and vice lives and shortening career spans of dium-sized, offshore, public-transport heli- incidents to most classes design/development engineers within one copter. This suffered a lightning strike re- of aircraft including large public trans- employer. Some manufacturers seek out sulting in damage to a composite tail rotor port airliners, transport helicopters, mili- long-retired engineering specialists to at- blade, which ultimately led to failure of the tary combat aircraft, and many general tend meetings with investigators in often tail rotor gearbox attachment making con- aviation types. Before joining the AAIB, vain attempts to recapture long-forgotten tinued flight impossible. he trained with the British Aircraft design data. Other manufacturers seem Although the gearbox fell from the py- Corporation. He later served as a design reluctant to part with information they lon at the end of the flight, somewhat mi- engineer on the Concorde SST. He holds a probably possess, either because they find raculously the hydraulic pipes did not ini- master’s degree in aircraft design and it technically embarrassing in the context tially fracture. Instead, they continued to pilot’s licenses on single- and multi- of the accident or for reasons about which support the mass of the gearbox for a brief engine GA aircraft and on helicopters. we can only speculate. The problem seems period. This preserved the longitudinal bal-

April–June 2008 ISASI Forum • 11 ance of the aircraft, en- nautical engineering companies in Europe abling a successful au- and North America. torotation to take place into a rough sea. Shortly The accidents afterward, the pipes The helicopter, an AS 332, lost part of a com- failed and the gearbox posite tail rotor blade as a result of a light- fell away and sank to the ning strike while descending to an offshore sea bed. The aircraft rig. Subsequent impact destruction to the drifted downwind until it remainder of the blade (see Figure 4), as also sank. Figure 1 the rotor struck the tail boom during gear- shows the aircraft some box separation, disguised the amount of ini- time between the loss of tial lightning damage. It can be seen in Fig- Figure 2 the gearbox and the final ure 5 that four of the blades have been de- sinking, shortly after all stroyed by this same impact mechanism, the passengers and crew although only the one on the left has any Figure 3 evacuated. evidence of the earlier lightning damage. The occupants es- It was required to establish the level and caped by dingy and were degree of initial lightning damage on this subsequently rescued. single blade in order to determine the se- Surprisingly, we were verity of the lightning strike that the blade successful in recovering suffered. This was necessary to establish both airframe and tail ro- the practical validity of the lightning certi- tor gearbox from two fication requirements to which the aircraft separate locations, both had been qualified. The loss of the machine at depths in excess of 700 had cast considerable doubt on the ad- feet (see Figures 2 and equacy of those requirements. It was feared 3). that aircraft operating at low levels, in win- Recordings of the tim- ter, in the temperate maritime conditions ing and location of the over the North Sea, were especially vulner- critical lightning strike able. At the time, this was the busiest area were obtained using meteorological record- namic response characteristics of the tail of offshore, long-range, public-transport ing equipment, as was a time-referenced re- boom and pylon structure of the rotorcraft. helicopter operation in the world. cording of the final radio distress call, made These characteristics significantly raised as the aircraft ditched, following the failure the stress levels in the gearbox attachments of the normal tail rotor gearbox attachment. resulting from rotor imbalance. Both investigations resulted The second accident was a fatal event to The second investigation, to the GA air- in development of methods a four-seat metal general aviation (GA) aircraft, took advantage of state-of-the-art tech- craft, which suffered the unusual phenom- niques to establish structural strength and that could be utilized in whole enon of a failure of the outboard section of aerodynamic loading figures. These were or more probably in part a wing, in a download sense. This occurred thought to be more accurate than those avail- during future investigations, while flying in smooth air in daylight visual able to the original aircraft designers. meteorological conditions. A good quality The expertise required to carry out the regardless of the size of radar recording and a reliable meteorologi- detailed calculations in support of these in- aircraft involved. cal after-cast enabled the airspeed history vestigations was provided by a number of to be calculated with an acceptable degree specialist analytical companies in the U.K. of accuracy. It was noted with some concern These have generally grown up during the that the speed, at the time of the failure, past 25 years. In addition, a U.K.-based, in- was significantly below the maneuver speed ternationally known academic establish- Tests on a number of ex-service blades of the aircraft. ment also supplied such assistance. The lat- were carried out at a lightning test facility Both investigations resulted in develop- ter has a wide range of expertise through to establish the extent of damage inflicted ment of methods that could be utilized in areas of structural design, flight mechan- by differing degrees of intensity of lightning whole or more probably in part during fu- ics, simulation, and dynamic load analysis. strikes. ture investigations, regardless of the size The specialist companies provide expertise It was found, from wreckage examina- of aircraft involved. Both investigations re- in areas ranging from finite element analy- tion, that imbalance following the strike had quired precise assessment of strength and sis to structural dynamics. One has specific created sufficient vibration to cause one of loadings in localized areas of structure. experience on maneuver load analysis of the three gearbox securing bolts to slacken. The first also required assessment of fast combat jet aircraft. They act as con- This both concentrated cyclic bending on loading applied as a result of tail rotor im- tract engineers to both major aircraft only two attachment lugs and altered the balance acting in conjunction with the dy- manufacturers and to other specialist aero- natural frequency of the tail boom/gearbox

12 • ISASI Forum April–June 2008 from timing of the final VHF lightning strike was now required to create Figure 4 crew distress call. The rotor loading to cause failure in the known time, speed was known from air- and a realistic assessment of the pure light- craft data. From these items ning damage required to cause this loss of information, it was possible could be made. By comparing this calculated to calculate the amount of im- mass loss with the damage inflicted by light- balance that provoked the ning tests on used blades, carried out ear- gearbox fatigue failure and lier using known electrical intensity char- must, therefore, have been acteristics, it was possible to determine the brought about by the lightning approximate magnitude of the lightning damage. strike. This, although confirming that the Figure 5 When first calculated, how- certification requirements then in force ever, without considering the were realistic in terms of magnitude for that dynamics of the tail boom, the flight environment, revealed significant mass loss from the blade, to drawbacks in the aircraft’s design process. create this imbalance, was It showed that the practical effects of bolt found to be slightly more than slackening under vibration loading, to- that resulting from damage gether with the similarity of natural fre- clearly caused finally by the quency of the structure to the rotor speed, collision between the blade had not been adequately taken into account and the tail boom. See again at the design stage. Figures 4 and 5. This damage Certification compliance merely called for had quite clearly only oc- an absence of severe structural damage in curred as the gearbox sepa- (static) lightning test conditions. It did not rated, some minutes after the call for a full assessment of the structural strike; something was un- behavior of the rotor system and mounting doubtedly wrong with the cal- after the limited lightning damage had oc- culated result. curred. No such assessment had apparently Figure 7 It was, therefore, decided been carried out on this aircraft type. that the dynamic characteristics of the tail boom/gearbox Second accident combination would be evalu- In the case of the GA aircraft, a PA28R-200- ated theoretically. This work 2, a finite element (FE) model of the wing was carried out using a bay in which the failure occurred was cre- manufacturer’s dimensioned ated using a manufacturer’s layout draw- Figure 6 layout drawing of the tail ing and measurements of panel thickness boom and skin thickness mea- made on the separated wing and a further surements made on the dam- sample wing. Figure 7 shows a visualization aged boom by ourselves. The of the model. mass of the gearbox was de- An evaluation of control responses was termined simply by weighing carried out, using a simple simulator, pro- combination. This alteration brought this the salvaged unit. grammed with a modified NASA computer structural frequency (in cycles per second) The new calculated tail boom dynamic model of the aircraft type. This was done to close to the rotational speed of the unbal- characteristics were confirmed by a reso- produce a realistic series of control column anced rotor with the damaged blade (in nance test of the rear structure of an in- displacement-time histories of pitch control revolutions per second). service aircraft while on the ground and inputs, creating a series of wing download- A finite element analysis of the gearbox were further corrected theoretically for the time (negative G) histories as well as other was carried out using actual measurements predicted effect of a single, loose tail rotor flight parameters. The span-wise negative of the casting to create the grid. In Figure 5, gearbox attachment. By this means, it was lift distribution was calculated and con- you see one of the visualizations of the gear- determined that the natural frequency of verted to engineering units. The time his- box showing the varying stress distribution the rear of the aircraft in cycles per second tory resulting in the highest negative load for a unit loading. The number of cycles to almost matched the rotor speed in revs/sec- factors achieved in the simulation series was failure was known, since the times of both ond. The cyclic forces applied to the two then used to factor the distributed forces. the strike and the final gearbox separation effective tail rotor gearbox attachments The result was used as the varying aerody- were known from recordings. The initial were thus found, as a result of these close namic force/time input to evaluate the be- event time was identified precisely using the frequency similarities, to be far greater than havior of the FE model under a varying atmospheric lightning recording equipment those initially calculated without taking ac- down-load. On carrying out this exercise, it available to the U.K. Met Office, while the count of the dynamics of the tail boom. was found that the theoretical wing strength failure time was established approximately Only a small mass loss resulting from the from the FE analysis was far in excess of

April–June 2008 ISASI Forum • 13 control wheel travel on an example of the type borrowed for measurement purposes). A persuasive scenario to ex- plain the occurrence, based on the nature and seating position of the aircraft occupants, in this dual- control machine was then devised. These two investigations demonstrate the way in which capabilities from part- ners outside the normal areas of Figure 7. Finite element analysis Figure 8 expertise usually visualization. called upon by investigators can be that required to carry the highest loads harnessed to re- implied by the results of the simulations. place data more Up to this point, only symmetrical pitch usually found maneuvers had been considered. It was from OEMs and realized, however, that even with those TC holders when forces calculated for such maneuvers act- such data are not ing in unison with forces resulting from a readily available. large simultaneous roll control input the Although the ab- load to fail the wing could not reasonably sence of manufac- be approached, much less achieved. The turer’s data may reason for the wing failure thus remained seem at first a entirely obscure. great handicap, A review of assumptions made to cre- the ever-increas- ate the finite element model was then car- ing power of mod- ried out; with a number of more pessimis- ern computers tic assumptions applied, the reduction in and the rising so- wing strength was still insignificant. Figure 9. Finite element analysis visualization. phistication of At this point, further specialist assis- commercially tance was sought. The company that was input, followed immediately by complete available analytical packages compensate consulted drew attention to the significance reversal of control inputs in both axes. Un- for much of this loss. These enable data to of inertia effects created by rapidly reversed der these influences, failure loads at the be generated and manipulated, which pro- control inputs. It was able to estimate the wing station where the actual aircraft struc- duce results that are no less accurate than approximate mass distribution of the wing ture failed could just be reached at the those achieved in the past by OEMs. These structure and also to create a NASTRAN/ known airspeed. The control input-time his- will have used methods that were state-of- PATRAN model of the machine, entirely by tories are shown graphically in Figure 8. the-art at the time of the aircraft’s initial measurement of a real example and use of Visualizations of the failure modes are design but may be two or more decades old published data relating to the type. This shown in Figures 9. at the time the accident occurs. enabled maneuver loads to be calculated for Calculation of control forces at this speed Investigations carried out using such continuously varying pitch and roll displace- indicated that these were sufficiently low to methods present a challenge to the manu- ments. It proved possible then to create a enable them to be easily generated by a facturers that frequently reengage more maneuver/time history that resulted in fail- front-seat occupant. (Control gearing was es- fully when they see that official investiga- ure of the finite element model as a result tablished by simply measuring control sur- tive bodies are serious about finding the root of full simultaneous pitch and roll control face angular movement for corresponding causes of such intricate accidents. ◆

14 • ISASI Forum April–June 2008 Critical Aspects of International Incident Investigations

The very fine line between incident and accident clearly emphasizes the importance of having well-trained airline investigators. By Deborah J. Lawrie, Robert N. van Gelder, and Jan Smeitink

(This article was adapted, with permission, and a mixture of slush and ice was found in operation was finished. (Kilfrost ABC 3 from the authors’ presentation entitled Cri- small areas on the top of the wings. The air- Type 2/50%). tical Aspects of International Incident In- craft was deiced, and the captain performed A short time later, the aircraft taxied for vestigations, presented at the ISASI 2007 a visual check of the wings after the deicing departure from Runway 36. A special pro- seminar held in Singapore, Aug. 27-30, 2007, which carried the theme “Interna- Deborah Lawrie is in human factors/ergonomics in 1982 tional Cooperation: From Investigation Australian born and joined KLM Royal Dutch Airlines Site to ICAO.” The full presentation includ- and holds a BSc and in 1978. He currently flies as a Boeing ing cited references index is on the ISASI bachelor of education 747-400 captain. During the last 17 website at www.isasi.org. degree from Melbourne years, Robert has held the positions of The authors are founders of the Indepen- University. She taught ergonomics engineer/human factors dent Safety Investigation & Consultation mathematics and physics specialist, line and simulator instructor, Services [ISIS] group, part of whose work and was a flying instructor from 1976 chairman of the Standardization is to teach others and, especially airline op- until 1979, when she joined Ansett to Committee on the Boeing 737, founder erators, how to investigate serious inci- become Australia’s first female airline and editor-in-chief of KLM’s flight dents.—Editor) pilot. She joined KLM Cityhopper in safety magazine in for SAFETY, and 1993, instructing on the Fokker 50 and chief investigator. his article is based on a case study of later operating the Fokker 70/100. In 1998 a serious deicing incident that had Deborah established the company’s Flight Jan Smeitink is a TT significant consequences for ground Safety Department, and held the position member of the manage- handling supervision and developed into a of safety manager and chief investigator ment team of the Dutch broad-based international investigation, for 8 years. She was chairman of the Safety Board, which lasting more than 2 years and conducted in ERA Air Safety Working Group from investigates transport accordance with ICAO Annex 13. 1998-2004. She is currently flying the accidents and incidents On Feb. 16, 2002, a Fokker 70 aircraft had A330 with KLM. and other types of been parked in Turin, Holland, overnight. disasters, serious accidents and inci- Rain and snow fell during the night with Robert van Gelder was dents, as well as crisis management light and variable winds. The temperature/ born in the Netherlands and disaster control. He joined KLM dew point ranged between 2/0°C and 0/-1°C, and received his flight shortly after graduating as a mechani- and enough fuel remained on board for the training at the Royal cal engineer from the polytechnic college return flight to Amsterdam the next day. Dutch Flight Academy in Arnhem, the Netherlands. He During the pre-flight inspection the next in 1973. He graduated flew as a flight engineer on the B-747- morning, ridges of ice 1.5 –2 cm thick were with a BSc honors from 200/300 and has more than 10,000 found under the leading edges of the wings, Loughborough University of Technology flying hours.

April–June 2008 ISASI Forum • 15 cedure with a right turn at 500 feet is specified in the case of engine failure during takeoff from this runway due to the close prox- imity of high terrain. The takeoff was performed using full thrust with the engine anti-ice on. The wind was from the northeast at 3 knots. There were scattered clouds at 500 feet and light rain. The temperature/ dew point was 1/0° C. All the engine indications were normal during the takeoff roll. But during the ro- tually surfaced on the Multi-Function Dis- PRECEDING PAGE: This photo illustrates tation the fan vibration in engine No. 1 in- play Unit, the crew then became aware that the icing conditions that existed on the creased, and at liftoff there was a sudden the only remaining engine was not function- evening of Feb. 16, 2002. loss of oil pressure and fuel flow to engine ing normally. The first officer later de- ABOVE: Damaged sustained by the Fokker No. 2 and the fan vibration in engine No. 1 scribed the situation as “the aircraft was not 70’s right engine when struck by clear increased above limits. We now know that flying really well and the engine did not feel ice from the wing. as the wings flexed during the rotation, smooth.” The captain declared a MAYDAY RIGHT: This iced wing is illustrative of the large pieces of clear ice separated from both and requested vectors to return for an ILS problem encountered by the Fokker 70 wings causing violent and immediate de- approach on Runway 36 at Turin. After be- involved in the noted incident. struction of the right engine and damage ing airborne for 29 minutes, the aircraft fi- to several fan blades in the left engine. The nally landed safely back on Runway 36. representative under whose supervision the following also occurred: company investigator would act as an advi- • Accessory gearbox and hydraulic pump The aftermath sor, the Turin situation called for an ap- housing were cracked. [Capt. Lawrie, as chief investigator for proach that deviated from the ICAO An- • Power lever transducer was hanging on KLM, was asked to assist in the incident nex 13 philosophy. its wiring. investigation. She relates her role.—Editor] In the Turin situation, a comprehensive • Gear box housing was cracked in two By the time I arrived in Turin with the knowledge and understanding of the ICAO places. technical pilot, investigation of the incident investigation process as well as knowledge • Throttle linkage was detached from the had already commenced and was under the about my entitlements and responsibilities fan case. control of the investigator-in-charge from and those of the other parties involved was The situation on the flight deck was com- the ANSV (Italian Aviation Safety Board). going to prove to be invaluable in what was plicated by the turn that was now neces- At this stage, it was unclear what had going to develop into a lengthy and contro- sary at 500 feet, a jammed fuel lever on the caused the damage to both engines. Other versial investigation. right engine, which disrupted the engine damage to the fuselage and the surface of shutdown procedure, and several other fail- the right wing led to initial speculation that The investigator as advisor role ures that occurred: an autothrottle failure, the damage to the left engine may have been The case study also shows that serious inci- an autopressurization failure, and eventu- caused by ingestion of debris from the cata- dent investigation is just as important as ally a fuel asymmetry warning. strophic failure of the right engine. an accident investigation and, therefore, The high-vibration warning on the left As chief investigator for the airline, I had should be performed as comprehensively engine was temporarily “hidden” by all the some previous internal incident investiga- and with the same allocation of resources other failures due to the priority allocation tion experience. Quite suddenly now, how- as if it had been an accident. In cases such of the aircraft’s warning system and insuf- ever, I found myself as the only party on as Turin where the operator had been for- ficient space available to display all the site, in what was to be an international in- tunate to escape disaster, then investigation warnings at the same time on the Multi- vestigation involving several parties and I of this event had the potential to reveal as Function Display Unit. Due to all the other was dealing directly with the investigator- much, if not more, about all the contribut- failures, the crew remained unaware of the in-charge. This type of situation is more ing factors that led up to it. high-vibration problem with the left engine likely to develop in the case of a serious in- Indeed, it was later revealed that the for the next 10 minutes. cident rather than an accident. While the Turin event had the same “footprint” as the When the high-vibration warning even- formal procedure calls for an accredited Scandinavian Airlines accident that involved

16 • ISASI Forum April–June 2008 to keep recording for several hours after the incident. After all the data were collected and extensive analysis of both engines had been performed by Rolls-Royce, the process of elimination led to the conclusion that the most probable cause of the event had been the ingestion of large amounts of ice by both engines. The focus of the investigation turned to the deicing operation, the post deicing inspection, and the operator’s supervision of ground handling. Deicing at European airports was a very controversial and high-profile safety concern at the time, and a few years earlier the an MD-81 that took off from Stockholm’s the IIC to interview the air traffic control- DAQCP (Deicing and Quality Control Pool) Arlanda Airport early in the morning of lers who were on duty in the tower at the had been established. The DAQCP was an Dec. 27, 1991. Vibrations from the MD-81’s time of the incident. organized group of operators who shared engines were noticed 25 seconds after be- I must emphasize that this event did not the auditing of several deicing contractors coming airborne. Approximately 1 minute have the high-profile media attention that throughout Europe. later, both engines failed. The aircraft was one associates with an accident. Also, the In the Turin investigation, there was con- committed to a forced landing in a field seriousness of the event didn’t start to fil- troversy over where it broke into three parts after the ter through to the interested parties until • knowledge of and training of the correct impact. Remarkably, in this accident there late in the day. The company reported the techniques for the removal of clear ice. were no fatalities, and later the investiga- matter to the Dutch investigation author- • ownership of the final responsibility for tion revealed that ice from the wings had ity, then known as the RVTV. The following the post-deicing inspection. entered both engines causing them to fail. day, representatives from Fokker and Rolls- • the operator’s contractual arrangement In Turin during the first hours of the in- Royce arrived in Turin. with the handling agent that performed the vestigation, the technical pilot and I worked The Italian Aviation Safety Board had deicing and the agent that performed the side by side with the Italian investigator-in- initiated the formal international investiga- inspection. charge (IIC). Our operational knowledge was tion, but our position as advisor to the Dutch • the separate arrangement between the very much appreciated, and we managed to accredited representative was not formal- handling agent performing the deicing and establish a good relationship with the IIC. ized until after we returned to Holland 2 the agent performing the post-deicing in- Aircraft documents and operating manuals days later. By the time we returned to Hol- spection. were identified and discussed, a detailed in- land we had • the structural safety deficit at an inter- spection of the cockpit was made, and a brief • established a good working relationship national regulatory level with no certifica- inspection of the engines and external con- with the IIC. tion rules for ground handling companies. dition of the aircraft was performed. • met several of the other parties who • evidence of previous substandard deic- The IIC arranged for us to inspect the would be involved in the investigation. ing operations in Italy. runway and the surrounding area at the • established our value as advisors in terms In the case of Turin, the company had a point where the aircraft had rotated. We of knowledge, expertise, and availability. written contract with a deicing agent but only retrieved pieces of engine acoustic lining An accredited investigator from Dutch a verbal contract with the post-deicing in- among other broken bits and pieces. It was ALPA was assigned to me, and together we specting company, which was a separate com- at this critical point in time, however, that acted as advisors to the RVTV. pany to that which performed the deicing. the driver of the airport safety car casually Among the vast quantity of collected data The crew documentation on board the mentioned that earlier that morning, just was information from the digital flight data aircraft indicated that the handling agent after the incident, he had found some very recorder, but information from the cockpit would perform the deicing and the post- large pieces of ice at the same location. To voice recorder was not available due to the deicing inspection, but in this case no post- close the day, we were invited to accompany jammed fuel lever that had caused the CVR deicing inspection was performed other

April–June 2008 ISASI Forum • 17 than the visual check performed by the captain. Further- more, several findings in rela- tion to training and contracts remained open from the deicing pool audit that had been conducted in January of the previous year. Tension between investigators was apparent and under- standable. Pending insurance claims and political issues also added pressure to the investigation. The importance of the role Right engine’s damaged fan blades. and the entitlements of the Dutch accredited representative were abso- vital facet of the operator’s flight safety opinion that there is an industrywide un- lutely vital to the progress of the investiga- program and more importantly the train- derestimation of the importance of well- tion. In turn, the Dutch accredited represen- ing should be within the reach of, and avail- performed incident investigations and qual- tative relied heavily upon the support, knowl- able to, all operators. ity report writing. edge, and objectiveness of his advisors. One of the most striking aspects of many In terms of an airline safety program, it Several analyses and recommendation formal accident investigation courses is that is important that meetings were convened, some of which the bulk of such courses is not relevant to • the seriousness of an event is recognized were held in The Hague and some in Rome. the airline investigator. Also many courses and assessed accurately by means of a com- The Dutch accredited representative could are out of the financial reach of smaller op- prehensive risk-assessment program. not attend all the meetings in Rome, so on erators and those who it could be argued • the airline must be prepared to partici- some occasions we were present at these may need it most. pate in investigations of serious incidents meetings as replacements. We were, there- Fortunately, most airlines tend not to with or without the assistance of the state fore, playing a variety of roles on different have accidents. If an airline does have an investigation authority. occasions throughout the investigation accident, the airline investigator will at best • if the state investigation authority con- ranging from a subordinate role to a lead- be an advisor and will certainly never be ducts an incident investigation, then the air- ership role. We had to balance diplomacy acting as an investigator-in-charge. line investigator should be aware of its re- with assertiveness and, above all, we had to Airlines do, however, have incidents from sponsibilities and entitlements. This point keep our focus on getting to the bottom of time to time and sometimes these incidents applies equally in an accident investigation. the true causes of the event. are serious. Often, though, due to staff We would argue that investigators should As this was an international investiga- shortages or other investigations already be trained how to recognize a serious inci- tion, the importance of a final report in the in progress, the national investigation au- dent, how to investigate a serious incident, English language was apparent. Because thorities do not have sufficient time or re- how to write a report that supports effective we were more fluent in English, the union sources to investigate all serious incidents recommendations, and to develop a sense of investigator and I were given the very im- and at best are sometimes only able to give when risk should be mitigated. Equally important job of writing the report under the limited attention. Even though the investi- portant is the airline investigator’s ability to supervision of the IIC. The report and its gation of serious incidents has been man- work with other investigators and the abil- recommendations would not only be impor- dated in the latest version of ICAO Annex ity to manage a small investigation team. We tant to our company but also to many other 13, the reality is that this task more often maintain that the proper and comprehensive operators who had a vested interest in this than not is allocated to the operator itself investigation of incidents is vital to the im- very critical safety issue. and, therefore, the quality of such an inves- provement of safety. tigation depends upon the training of the In the case of Turin, the report also ana- Investigator training operator’s investigators. There is no doubt lyzed and produced recommendations in As illustrated by the example in Turin, that valuable lessons can be learned from regard to proper training of airline investigators is a incident investigations, and we are of the • fueling policy,

18 • ISASI Forum April–June 2008 • crew hand-over procedures, We believe that accidents such as the one at Cali • preflight inspections, • the Deicing and Quality Control Pool au- and serious incidents such as the one at Turin clearly diting system, demonstrate the very fine line between incident and • organization and management of out station ground handling, and accident and clearly emphasize the importance of • internal distribution and control of com- having well-trained airline investigators. pany documentation. In view of the potential value of well-for- • Better capacity and more time to concen- • to create a course that is portable. mulated recommendations that arise from trate on and discuss “regional” issues. We believe that the delivery and teach- a comprehensive investigation, it makes • Less costs per head for the company. ing methods are just as important as the sense to give consideration to affordable and • Less down time for personnel because content of the course. Lecturers are trained appropriate investigator training courses. traveling away from the home base was not in teaching skills in line with recognized We also believe that due consideration required. university teacher training methods. The should be given to airline investigator pools • Increased flexibility for the company in ISIS course is highly interactive, and the or investigator exchange programs. case of production problems. number of attendees is restricted to smaller Our belief is that if investigator pools or Why a stand-alone incident investigation groups in order to guarantee individual at- exchange programs existed then several course and not an integrated accident in- tention and feedback. smaller airlines of limited resources and vestigation course? We place a very high value on incident capability would benefit enormously from • Airlines have more incidents than acci- investigations, not only from the cost aspect the opportunities—not only for their inves- dents, and proper investigation of an inci- for smaller companies but also for the added tigators to improve their skills by working dent can help to prevent an accident. value to safety that will come with compre- along side more experienced investigators, • Investigation training also requires con- hensive, well-performed investigations and but also that all companies would benefit solidation, and working with other experi- associated quality reports. from an exchange of ideas and incident in- enced investigators and advanced training It is often the case that many serious investigations could be performed more thor- is only of value after a suitable consolida- cidents will have the same “footprint” as an oughly and proficiently. tion period. accident but that a stroke of luck or good • Cost, spread of costs, employees are only fortune breaks the error chain and an acci- Interactive and customized absent from duties for 1 week at a time in- dent is avoided. training benefits stead of the usual 2 weeks or more. This was seen in the case of the Fokker In 2003 the ISIS incident investigation • Specific learning—more concentration 70 icing incident at Turin. Many accidents, course was developed to train investigators on the topics and disciplines that are rel- on the other hand, would have or could have in order that they may lead and manage an evant to airline operations. been serious incidents save for one factor incident investigation and that they may be • Learning over a longer period of time plus such as in the case of Cali when the speed able to perform the role as advisor in an the opportunity to revise and update previ- brakes remained extended when the B-757 accident or incident investigation. ous learning by doing the accident investi- attempted to clear the high terrain. If Cali The specific advantages of in-house gation training module 6 months to 1 year had been a serious incident and not an acci- courses involved have been after the incident investigation module. dent, then the investigation of this event • More people were trained and ready to Why did ISIS set up this course, while would have been vitally important. The in- perform investigations, safety assessments, there are already other courses available? teresting thing, however, from our point of and analysis. We wanted view is that had Cali only been an incident, • Persons were trained in the same “vein” • to see more emphasis placed on incident the investigation may have had to have been and were therefore able to think on the investigation. performed by the airline itself. same wavelength. • to see the inclusion of more relevant ma- We believe that accidents such as the one • Persons from several different depart- terial and to give more hands-on practice. at Cali and serious incidents such as the one ments were trained together, which in- • this type of training to be available for at Turin clearly demonstrate the very fine creased their individual knowledge and un- all operators, large and small. line between incident and accident and derstanding of one another’s roles within • investigators to be able to recognize the clearly emphasize the importance of hav- the company. intrinsic value of other investigation reports. ing well-trained airline investigators. ◆

April–June 2008 ISASI Forum • 19 passage of an item through its various own- ers.” Adapted for the context of recorded flight data, the definition becomes “a record Flight Data– of a physical measurement or system state and the changes to that record as it passes What Every Investigator through various system components until it is interpreted for an investigation.” Consider the “Selected Heading” ex- Should Know ample above and as shown in Figure 1. The The correct interpretation of flight data and/or audio data flight crew uses the heading window on the requires a full understanding of the entire signal path from mode control panel (MCP) to choose a flight heading. The MCP transmits this value to measurement to recorder to investigator. the flight control computer (FCC). The By Michael R. Poole, Managing Partner, Flightscape, and Simon Lie, FCC uses the value for computing the cor- Associate Technical Fellow, Boeing Air Safety Investigation rect flight director and autopilot behavior.

(This article was adapted, with permission, Consider two different parameters that are Mike Poole is a principal from the authors’ presentation entitled recorded on certain B-737 aircraft: Selected at Flightscape, a flight Flight Data—What Every Investigator Fuel Flow and Selected Heading. In the sciences company Should Know, presented at the ISASI 2007 former, the “Selected” indicates that mul- specializing in flight data seminar held in Singapore, Aug. 27-30, tiple fuel flow readings from different sen- analysis software. He 2007, which carried the theme “Interna- sors are available and this particular value was with the TSB of tional Cooperation: From Investigation has been judged to be the most accurate Canada for more than 20 Site to ICAO.” The full presentation includ- and thus has been selected for display to years where he served as the head of the ing cited references index is on the ISASI the flight crew. In the latter, “Selected” Flight Recorder and Performance website at www.isasi.org.—Editor) means the target value of heading chosen Laboratory and the Flight Recorder by the flight crew via the autoflight mode Group chairman on all major accidents. light data are becoming more readily control panel. As these two examples dem- He also has represented Canada as the accessible and are increasingly being onstrate, scientific rigor requires a full un- national expert panel member to the Fused for investigation and airline derstanding of the origin of flight data and International Civil Aviation safety programs. Modern aircraft record a how the data were processed. Organization’s Flight Recorder Panel. huge amount of data compared to just a few It is important that investigators and years ago, but even in the most advanced airline Flight Operations Quality Assur- Simon Lie is an associate aircraft recording systems, significantly ance (FOQA) analysts appreciate the prov- technical fellow and air less than 1% of the available data are actu- enance of the flight data, especially when safety investigator at ally recorded. The challenge of analyzing drawing substantive conclusions. There is Boeing Commercial flight data is to recreate an accurate under- anabundance of flight data analysis tools Airplanes. He has an standing of an event from that small per- that are becoming progressively more au- academic background in centage of the available data. tomated, which in turn increases the po- composite structures and The scientific evaluation of data requires tential to mislead. aeroelasticity and spent 10 years of his an understanding of the origin, or prov- career at Boeing troubleshooting enance, of the data and how the data were Provenance predelivery issues on the B-747, -767, and processed. Both authors have seen profes- There are many examples where the cor- -777 aircraft. Lie led Boeing support of sional investigators reach mistaken conclu- rect interpretation of an FDR recording numerous accident investigations sions when reviewing recorded flight data requires a full understanding of the prov- including Singapore Airlines B-747 without fully understanding the origin and enance including the methods employed by Taipei 2000, China Airlines B-747 history. the replay ground station. According to the Taiwan 2002, Flash Air B-737 Egypt As parameters proliferate, even the nam- Oxford English Dictionary, provenance is 2004, and Mandala Airlines B-737 ing of parameters can lead to confusion. “a record of the ultimate derivation and Indonesia 2005.

20 • ISASI Forum April–June 2008 Figure 1. A graphical depiction of the provenance of the selected heading parameter Charles de Gaulle Airport (CDG), France, recorded on a B-737-300 FDR. The data originate in the mode control panel and pass with a stopover at Cairo International Air- through a number of distinct transformations before being utilized during an port (CAI) for refueling. Flight 604 de- investigation. parted from SSH Airport with 2 pilots (captain and first officer), 1 observer, 4 cabin In addition, the FCC transmits the value to of ones and zeros. The data are subse- crew, 6 off-duty crew members, and 135 pas- the digital flight data acquisition unit quently extracted and converted from raw sengers on board. The airplane was de- (DFDAU). binary format back into engineering units stroyed due to impact forces with the Red Continuing to follow the signal chain, we (i.e., degrees). The converted value is rep- Sea with no survivors. find that the DFDAU stores the values it resented as a plot, table, animation, or pos- The airplane had departed from SSH receives from the FCC until the value is sibly another format. Finally, the data rep- Runway 22R and was airborne at 02:42:33 scheduled to be written to the FDR. The resentation is interpreted by the accident UTC, approximately 2½ minutes prior to FDR writes the value to either magnetic or incident investigator. the crash, and had been cleared for a climb- tape or solid-state memory as a sequence In theory, each parameter may have its own ing left turn to intercept the 306 radial from unique signal chain. SSH VOR station located just north of Run- In practice, param- way 22R. eters that have the The FDR and CVR were subsequently same source often recovered from a depth of more than 1,000 share the same chain meters and provided data used during the —but not always. At investigation. The airplane began the left each step of the turn but then rolled out of the left turn and chain, there is the po- into a right bank that eventually reached tential for a change to 110° right bank. A recovery attempt was the signal. Therefore, made but was not completed before the air- each step must be plane descended into the Red Sea. fully understood as The FDR recorded that the departure both intended and was flown with the use of the captain’s and unintended changes first officer’s flight directors in heading se- can affect the results. lect mode. In this mode, the flight director provides roll guidance to turn the airplane B-737-700 toward and hold a “selected heading” set example by the flight crew on the mode control panel. On Jan. 3, 2004, Accordingly, investigative attention turned about 02:45:06 to the recorded values of selected heading Figure 2. Altitude, airspeed, heading, and selected heading UTC, 04:45:06 local on the FDR. parameters during the accident flight. Selected heading is time, Flash Airlines Figure 2 depicts the airplane heading, recorded once every 64 seconds. The other three parameters are Flight FSH604, a selected heading, altitude, and airspeed recorded every second. The value at time 123 seconds is unusual Boeing 737-300, during the accident flight. Heading, com- as a value equal to the runway heading (220°) would be expected Egyptian registra- puted airspeed, and altitude are recorded at this phase of flight. The unorthodox arrangement of the tion SU-ZCF, each second. Selected heading is recorded heading and selected heading scales (from higher values to lower crashed into the once every 64 seconds. Standard practice values) is intentional. By convention, parameter values that result Red Sea shortly af- calls for setting the selected heading equal in or result from a right turn tend toward the bottom of the page. ter takeoff from to runway heading during take off. At time The FDR and CVR were subsequently recovered from a depth of Sharm el-Sheikh 59 seconds, before the airplane turns onto more than 1,000 meters and provided data used during the International Air- the runway, the recorded value of selected investigation. The airplane began the left turn but then rolled out port (SSH) in south heading was 220° (runway heading) as ex- of the left turn and into a right bank that eventually reached 110° Sinai, Egypt. The pected. At time 123 seconds, just prior to right bank. A recovery attempt was made but was not completed flight was a passen- rotation, the recorded value was 360°. before the airplane descended into the Red Sea. ger charter flight to Later during the flight, the recorded val-

April–June 2008 ISASI Forum • 21 Figure 3. Altitude, airspeed, heading, and selected heading Figure 4. A portion of the same flight depicted in Figure 3 (note parameters during an earlier flight. Although selected heading change in time scale). The unexpected 360° values of selected generally follows actual heading as would be expected, there heading alternate with values coinciding with the actual are repeated instances where unexpected values of 360° are airplane heading (expected values). recorded. ues were to the left of the airplane heading, time scale. The unexpected 360-degree val- also discovered in the selected course #1 as would be expected during a left turn. The ues can be seen to alternate with values co- parameter on both airplanes but not in the 360° value was unusual as the expected inciding with the actual airplane heading. selected course #2 on either airplane. Based value would still be runway heading at this A common practice among DFDAU on these discoveries, the possibility arose point of the takeoff roll. The recorded se- manufacturers is to use alternating patterns that some sort of design characteristic was lected heading data could have indicated an to indicate errors in the FDR data. For ex- responsible for the observed data. Perhaps unusual procedure by the flight crew, a ample, “stale data” occur when a source there was some difference in the way the malfunction of the mode control panel or stops transmitting data to the DFDAU or selected heading and selected course #1 flight control computer, or something else. the transmitted data are not received by the parameters were processed compared to Thus, one focus of the investigation was to DFDAU. Consultation with the manufac- the selected course #2 data that would ex- understand the actual reason for the un- turer of the DFDAU confirmed that the al- plain the anomaly. usual reading. ternating pattern observed in the FDR data Accordingly, the inquiry focused on how When examining unexpected FDR data, from the accident flight was an error code the DFDAU detected stale data. Accord- a common practice is to use the entire 25- indicating “stale data,” which originated in ing to the DFDAU manufacturer, stale data hour record to determine if the unusual the DFDAU. The stale data error code is are detected as follows: behavior has been present on previous an alternating sequence of 409510 counts • The DFDAU uses an 8-bit counter to track flights. Figure 3 depicts the same four pa- (i.e., 1111111111112) and the last value re- the number of data samples it has received rameters from an earlier flight recorded on ceived. For selected heading, 409510 counts from the source (in this case the FCC). the FDR. The recorded values of selected converts to 360°, therefore the stale data • When scheduled to write a value to the heading generally followed the actual head- error code consists of recorded values of FDR, the DFDAU compares the value of ing (as expected), but there were repeated 360° alternating with the last value received. the counter to the value of the counter the instances where the two differed and the If the inquiry had ended here, one might last time a sample was sent to the FDR. selected heading was recorded as 360°. Dur- conclude that the FCC had malfunctioned • If either the counter value or the data ing some of the times that the 360° values as evidenced by the apparent lack of se- value is different, the DFDAU concludes the were recorded, the airplane was flying on a lected heading transmission to the DFDAU. data are fresh. If both the counter value and heading of approximately 315° with the au- Such a conclusion would be incorrect. the data value are the same, the DFDAU topilot engaged in heading select mode. In addition to the 25 hours of FDR data concludes the value is stale. After three con- With the selected heading 45° to the right available from the accident airplane, the secutive stale samples, the DFDAU begins of the actual heading, the airplane would Egyptian MCA provided 25 hours of FDR writing the stale data error code until ei- have been expected to begin a right turn data from the sister ship. An examination ther the counter value or data value change. toward 360°. However, no such behavior was of that data confirmed the same behavior— Consulting with the FCC manufacturer, observed in the recorded data. selected heading occasionally alternated it was determined that selected heading and Figure 4 depicts a portion of the same between an expected value and 360°. As selected course #1 were transmitted to the flight as shown in Figure 3 at a different shown in Figure 5, the same behavior was DFDAU at a rate of 20 Hz. Thus, the

22 • ISASI Forum April–June 2008 640 to 1. ceived. Figure 6 depicts The anomalies in the selected heading the behavior of the 8- and selected course #1 parameters oc- bit counter when uti- curred frequently but not in every instance lized for selected during which the above conditions are met. course #2 (receive to The last step in the inquiry determined that transmit ratio of 640 the exact receive-to-transmission ratio de- to 1). The capacity of pended upon the relative timing between an 8-bit counter is 0- the FCC internal clock and the DFDAU 255 or 256 distinct val- internal clock, known as jitter. Occasionally, ues. During normal the DFDAU would detect 1,279 or 1,281 operation, the 8-bit samples instead of 1,280 in which case the counter will reach its data would be treated as fresh. maximum value and Once the behavior of the stale data de- “roll over” back to tection algorithm was understood, it was a zero at least twice and simple matter to correct the FDR data to possibly three times accurately reflect the selected heading val- as the 640 samples ues transmitted by the FCC. The DFDAU Figure 5. Similar to selected heading, unexpected values of are received by the will only detect stale data if the parameter 360° were also observed in the selected course #1 parameter DFDAU between value itself is unchanged. Therefore, it was but not in selected course #2. All three parameters are re- each sample trans- possible to conclude that the selected head- corded once every 64 seconds. FDR data from the accidents mitted to the FDR. ing transmitted by the FCC that resulted airplane’s sister ship exhibited the same unexpected values in Regardless of the in the 360° value recorded on the FDR must the selected heading and selected course #1. value of the counter have been the same as the previously re- when a sample is corded value—220, the runway heading. transmitted, the counter will be at a differ- The investigation concluded that the DFDAU received selected heading data ent value (that differs by approximately 128 anomaly in the stale data detection capabil- once every 50 ms and transmitted it once from the previous value) when the next ity of the DFDAU was responsible for the every 64 seconds—a ratio of 1,280 to 1. In sample is transmitted. The result is that the unexpected value of selected heading re- contrast, selected course #2 was transmit- DFDAU can correctly determine if the data corded on the FDR and that the actual value ted by the FCC (and received by the are fresh or stale. of selected heading at this time was 220°. DFDAU) at a rate of 10 Hz for a ratio of Applying the same analysis to selected The corrected value shown in Figure 8 (see heading and selected page 30) depicts the data used for the analy- course #1 yields a sis portion of the investigation. different result. Fig- As often occurs, this investigation uncov- ure 7 depicts the situ- ered a finding not related to the accident ation when the ratio itself—that the DFDAU did not correctly of receive-to-trans- process data when the receive interval-to- Figure 6. Behavior of an 8-bit counter when the receive-to- mit interval is 1,280 transmit interval ratio was a multiple of 256. transmit interval is 640 to 1 as is the case for selected course to 1. In this case, the A full understanding of the provenance of #2. The capacity of an 8-bit counter is 2565. During normal counter rolls over ex- the FDR data allowed for the correct in- operation, the counter will “roll over” twice (or possibly three actly five times be- terpretation of that data for subsequent use times) between each transmission. Regardless of the value of tween each transmit- in the analysis of the accident. the counter when a sample is transmitted, the counter will be at tal to the FDR. Nor- Every investigator or analyst who uses a different value (that differs by approximately 128 from the mal operation will flight data should know that the correct previous value) when the next sample is transmitted with the result in the counter interpretation of flight data requires a full result that the DFDAU correctly detects that the data are fresh. value being the same understanding of the provenance of the when successive data. Each step in the signal chain from samples are trans- measurement to transmission, recording, mitted to the FDR. decoding, conversion, and the final repre- If the parameter sentation can introduce unintended value has not changes and thus the potential for error. changed, the The example discussed above demon- Figure 7. Behavior of an 8-bit counter when the receive-to- DFDAU will incor- strates unintended changes introduced on transmit interval is 1,280 to 1 as is the case for selected rectly detect that the board the accident aircraft. Such unin- heading and selected course #1. The counter will roll over data are stale, even tended changes can also occur during the exactly five times between each transmission. The result is that though the correct subsequent recovery and conversion pro- the DFDAU incorrectly detects that the data are stale during number of samples cesses. Determining the provenance of normal operation. (1,280) has been re- (continued on page 30)

April–June 2008 ISASI Forum • 23 ISASI ROUNDUP

‘Investigation: The Art and the Science’

ISASI 2008, the Society’s 39th annual ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ introduces the need for new techniques international seminar on air accident ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ and support systems to ensure a compre- investigation, is now open for registration, hensive and professional investigation. according to Barbara Dunn, seminar The art of successful investigation chairperson and president of the Cana- requires the creative application of dian SASI (CSASI), which is hosting the personal knowledge and skills and the event to be held in Halifax, Nova Scotia, development of new concepts to keep pace September 8-11. In addition, the Seminar with a rapidly changing industry world Committee is honoring the 100th anniver- wide.” sary of the construction of the Silver Dart He adds, “The Seminar Committee was through its representation in the ISASI tion, P.O. Box 16032, Albuquerque, NM, looking for papers that would deal with the 2008 logo. The aircraft, credited with 87191 USA. hard and soft aspects of investigation, in completing the first controlled power The seminar will be held at the Halifax particular, new ideas that will lead us to flight in Canada and the British Empire, Marriott Harbourfront Hotel. The ISASI improved investigation whether it is was built in 1908 by the Aerial Experi- delegate room rate is $185 Canadian for techniques, management, process, ment Association chaired by Dr. either a single or double and is subject to technology, factual analysis, high tech or Alexander Graham Bell. John A.D. taxes. The special rate is valid to August 7 low tech. The subject matter could be as McCurdy piloted the historic flight on and is available from September 2-16. No broad as the imagination or expertise of Feb. 23, 1909. provisions exist for special rates on the presenter. The Technical Committee The seminar program registration fee upgrade rooms. Rated as a AAA 4- wanted to reach beyond the normal papers (in U.S. dollars) by August 10, is member, Diamond hotel, it is situated in the heart and explore new ideas. We were also very $525; student member, $200; non-member, of downtown, only steps away from the interested in hearing from full-time $570. If registration is made after August city’s top attractions, including the investigators or agencies that have recent 10, the fees are $575, $225, and $625, business district and World Trade & experience with new techniques or respectively. Day pass fee for any of the Convention Center. The hotel is known processes and their experience in applying three days is $200 by August 10, after for its upgraded business amenities, as them. Some ‘soft side’ subjects we were that date $225. The member fee for either well as 17,000 square feet of flexible interested in were subjects ranging from of the two September 8 tutorials is $125 meeting and social-event space. Delegates dealing with the news media, relatives, and by August 10 and $150 after that date; should deal directly with the Halifax interview techniques.” student member, $75 and $100. The Marriott regarding their accommodation The Committee has received more than companion fee is $320 by August 10 and arrangements. The hotel registration 40 proposals for papers from a number of $350 after that date. Registration form is available through a link accessed qualified speakers on a wide range of cancellations made before July 10 will ○○○○○○○○○○○○○○○○○○○○○○○○○○ through the ISASI 2008 seminar website subjects. About 25 of these proposals will incur a $10 fee. Cancellations between (www.isasi.org). be presented in Halifax following a July 27 and August 10, will incur a $75 thorough assessment by the ISASI 2008 fee. There will be no refund of fees for Program plan Papers Selection Committee, which cancellations after August 10. The seminar program will follow the reflects the international aspect of ISASI. The Canadian Society has established a established format of past seminars, with Joining Stewart as members of the detailed and easy-to-manage website 1 day devoted to two tutorial workshops Committee are Barbara Dunn (Canada), accessible through the ISASI website, and 3 days of technical paper presenta- Seminar Committee chair; Nick Stoss www.isasi.org. All areas of delegate tions in plenary session. National society, (Canada), tutorial chair; Ron Schleede interest are easily identified and accessed committee, and working group meetings (USA); Marcus Costa (ICAO); Claudio on the site. A seminar registration form will also be scheduled. ISASI 2008 carries Pandolfi (Chile); Wing Keong Chan may be found on the website and it may the theme “Investigation: The Art and (Singapore); Danny Ho (Taiwan); Martine ○○○○○○○○ be submitted electronically. A copy of the the Science,” that, says Jim Stewart, Del Bono (France); David King (UK); and seminar registration form is also re- technical program chairman, “reflects the Michael Walker (Australia). printed on page 25. Either registration complexity and challenge of today’s The 1-day tutorial sessions will include form may be downloaded or clipped out investigation process. As science advances two workshops. The first will center on and mailed to ISASI Seminar Registra- aviation technology, that same science Safety Management Systems and the

24 • ISASI Forum April–June 2008 39th Annual International Society of Air Safety Investigators Seminar Sept. 8–11, 2008, Marriott Harbourfront Hotel, Halifax, Nova Scotia, Canada Delegate Registration Form and Fee Summary (US$) Yes, please register me for the 39th Annual International Society of Air Safety Investigators Seminar! You can register by e-mailing, mailing, or faxing this completed form to the information below. Please complete one form for the primary individual attending. Exhibitors and companions have a separate registration form. Note: Please print all information on this form. This form may be reproduced as necessary. Cancellations made before July 10, 2008, will incur a $10 fee. Cancellations between July 27, 2008, and Aug. 10, 2008, will incur a $75 fee. There will be no refund of fees if cancelled after Aug. 10, 2008. However, substitutions are permitted at any time. Make sure to include the fees for any optional programs in the total amount being paid.

Please Complete All Areas as Appropriate Is this Your First Seminar? ❏ Ye s ❏ No ISASI Member? ❏ Ye s ❏ No If yes, please complete the member information below: Member number ______Society, chapter, or region: ______❏ Mr. ❏ Ms. ❏ Mrs. ❏ Dr. ❏ Other (If “other” please specify) ______First name ______Middle initial ______

Last name ______Company or organization ______Position, title, or job ______Address line 1 ______

Address line 2 ______City ______State or province ______Country ______ZIP or postal code ______Telephone number ______FAX number ______

E-mail address ______Special meals request (All requests will be honored if possible) ______Name and company as you want it on the badge: ______✁ Registration type Before Aug. 10 After Aug. 10 Optional programs Before Aug. 10 After Aug. 10 ❏ ISASI member US$525 US$575 ❏ Tutorial (Monday, Sept. 8) US$125 US$150 ❏ ISASI student member US$200 US$225 ❏ Tutorial (student member) US$275 US$100 ❏ Not an ISASI member US$570 US$625 Please select one tutorial: Delegate nominated by sponsor (free) ❏ Tutorial #1—Conducting Safety Investigations in a The above registration includes the reception (Mon.), fun night (Tues.), and Safety Management (SMS) Environment ❏ banquet (Thurs.). Please check below if not attending: Tutorial #2—Investigating General Aviation Accidents ❏ Reception ❏ Fun Night ❏ Banquet ❏ Companion Program ❏ Day pass only (per day) US$200 US$225 Check day(s): ❏ Tuesday ❏ Wednesday ❏ Thursday (per person) US$125 US$150 ❏ Banquet only (US$100) ❏ Tuesday fun night (US$100) Note: Please fill out the Companion registration for each companion. ❏ Welcome reception (Monday) (US$90) # of Companion Programs ______

Billing information ❏ Charge my credit card: ❏ AmEx ❏ VISA ❏ MasterCard Name on card ______Card number ______Expiration date ______Card code ______❏ Send by mail: ❏ Payment by check ❏ Company purchase order P.O. # ______TOTAL IN US$ ______Note: Credit card name must be listed on the card. Card billing address must match address listed above in registration. The card code is a four-digit number on the front of an American Express card or a three-digit number on the back of a VISA or MasterCard.

Mail to: ISASI Seminar Registration TEL: +1 (888) 292-2129 (US and Canada) FAX to: +1 (505) 292-2017 P.O. Box 16032, Albuquerque, NM 87191 USA TEL: +1 (505) 299-1690 E-mail to: [email protected]

Signature (required for credit card)______

April–June 2008 ISASI Forum • 25 ISASI ROUNDUP Continued . . .

second will deal with investigating general ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ guides will no doubt relate stories of the ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ temperatures in spring are from 2° to 9° C aviation accidents. days of rum running and privateering as (35.6° to 48.2°F); summer from 16° to 24° C Tutorial No. 1, “Conducting Safety they move through the city streets on (60.8° to 75.2°F); fall about 18° C (64.4°F); Investigations in a Safety Management Tuesday. The tour will travel to winter about -3° C (26.6°F). Weather Systems (SMS) Environment,” will seek Fisherman’s Cove for a two-course lunch forecasts are given in Celsius measure- to answer the questions “What will the enhanced by views of Nova Scotia’s ments. For approximate temperature impact of SMS be on safety investiga- Eastern Passage. conversion: Fahrenheit to Celsius: subtract tions?” and “How do we investigate under On Wednesday, the group will tour the 30 and divide by 2. Celsius to Fahrenheit: the shadow of SMS?” The seminar rugged and beautiful South Shore of multiply by 2 and add 30. website contains more information on the Nova Scotia. The first stop is Airports—Most air traffic comes through contents of the tutorials. Lunenburg, which has been called “the Halifax Stanfield International Airport, prettiest town in Canada.” Settled in the which is the Atlantic Canadian center for Social programs mid 1750s by Germans and Swiss, its domestic, regional, and international In keeping with ISASI tradition, the citizens still retain one of the most flight service. With more than 600 flights seminar social program will start with a interesting accents in North America. a week, travelers can reach Halifax on welcome reception on Monday evening, Now a bustling fishing port, the town’s direct flights from many Canadian, U.S., September 8. This is an ice-breaker social, distinctive architecture and extraordi- European, and Caribbean destinations. providing an opportunity to meet with old nary scenic beauty are a colorful Air carriers serving Halifax include Air and new friends. On Tuesday evening a reminder of its maritime heritage. Canada, Air Canada Jazz, WestJet, special dinner is planned that will allow After lunch at Lunenburg’s The Old Continental Express, Delta, United, attendees to experience some Canadian Fish Factory comes a quick stop at Northwest, American Eagle, Air St. history at the Pier 21 National Historic Mahone Bay with its many cottage Pierre, Condor, Zoom, Go Travel Direct Site. This Canadian equivalent to Ellis industries, craft shops, and famous Vacations, Provincial, Sunwing, and Island welcomed newcomers to Canada “Three Churches.” Then it is on to Skyservice. from 1919 to 1972. Immigration data- Peggy’s Cove, an artists’ and explorers’ Customs and immigration—Immigra- bases, tourist information, and gift and paradise for well more than 150 years. tion regulations: American citizens (or coffee shops are a few of the attractions This picture-postcard village stands on permanent residents) entering Canada by that await guests at the Chrysler Canada solid rock above the crashing surf. The air require valid passports, and as of Welcome Pavilion. Guests will enjoy a coastline is famous for pirates, ship- January 31 valid passports are required reception in Exhibition Hall before being wrecks, rum running, and sunsets. Full when entering by land and sea. Visitors seated in Heritage Hall for a traditional details of the Companion’s Program are from a country other than the United East Coast lobster dinner. available on the ISASI 2008 website. States must carry a valid passport and, in Wednesday evening will be a free night ○○○○○○○○○○○○○○○○○○○○○○○○○○ ○○○○○○○○○○○○○○○○○○○○○○○○○○ certain cases, a visa to be eligible to enter permitting attendees to explore the many Nova Scotia fast facts Canada. All persons entering Canada fine restaurants found in Halifax. The Capital city—Halifax Regional must fill out a declaration for Canada Awards Banquet, at which ISASI’s Municipality. Customs. Jerome F. Lederer Award presentation is Population—Halifax Regional Munici- Currency—Canada’s currency is based made, will be held on the Thursday pality 382,203, Nova Scotia 934,405. on the decimal system, with 100 cents to evening at the Harbourfront Marriott. Languages—The official languages of the dollar. The usual post-seminar optional tour on English and French are spoken through- Sales tax and rebates—The harmonized Friday is not being offered. out Nova Scotia. sales tax (HST) is applied at a single rate Time zone—Nova Scotia is on Atlantic of 13% to a base of goods and services. Companion’s Program Daylight Time, which is 4 hours earlier Foreign visitors may be entitled to claim a The Companion’s Program, organized by than Greenwich Mean Time and 1 hour rebate of the HST paid. Accommodations Gail Stewart and Paula Demone, includes later than North America’s Eastern Time in Halifax charge a hotel levy of 2% on a deluxe Halifax City tour on Tuesday and Zone. Daylight Saving Time took effect room rates to assist in marketing Halifax a visit to the South Shore on Wednesday. the second Sunday of March and contin- as a business and leisure destination. Both are all-day tours and promise ues to the first Sunday of November. Hospital/medical services—Visitors to exciting historical narratives. Kilted Climate and weather—Average daily Canada are strongly urged to obtain

26 • ISASI Forum April–June 2008 2007 Annual Seminar Proceedings Now Available

Active members in good standing and page at http://www.isasi.org. The seminar US$75 fee. A limited number of paper corporate members may acquire, on a papers can be found in the “Members” copies of Proceedings 2007 are no-fee basis, a copy of the Proceedings section. Alternatively, active members available at a cost of US$150. Checks of the 38th International Seminar, may purchase the Proceedings on a CD- should accompany the request and be held in Singapore Aug. 27-30, 2007, by ROM for the nominal fee of $15, which made payable to ISASI. Mail to ISASI, downloading the information from the covers postage and handling. Non-ISASI 107 E. Holly Ave., Suite 11, Sterling, appropriate section of the ISASI web members may acquire the CD-ROM for a VA USA 20164-5405.

Preface: Welcome to Singapore Using Checklists as an Investigator’s Tool SESSION 5—Moderator Danny Ho By Frank Del Gandio, President, ISASI By Al Weaver International Cooperation and Challenges: Opening Address: Importance of Interna- Understanding Cross-Cultural Issues tional Cooperation in Aircraft Accident SESSION 3—Moderator Alan Stray By Dr. Wen-Chin Li, National Defense Investigation Finding Nuggets: Cooperation Vital in Efforts University; Dr. Hong-Tsu Young, Taiwan, By Raymond Lim, Minister for Transport to Recover Buried Data ASC; Thomas Wang, ASC; and Dr. Don and Second Minister for Foreign Affairs, By Christophe Menez and Jérôme Projetti, BEA Harris, Cranfield University Singapore International Investigation: General Aviation Very Light Jets: Implications for Safety Keynote Address: Sharing Experience Accident in Atlantic Waters And Accident Investigation And Knowledge By Joseph Galliker, ASC International, Inc. By Dr. Robert Matthews, Ph.D., FAA By Mark V Rosenker, Chairman, U.S. Standardizing International Taxonomies for Enhanced Airborne Flight Recorder National Transportation Safety Board Data-Driven Prevention (EAFR)—The New Black Box Lederer Award Recipient: ‘Independence By Corey Stephens, Air Line Pilots Association; By Jim Elliot, G.E. Aerospace and Integrity’ Mark Tom McCarthy Oliver Ferrante, BEA; Kyle Olsen, FAA; and RSAF: Analysis and Investigation; Tools By Esperison Martinez, Editor Vivek Sood, FAA and Techniques Midair Collision Over Brazilian Skies— By Lt. Col. Suresh Navaratnam, Republic of SESSION 1—Moderator David McNair A Lesson to Be Learned Singapore Air Force (RSAF) Royal Australian Navy Sea King Accident By Col. Rufino Antonio da Silva Ferreira, José Wet Runway Accidents—The Role of Investigation—Indonesia April 2, 2005 Mounir Bezerra Rahman, and Carlos Eduardo Fatigue and Coercive Habits By Nicholas Athiniotis and Domenico Magalhães da Silveira Pellegrino, Brazilian By Capt. A. Ranganathan Lombardo, Defence Science and Technology Aeronautical Accident Investigation Commis- Organization, Australia sion (CENIPA); William English, NTSB; and SESSION 6—Moderator David King Russia/France: Safety and Cultural Nick Stoss, TSB Canada ISASI International Working Group on Challenges in International Investigations Human Factors: A Progress Report By Alexey N. Morozov, Interstate Aviation SESSION 4—Moderator Richard Breuhaus By Capt. Richard Stone, ISASI and Dr. Committee and Sylvain Ladiesse, BEA Convair 580 Accident Investigation: A Study in Randy Mumaw, Boeing International Cooperation Paves the Synergy International Coorperation During Recent Runway for a Safer Sky By Ian McClelland, TAIC, New Zealand Major Aircraft Accident Investigations in By Guo Fu, East China Administration, Tenerife to Today: What Have We Done in Nigeria CAAC 30 Years To Prevent Recurrence? By Dennis Jones, NTSB By Ladislav Mika, Ministry of Transport, Czech Critical Aspects of International Incident SESSION 2—Moderator Sue Burdekin Republic, and John Guselli, JCG Aviation Investigations Winter Operations and Friction Measure- Services By Deborah J. Lawrie, Robert N. van Gelder, ments Flight Data: What Every Investigator and Jan Smeitink, Independent Safety By Knut Lande, Accident Investigation Should Know Investigation & Consultation Services Board, Norway By Michael Poole, Flightscape, Inc., and Simon National Transportation Safety Commit- Utilization of the Web-Based GIS to Assist Lie, Boeing tee of Indonesian Presentation Aviation Occurrence Investigation Sound Identification and Speaker Recogni- By Tatang Kurniadi, Chairman, National By Tien-Fu, Yeh, Wen-Lin Guan, and Hong tion for Aircraft Cockpit Voice Recorder Transportation Safety Committee, Indonesia T. Young, Aviation Safety Council By Yang Lin, Center of Aviation Safety Going the Extra Mile Use of Reverse Engineering Techniques to Technology, CAAC and Wu Anshan and Liu By Donald F. Knutson (Accepted for Generate Data for Investigations Enxiang, General Administration of Civil presentation, but not orally delivered due to By Peter Coombs, AAIB, UK Aviation of China, CAAC exigent circumstances.) ◆ health insurance before leaving their ○○○○○ ISASI Reachout Enters 8th ○○○○○ world. Ladislav Mika, the first host, and home country. Canadian hospital and Prague, the Czech Republic, where the medical services are excellent, but a Year in Need of Funds first workshop was held in May 2001, will hospital stay can be costly without ○○○○○○○ It may seem hard to believe but on May 7, ○○○○○○○ forever be linked to the success of the adequate insurance coverage. Visitors the ISASI Reachout program enters its Reachout program. taking prescribed medications are advised eighth year of operation. During that The past 12 months saw a flurry of to take a copy of the prescription should it time, volunteer instructors have delivered activity as ISASI instructors traveled need to be renewed during the trip. ◆ 25 successful workshops around the extensively around the world—

April–June 2008 ISASI Forum • 27 ISASI ROUNDUP New Members CORPORATE Houston, Mark, D., Hawkes Bay, Continued . . . Finnish Military Aviation Authority New Zealand Major (Eng.) Kimmo T. Nortaja, Leading Husain, Muhammad, M., Arad, Kingdom Aviation Accident Investigator of Bahrain Mikko T. Hietanen, Chief of FAIF Jackson, Robert, J., Issaquah, WA, USA Flight Safety Jay, Susan (Sue), M., Orange Park, FL, USA Northwest Airlines Kelman, Sarah, J., Cambridge, United Lisa G. Brockenbrough, Director, Flight Kingdom Safety and Industry Affairs Kunteson, Randall, R., Deatsville, AL, USA

○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ Todd Tilbury, Manager, Flight Safety Kwajan, Francis, A., Grand Central Station, • Reachout #21 Kiev, Ukraine, October Dombroff Gilmore Jaques & French P.C. NY, USA 2007 Mark A. Dombroff, Managing Partner Labrucherie, Stephan, Abu Dhabi, United • Reachout #22 Santiago, Chile, Novem- Dane B. Jaques, Partner Arab Emirates General Aviation Manufacturers Association Lammas, Janet, I., Wellington, New Zealand ber 2007 W. Casey Kinosz, Manager of Operations Mayfield, Richard, F., Renton, WA, USA • Reachout #23 Karachi, Pakistan, No- Jens C. Hennig, Vice-President Miller, Andrew, B., Christchurch, vember 2007 of Operations New Zealand Miranda, Ray, M., Kajang, West Malaysia • Reachout #24 Abu Dhabi, UAE, Novem- INDIVIDUAL Moussa, Ali, M., Beirut, Lebanon ber 2007 Adkins, John, M. ,Daytona Beach, FL, USA Naseer, Ahsan, Abu Dhabi, United Arab • Reachout #25 Brisbane, Australia, Al Marzouqi, Fouad, B., Abu Dhabi, United Emirates Arab Emirates Nitta, Takashi, Toyonaka, Osaka, Japan March 2008 Al Said, Saud, T., Sharjah, United Arab Noe, II, Ronald, E., Wichita, KS, USA Currently scheduled workshops Emirates Nutter, Christopher, G., Federal Way, include— Al-Awadhi, Kamil, H., Quadsiah, Kuwait WA, USA Alegado, Jaime, V., Doha, Qatar Ordor, Maria, U., Mafoluku-Lagos, Nigeria • Reachout #26 Seattle, USA, April 2008 Anglin, Lori, M., Renton, WA, USA Pinzón Müller, Eduardo, Bogota, Colombia, • Reachout #27 Mumbai, India, April Awbery, Natalie, Godalming, Surrey, South America 2008 United Kingdom Pooley, Edward, J., Isle of Man, British Isles Balentine, Chad, J., Winchester, VA, USA Qarashi, Baha, G., Abu Dhabi, United • Reachout #28 Bahrain, UAE, May Barros, Andrés, Santiago, Chile Arab Emirates 2008 Barth, Thomas, H., Denver, CO, USA Richter, Mark, W., Hanover Park, IL, USA • Reachout #29 Karachi (Unscheduled) Benassi, Stefano, VR, Italy Rigby, Kevin, T., Pensacola, FL, USA Bonnassie, Fabien, Montreuil, France Rivera-Martinez, Alimael, Prescott Valley, But ISASI Reachout Committee Boyd, Rodney, L., Wylie, TX, USA AZ, USA chairman, Jim Stewart, sees problems in Cabel Francis, Abu Dhabi, United Arab Rogachuk, Ted, M., Bay Village, OH, USA the future for the Reachout program if Emirates Ronaldson, George, Aberdeen, United Chen, Felix, S.K., Singapore Kingdom funds are not received to replenish the Chiang, Vin, C., Daytona Beach, FL, USA Ross, Cameron, C., Southlake, TX, USA main Reachout account. “We have been de Kock, André, L., Montréal, Canada Satti, Juan, A., Castelar, Argentina very successful in obtaining local De La Vauvre, Gaetan, Evanston, IL, USA Savage, Eric, R., Prescott, AZ, USA De Waal, Johan, G., Abu Dhabi, United Schonhardt, Carlos, F.G., Rio De Janeiro, sponsorship for individual workshops,” Arab Emirates Brazil Stewart said recently. “We have been Decatur, Danielle, M., Ocean, NJ, USA Senthil, M.S., Gopalakrishnan, Seeb Int’l particularly blessed with dynamic Dyer, Brian, R., Daytona Beach, FL, USA Airport, Sultanate of Oman Eisenman, Nathaniel, B., Manassas, USA Shappee, Eric, H., Salina, KS, USA individuals volunteering to host our Elgindi, Hisham A. Aziz, Abu Dhabi, Shearer, Daniel, J., Daytona Beach, FL, USA workshops and with national and United Arab Emirates Skoor, Erich, J., Prescott, AZ, USA regional airlines providing air travel.” Engroba, Juan, D., Buenos Aires, Argentina Smith, Sean, J., Daytona Beach, FL, USA Farmiga, Samuel, R., West Chester, OH, USA Stauffer, Kent, M., Stow, OH, USA Stewart pointed out that their support Fragoso, Jefferson, V., Sao Paulo, Brazil Talay, James, E., Mission Viejo, CA, USA has been considerable in many cases and Franklin, Ryan, H., Ormond Beach, FL, USA Torres, Roberto, H., Ormond Beach, without that support the program could ○○○○○○○○○○○○○○○○○○○○○○○○○○ Gipson, Paul, Scottsdale, AZ, USA FL, USA Goetz, Jean-Baptiste, Abu Dhabi, United Trono, Peter, J., Pearblossom, CA, USA not have been able to sustain the level of Arab Emirates Yusof, Mohd Hisham Md., Abu Dhabi, activity it has to date. He concluded, Graydon, Mike, Abu Dhabi, United Arab United Arab Emirates “But our ability to support our instruc- Emirates Zanolli, Stefano, Roma, Italy Holland, Dennis, A., Northamptonsire, Zwager, Finn, Dubai, United tors is not as strong as I would like. United Kingdom Arab Emirates ◆ “We need to top up our emergency fund to ensure that our instructors are not paying their own way,” Stewart said. with both he and the ISASI treasurer use of organizational materials and staff, “That Reachout fund is kept in a separate having to jointly approve expenditures and many other small and large contribu- account at the ISASI head office. We use before the fact. tions. Stewart said, “Without the contin- that fund to reimburse instructors when ISASI Reachout was originally ued support of sustaining organizations local sponsors can not cover all of the conceived as a way for ISASI corporate like ICAO, the Air Line Pilots Associa- associated costs,” Stewart pointed out, members to participate in the work of the ○○○○○○○ tion, and Continental Airlines, our job “and that fund is running dangerously low Society. To a great extent, some ISASI would be much more difficult.” as a result of our activity levels.” Stewart member corporations have responded in ISASI Reachout needs to increase the also pointed out that access to the kind and provide access to airline seats, number of sustaining sponsors and needs emergency funds is strictly controlled free access to documents and manuals, to receive additional “one time” donations

28 • ISASI Forum April–June 2008 REMINDER

ISASI annual dues were due in January. For those members who may not have yet made the payment, please contact Ann Schull at [email protected] or call 703-430- 9668 to make payment arrangements. If payment is not received, the affected member will be placed ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ sponsored a presentation to the campus ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ provenance and record search by which in an inactive status. ◆ by retired NTSB senior air safety we have been able to document what is investigator Greg Feith. Members from known about this long-lost young pilot. the Chapter have been assisting regularly We are currently making plans to return from a broader base of corporations. in maintaining and refurbishing accident the ID to Shipley’s family.” “When we started the Reachout program, scenarios in the Aircraft Accident we established a one-time Charter Laboratory. The Chapter is sponsoring a New Book Outlines Sponsor category for original sponsors,” trip to Williams-Gateway Airport in May Stewart said. for those members who would like to ‘The Limits of Expertise’ “We can’t bring that category back, but experience an altitude chamber ride. The The Limits of Expertise: Rethinking we will create a new category, ‘Sustaining student section has just held officer Pilot Error and the Causes of Airline Sponsor,’” he said. “Sustaining Sponsors elections, and Erich Skoor (ST5519) will Accidents, by ISASI member Dr. Loukia are those sponsors we can count on to be be the new president. Loukopoulos and coauthors Benjamin A. there when needed throughout the year.” On Nov. 16, 2007, the Chapter con- Berman and R. Key Dismukes, argues For those corporations or individuals ducted an expedition to the World War II that human skill and vulnerability to error who wish to contribute to the Reachout B-24 crash site on Humphries Peak near are closely linked. program, even a one-time contribution Flagstaff, Ariz. This was the single In paperback, it is 352 pages, published will help, no matter how large or small. largest trip to the crash site, with 35 Contributions should be payable to ISASI people participating. The aircraft crashed Reachout and sent to ISASI headquar- during a night training mission in ters at Park Center, 107 East Holly Ave., September 1944. All eight airmen aboard MOVING? Suite 11, Sterling, VA, 20164-5405. were killed. The wreckage is located In a final comment, Stewart again between 11,000 and 11,500 feet on the Please Let Us Know thanked the instructors, the local hosts, mountain and, due to the terrain, is Member Number______and the current Sustaining Sponsors for difficult to access. their continued support. “They are the Most of the wreckage is still there, and Fax this form to 1-703-430-4970 or mail to ISASI, Park Center reason we are here today,” he said. “Their one of the long-term projects the Arizona 107 E. Holly Avenue, Suite 11 contribution has been large, their Chapter has undertaken is to conduct Sterling, VA USA 20164-5405 commitment strong. You can help by periodic surveys of the wreckage layout Old Address (or attach label) joining the ISASI Reachout sponsorship and positions of major pieces. Name ______group as a ‘Sustaining Sponsor’ or a one- Waldock says, “We have been up there time contributor. Make your contribution five previous times, but this time we had a Address ______○○○○○○○○○○○○○○○○○○○○○○○○○○ ○○○○○○○○○○○○○○○○○○○○○○○○○○ today,” he concluded. ◆ very significant occurrence. Our student City ______section president, Cavi Freeland, was State/Prov. ______helping examine an area where compo- Arizona Chapter Zip ______nents from the front of the fuselage had Continues Expansion been identified. She noticed an odd piece Country ______The student section of the Arizona chapter just below the surface of the soil and New Address* continues to expand. Chapter President teased it up with a finger. It was a gold Name ______Bill Waldock says, “We currently have 16 airmen’s style ID bracelet piece with the Address ______active members, with a big push for them name ‘Ray Shipley’ engraved on one side to join the International Society. Currently, and ‘DRU’ on the other. The crew City ______the students meet every 2 weeks in the manifest shows Flight Officer Ray State/Prov. ______Robertson Aviation Safety Center at Shipley as the ‘second pilot’ on the Zip ______Embry-Riddle’s Prescott Campus. We aircraft. Research has shown that his Country ______have had several guest speakers, including wife’s name was “Drucilla” and that he Prof. Denny Lessard and FAA Mainte- has living relatives in Garland, Tex. Craig E-mail ______nance Inspector Pete Kelly.” Fuller of Aviation Archaeology Investiga- *Do not forget to change employment and In November 2007, the Chapter tion and Research has assisted with the e-mail address.

April–June 2008 ISASI Forum • 29 by Ashgate Publishing Company. It is ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ ○ ○ ○ ○ Instead, the authors emphasize the available through Amazon.com, where interactions of subtle variations in task sufficient reviews are also posted. One of demands, incomplete information ISASI Information those reviews, by Benjamin Daley, cites it available to pilots, and the inherent as an “Outstanding and original book… nature of skilled performance.” that show[s] that the presence and inter- He adds, “The book will be informative action of factors contributing to error is for diverse readers in the air transport probabilistic rather than deterministic.” industry, including operational staff, Stephen Wilkinson in his review of the researchers, safety analysts, accident book published in Air and Space maga- investigators, designers of systems and OFFICERS zine noted that the book “simply examines procedures, training providers, and President, Frank Del Gandio ([email protected]) the 19 serious accidents suffered by major students…. Executive Advisor, Richard Stone U.S. air carriers between 1991 and 2000 in “The main significance of this book is ([email protected]) Vice-President, Ron Schleede which the NTSB cited crew error as in its re-framing of the causes of airline ([email protected]) playing a central role.” accidents: the authors argue that, if we Secretary, Chris Baum ([email protected]) Daley, in his review, says: “While the must continue to conceive of airline Treasurer, Tom McCarthy NTSB must determine the probable accidents in terms of deficiency, then that ([email protected]) cause of each specific accident, the deficiency should be attributed to the authors take a different approach: would overall air transport system. Such an COUNCILLORS Australian, Lindsay Naylor other pilots be vulnerable to making the approach can contribute to aviation ([email protected]) kinds of errors made by the accident crew safety by providing a foundation for Canadian, Barbara Dunn ([email protected]) and, if so, why? This original approach improving equipment, training, proce- European, Anne Evans reveals factors that make all pilots dures, and organizational policy. In so ([email protected]) International, Caj Frostell vulnerable to specific types of error in doing, it is possible to reduce the ([email protected]) certain situations. In adopting this frequency of ‘system accidents’ and to New Zealand, Vacant United States, Curt Lewis approach, the authors challenge the devise adequate protection against the ([email protected]) assumption that, if expert pilots make types of errors to which many, if not all, errors, this is evidence of their lack of pilots—as well as many other experts— NATIONAL AND REGIONAL skill, vigilance, or conscientiousness. are vulnerable.” ◆ SOCIETY PRESIDENTS Australian, Lindsay Naylor ([email protected]) Canadian, Barbara M. Dunn Flight Data—What Every Investigator Should Know (from page 23) ([email protected]) European, David King ([email protected]) each parameter and Latin American, Guillermo J. Palacia (Mexico) understanding the New Zealand, Peter Williams capabilities and re- ([email protected]) Russian, Vsvolod E. Overharov play processes with- ([email protected]) in the analysis soft- SESA-France Chap.,Vincent Fave ([email protected]) ware is a necessary United States, Curt Lewis step in interpreting ([email protected]) flight data. ◆ UNITED STATES REGIONAL (Space restrictions CHAPTER PRESIDENTS Alaska, Craig Bledsoe required deletion ([email protected]) from the authors’ Arizona, Bill Waldock ([email protected]) presented paper Dallas-Ft. Worth, Curt Lewis their discussion ([email protected]) Florida, Ben Coleman dealing with the ([email protected]) processing of flight Great Lakes, Matthew Kenner ([email protected]) data through two Los Angeles, Inactive methods: Convert- Mid-Atlantic, Ron Schleede ([email protected]) ing the binary ones Northeast, David W. Graham and zeros into engi- Figure 8. The same data as depicted in Figure 1 corrected to ([email protected]) Pacific Northwest, Kevin Darcy neering units data account for the anomaly discovered in the way the DFDAU ([email protected]) and producing com- processed the selected heading data. These were the data used Rocky Mountain, Gary R. Morphew ([email protected]) ma separated vari- for the analysis portion of the investigation. San Francisco, Peter Axelrod ables files and inter- ([email protected]) Inactive actively working with the binary data on a be viewed on the ISASI web page in Pro- Southeastern, demand basis. The unused material may ceedings 2008, page 95.—Editor)

30 • ISASI Forum April–June 2008 COMMITTEE CHAIRMEN American Underwater Search & Survey, Ltd. Independent Pilots Association Audit, Dr. Michael K. Hynes AmSafe Aviation Int’l Assoc. of Mach. & Aerospace Workers ([email protected]) Aramco Associated Company Interstate Aviation Committee Award, Gale E. Braden ASPA de Mexico Irish Air Corps ([email protected]) Association of Professional Flight Attendants Irish Aviation Authority Ballot Certification, Tom McCarthy Atlantic Southeast Airlines—Delta Connection Japan Airlines Domestic Co., LTD ([email protected]) Australian Transport Safety Bureau Japanese Aviation Insurance Pool Board of Fellows, Vacant Aviation Safety Council Jeppesen Bylaws, Darren T. Gaines JetBlue Airways ([email protected]) Avions de Transport Regional (ATR) Code of Ethics, Jeff Edwards BEA-Bureau D’Enquetes et D’Analyses Jones Day ([email protected]) Board of Accident Investigation—Sweden KLM Royal Dutch Airlines Membership, Tom McCarthy Boeing Commercial Airplanes Korea Air Force Safety Ctr. ([email protected]) Bombardier Aerospace Regional Aircraft Korea Aviation & Railway Accident Investigation Board Nominating, Tom McCarthy Bundesstelle fur Flugunfalluntersuchung—BFU Kreindler & Kreindler, LLP ([email protected]) Cathay Pacific Airways Limited L-3 Communications Aviation Recorders Reachout, James P. Stewart Cavok Group, Inc. Learjet, Inc. ([email protected]) Lockheed Martin Corporation Seminar, Barbara Dunn Centurion, Inc. ([email protected]) Charles Taylor Aviation, Singapore Lufthansa German Airlines China Airlines MyTravel Airways WORKING GROUP CHAIRMEN Cirrus Design National Aerospace Laboratory, NLR Civil Aviation Safety Authority Australia National Air Traffic Controllers Assn. Air Traffic Services, John A. Guselli (Chair) National Business Aviation Association ([email protected]) Colegio De Pilotos Aviadores De Mexico, A.C. Ladislav Mika (Co-Chair) ([email protected]) Comair, Inc. National Transportation Safety Board Cabin Safety, Joann E. Matley Continental Airlines NAV Canada ([email protected]) Continental Express Nigerian Ministry of Aviation and Accident Corporate Affairs, John W. Purvis COPAC/Colegio Oficial de Pilotos de la Aviacion Comercial Investigation Bureau ([email protected]) Cranfield Safety & Accident Investigation Centre Northwest Airlines Flight Recorder, Michael R. Poole Curt Lewis & Associates, LLC Parker Aerospace ([email protected]) DCI/Branch AIRCO Phoenix International, Inc. General Aviation, William (Buck) Welch Pratt & Whitney ([email protected]) Defence Science and Technology Organization (DSTO) Government Air Safety, Willaim L. McNease Delta Air Lines, Inc. Qantas Airways Limited ([email protected]) Directorate of Aircraft Accident Investigations— Qatar Airways Human Factors, Richard Stone Namibia Qwila Air (Pty), Ltd. ([email protected]) Directorate of Flight Safety (Canadian Forces) Raytheon Company Investigators Training & Education, Directorate of Flying Safety—ADF Republic of Singapore Air Force Graham R. Braithwaite Dombroff Gilmore Jaques & French P.C. Rolls-Royce, PLC ([email protected]) Dutch Airline Pilots Association Royal Netherlands Air Force Dutch Transport Safety Board Royal New Zealand Air Force EL AL Israel Airlines RTI Group, LLC CORPORATE MEMBERS Embraer-Empresa Brasileira de Aeronautica S.A. Sandia National Laboratories AAIU Ministry of Transport Bulgaria Embry-Riddle Aeronautical University SAS Braathens Accident Investigation Board, Finland Emirates Airline Saudi Arabian Airlines Accident Investigation Board/Norway Era Aviation, Inc. SICOFAA/SPS Accident Investigation & Prevention Bureau European Aviation Safety Agency Sikorsky Aircraft Corporation Aeronautical & Maritime Research Laboratory EVA Airways Corporation Skyservice Airlines, Ltd. AeroVeritas Aviation Safety Consulting, Ltd. Exponent, Inc. Singapore Airlines, Ltd. Aerovias De Mexico, S.A.De C.V. Federal Aviation Administration SNECMA Moteurs Air Accident Investigation Bureau of Singapore Finnair Oyj South African Airways Air Accident Investigation Unit—Ireland Finnish Military Aviation Authority South African Civil Aviation Authority Air Accidents Investigation Branch—U.K. Flight Attendant Training Institute at Melville College Southern California Safety Institute Air Canada Pilots Association Flight Safety Foundation Southwest Airlines Company Air Line Pilots Association Flight Safety Foundation—Taiwan Southwest Airlines Pilots’ Association Air New Zealand, Ltd. Flightscape, Inc. Star Navigation Systems Group, Ltd. Airbus S.A.S. Galaxy Scientific Corporation State of Israel Airclaims Limited General Aviation Manufacturers Association Transport Canada Aircraft Accident Investigation Bureau—Switzerland GE Transportation/Aircraft Engines Transportation Safety Board of Canada Aircraft Mechanics Fraternal Association Global Aerospace, Inc. U.K. Civil Aviation Authority Aircraft & Railway Accident Investigation Commission Gulf Flight Safety Committee, Azaiba, Oman UND Aerospace Airservices Australia Hall & Associates, LLC University of NSW Aviation AirTran Airways Hellenic Air Accident Investigation University of Southern California Alaska Airlines & Aviation Safety Board Volvo Aero Corporation Alitalia Airlines—Flight Safety Dept. Honeywell WestJet ◆ All Nippon Airways Company Limited Hong Kong Airline Pilots Association Allied Pilots Association Hong Kong Civil Aviation Department American Eagle Airlines IFALPA

April–June 2008 ISASI Forum • 31 WWHO’SHO’S W WHOHO

WestJet Navigates the SMS Airways ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ ○ ○ ○ ○ ○ ○ ○ ○ (Who’s Who is a brief profile of, and pre- ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ A key milestone was the deployment of reviews the report and implements pared by, the represented ISASI corporate the company safety database in April corrective actions, as required, according member organization to enable a more 2006. Hosting more than 60 users, the to time lines specified in the company thorough understanding of the organiza- safety database supports WestJet’s SMS safety manual. Follow-up assessment of tion’s role and functions.—Editor) processes, which manage both proactive corrective action effectiveness is then and reactive safety reports in a similar conducted and the entire process is closed ased in Calgary, Alberta, WestJet is manner. Other database features include off by the company’s Safety Management Canada’s original low-fare airline. online safety reporting, corrective action Committee, which is chaired by its BWestJet took flight on Feb. 29, tracking, permanent occurrence and accountable executive. 1996, serving five destinations with three investigation logs, data analysis tools, and WestJet’s SMS is very specific as to the Boeing 737-200 aircraft and 220 employ- roles of the safety and leadership teams. ees. As of Nov. 30, 2007, WestJet has Safety teams facilitate development, grown, offering scheduled and charter implementation, and operation of safety service to 38 cities across North America management processes; however, the and the Caribbean with a fleet of 70 scheduling of audits and repetitive tasks. senior leadership team retains overall Boeing “next generation” 737 aircraft and Common standards for operational risk responsibility for all operational risk a team of more than 6,700 employees. management have been implemented management processes and risk control WestJet’s fleet is the most modern and across the company. Once a hazard is activities. SMS elements have been efficient in North America, featuring identified, it receives an initial risk integrated into the company’s orientation, blended winglets on all B-737-700 and B- classification that drives the follow-up strategic planning, and training pro- 737-800 aircraft that minimize fuel process and time line. Risk classifications grams. Extensive use is made of interac- consumption and environmental impact. are validated by expert consensus or tive computer-based training and The company has plans to grow its fleet to through full investigations, which are electronic documentation so that employ- more than 100 aircraft by 2012. required for all elevated risk reports. ees throughout the network can access The company has experienced expo- A typical investigative follow-up sees a SMS resources and apply SMS principles nential growth over its short 11½ years in department safety team log the occur- in their daily activities.

operation. Year 2006 revenue was ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ rence report, communicate with the With an ingrained mission to “enrich reported at $1.8 billion, available seat reporter, assign an initial risk rating, the lives of everyone in WestJet’s world miles at 12.5 billion, and revenue passen- conduct an investigation, convene a safety by providing safe, friendly, and affordable ger miles at 9.8 billion. More than 10 review team, and publish a report of air travel,” WestJet is integrating SMS million guests (known as passengers with findings and causes. Following this throughout the organization as it strives most other carriers) were carried in 2006. process, a management representative to be the No. 1 choice for air travelers. ◆ WestJet credits its business success in large part to the power of its positive corporate culture. The work environment ISASI NON-PROFIT ORG. is positive, with management empowering 107 E. Holly Ave., Suite 11 U.S. POSTAGE PAID its people to make the best decisions for Sterling, VA 20164-5405 USA Permit #273 the company, its guests, and its sharehold- CHANGE SERVICE REQUESTED ANNAPOLIS, MD ers. WestJet’s people-oriented environment has been a key factor in its successful implementation of Safety Management Systems (SMSs) on which it embarked in 2005 under the Canadian aviation regulations. Throughout the last 2 years, WestJet has documented existing safety management processes, implemented new ones, and built an infrastructure to support all SMS elements. Full SMS implementation is scheduled for the autumn of 2008.

32 • ISASI Forum April–June 2008