飛航安全與人為因素 行政院飛航安全委員會 報告人 王興中 飛航安全與人為因素 行政院飛航安全委員會 報告人 王興中

Flight Safety 飛航安全

U.S. General Aviation U.S. Navy/Marine Corps U.S. Air Force Source: Boeing 10 20 30 40 50 1960 1970 1980 1990 Scheduled Air Carrier Accidents/100,000 flight hours U.S. General Aviation Source: NTSB Improvements in aviation safety, however, are not unique to commercial carriers. General aviation deaths and fatal accident rates in the U.S. declined to a 15-year low in 1996, with only 1.51 accidents occurring per 100,000 flight hours (NTSB, 1997). Aviation accidents within the U.S. military (i.e., Army1, Navy, Air Force, and Marine Corps) have also decreased steadily over the past 2 decades. The rate of major accidents in the U.S. military, calculated as the number of accidents per 100,000 flying hours, declined from about 4.3 in 1975 to 1.5 in 1995. In fact, if one were to examine any Service organization or even the civilian sector they would all look essentially the same (USN/USMC - upper right; USAF - upper left; commercial airlines - bottom). Specifically, they all reveal the same downward trend throughout the 50s, 60s and into the early 70s. Many have attributed this stark decline in the overall mishap rate to improved design, materials, training, and the implementation of standardized training programs. Notably, however, all the graphs show the same “flattening” of that positive downward trend in the mishap rate over the last couple decades. 1 The U.S. Army rates between 1950 and 1972 were unavailable at the time of publication. Accidents/100,000 flight hours U.S. Navy/Marine Corps Source: U.S. Naval Safety Center Accidents/100,000 flight hours U.S. Air Force Source: U.S. Air Force Safety Center

在一次飛航中死亡的機率 搭乘全世界飛安紀錄最好的前25名航空公司 七百七十一萬分之一 搭乘全世界飛安紀錄最差的25名航空公司 五十五萬八千分之一 The red (bottom) line in this graph shows worldwide trends in aviation accident rates, as well as projected accident rates through the year 2010. The green (middle) line depicts the traffic growth which is expected to increase dramatically over the next 10 years. The blue (top) line shows the predicted increase in accident frequency due to the rapid industry expansion. Note that this predicted increase is based on the current accident rate; therefore, even if the accident rate stays the same over the next decade, the raw number of accidents will increase markedly. Furthermore, as can be seen from the graph, there may be as many as 52 accidents a year worldwide during the first decade of the new century. This translates into an astonishing one accident a week. Note. Graph adapted from Flight Safety Foundation (1997). Values plotted in the graph are estimates based on industry statistics.

飛航架次與失事率 Departure (millions) / Rate per million Accidents Year 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Year 5 10 15 20 25 30 35 Departure (millions) / Rate per million 40 50 60 70 Accidents Traffic Growth Accident Rate 1 2 3 1 Based on current accident rate 2 Based on industry estimates 3 Based on current accident rate Number of Commercial Jet Accidents, Accident Rate and Traffic Growth - Past, Present and Future The red (bottom) line in this graph shows worldwide trends in aviation accident rates, as well as projected accident rates through the year 2010. The green (middle) line depicts the traffic growth which is expected to increase dramatically over the next 10 years. The blue (top) line shows the predicted increase in accident frequency due to the rapid industry expansion. Note that this predicted increase is based on the current accident rate; therefore, even if the accident rate stays the same over the next decade, the raw number of accidents will increase markedly. Furthermore, as can be seen from the graph, there may be as many as 52 accidents a year worldwide during the first decade of the new century. This translates into an astonishing one accident a week. Note. Graph adapted from Flight Safety Foundation (1997). Values plotted in the graph are estimates based on industry statistics.

2002年全世界失事紀錄 01/14 Ibertrans Aerea Embraer 120RT Zaldivar, Spain (3/3)-(0/0)-(3/3) 01/15 Procuraduria General De Havilland DHC-6 Chilpancingo, Mexico (0/4)-(0/0)-(3/15) 01/16 Garuda Indonesia Boeing 737-3Q8 Klaten, Indonesia (1/0)-(0/0)-(6/54) 01/17 Petroproduccion Fairchild FH-227E El Tigre, Colombia (5/21)-(0/0)-(5/21) 01/28 TAME Ecuador Boeing 727-134 Cumbal Volcano, (9/83)-(0/0)-(9/83) 02/07 Volare Aviation Antonov AN-12BP Agadir, Morocco (8/0)-(0/0)-(8/0) 02/12 Iran Air Tours Tupolev TU-145M Khorramabad, Iran (12/107)-(0/0)-(12/107) 02/15 Tiramavia Antonov 12BP Monrovia-Roberts (1/0)-(0/0)-(8/0) 03/15 Aerotaxi (Cuba) Antonov AN-2 Baez, Cuba (2/14)-(0/0)-(2/14) 03/17 Djibouti Airlines Let 410 Djibouti, Africa (4/0)-(0/0)-(4/0) 04/12 Tadair Swearingen 226 Palma de Mallorca, Spain (2/0)-(0/0)-(2/0) 04/15 Air China Boeing 767-2J6ER Pusan, South Korea (11/117)-(0/0)-(11/166) 04/19 SELVA Colombia Antonov AN-32A Popayan, Colombia (0/3)-(0/0)-(3/5) 05/04 EAS Airlines BAC One-Eleven Kanos, Nigeria (7/64)-(0/0)-(8/69) 05/07 EgyptAir Boeing 737-566 Tunis, Tunisia (3/11)-(0/0)-(6/56) 05/07 China Northern McDonnell Doulgas Yellow Sea – Dalian, China (9/103)-(0/0)-(9/103) 05/21 Sky Executive Airlines Let 410UVP Calabar, Nigeria (5/0)-(0/0)-(5/0) 05/25 Trigana Air Service De Havilland Nabire, Indonesia (6/0)-(0/0)-(6/0) The red (bottom) line in this graph shows worldwide trends in aviation accident rates, as well as projected accident rates through the year 2010. The green (middle) line depicts the traffic growth which is expected to increase dramatically over the next 10 years. The blue (top) line shows the predicted increase in accident frequency due to the rapid industry expansion. Note that this predicted increase is based on the current accident rate; therefore, even if the accident rate stays the same over the next decade, the raw number of accidents will increase markedly. Furthermore, as can be seen from the graph, there may be as many as 52 accidents a year worldwide during the first decade of the new century. This translates into an astonishing one accident a week. Note. Graph adapted from Flight Safety Foundation (1997). Values plotted in the graph are estimates based on industry statistics.

2002年全世界失事紀錄 05/25 China Airlines Boeing 747-209B Taiwan Strait – Penghu, (19/206)-(0/0)-(19/206) 06/01 Airquarius Aviation Hawker George, South Africa (2/1)-(0/0)-(2/1) 06/17 Hawkins and Powers Lockheed Walker, California – U.S.A. (3/0)-(0/0)-(3/0) 07/01 DHL Express Boeing 757 Freighter Ueberlingen, Germany (2/0)-(0/0)-(2/0) 07/04 New Gomair Boeing 707 Bangui, Central African (16/7)-(1/1)-(17/8) 07/16 Britten-Norman BN-2B Borneo Jungles, (2/7)-(0/1)-(2/8) 07/17 Skyline Airways DeHavilland Surkhet, Nepal (2/2)-(0/0)-(2/2) 07/26 FedEx Boeing 727 Tallahassee, Florida, (0/0)-(3/0)-(3/0) 07/27 Ukraine Air Force Sukhoi SU-27 Lviv, Ukraine (0/0)-(2/0)-(2/0) 07/28 Pulkovo Airlines IL-86 Moscow, Russia (14/0)-(2/0)-(16/0) 08/22 Shangri La Air DHC-6-300 Pokhara, Nepal (3/15)-(0/0)-(3/15) 08/29 Vostok Aviakompania Antonov AN-28 Ayan, Russia (2/14)-(0/0)-(2/14) 08/30 Rico Linhas Aereas Embraer 120ER Rico Branco, Brazil (3/20)-(0/0)-(3/28) 09/14 TOTAL Linhas Aereas ATR-42-312 Paranapanema, Brazil (2/0)-(0/0)-(2/0) 10/01 India Military - Navy Ilyushin IL-38/ Vasco, India (12/0)-(0/0)-(12/0) 10/23 Tretyakovo Air Ilyushin IL-62M Bishkek, Kyrgyzstan (0/0)-(0/0)-(11/0) 10/25 Private Charter Beech King Air A100 Eveleth, Minnesota (2/6)-(0/0)-(2/6) 11/06 Lux Air Fokker 50 Luxembourgh (2/18)-(2/0)-(3/19) The red (bottom) line in this graph shows worldwide trends in aviation accident rates, as well as projected accident rates through the year 2010. The green (middle) line depicts the traffic growth which is expected to increase dramatically over the next 10 years. The blue (top) line shows the predicted increase in accident frequency due to the rapid industry expansion. Note that this predicted increase is based on the current accident rate; therefore, even if the accident rate stays the same over the next decade, the raw number of accidents will increase markedly. Furthermore, as can be seen from the graph, there may be as many as 52 accidents a year worldwide during the first decade of the new century. This translates into an astonishing one accident a week. Note. Graph adapted from Flight Safety Foundation (1997). Values plotted in the graph are estimates based on industry statistics.

2002年全世界失事紀錄 11/07 Dirgantara Air Service Britten-Norman - 2A Tarakan Indonesia (1/6)-(0/0)-(1/9) 11/08 Nepal Airways Harbin Yunshuji Y-12 Jomsom, Nepal (0/0)-(?)-(3/16) 11/09 Tymen Antonov AN-26 Antalya, Turkey (0/0)-(8)-(9/19) 11/11 Laoag International Fokker F27 Manila, Philippines (1/18)-(?)-(5/29) 11/28 Eagle Aviation Let 410 Masai Mara, Kenya (1/7)-(0/0)-(2/17) 12/03 C F F Air Learjet 36A Astoria, OR (0/0)-(0/0)-(2/2) 12/09 Raytheon Aircraft Beechcraft 1900C Eagleton, AR (2/1)-(0/0)-(2/1) 12/21 Transasia Airways ATR-72-202 Penghu Islands (2/0)-(0/0)-(2/0) 12/23 Aeromist Kharkiv Antonov 140 Baghrabad, Iran (6/38)-(0/0)-(6/38) 12/24 North Flying SA.227AC Metro III Aberdeen-Dyce Airport, UK (0/0)-(2/0)-(2/0) The red (bottom) line in this graph shows worldwide trends in aviation accident rates, as well as projected accident rates through the year 2010. The green (middle) line depicts the traffic growth which is expected to increase dramatically over the next 10 years. The blue (top) line shows the predicted increase in accident frequency due to the rapid industry expansion. Note that this predicted increase is based on the current accident rate; therefore, even if the accident rate stays the same over the next decade, the raw number of accidents will increase markedly. Furthermore, as can be seen from the graph, there may be as many as 52 accidents a year worldwide during the first decade of the new century. This translates into an astonishing one accident a week. Note. Graph adapted from Flight Safety Foundation (1997). Values plotted in the graph are estimates based on industry statistics.

What is the solution??

早期數據 第一次世界大戰時 英國皇家空軍飛行員陣亡的原因 敵人打落 飛機機械或結構失效 人員失誤

失事統計 Accidents by Primary Cause Hull Loss Accidents – Worldwide Commercial Jet Fleet – 1992 Through 2001

Organizational Factors 人為因素 Human Factors Ergonomics Aeromedical Issues Organizational Factors CRM Pilot Error A comprehensive Human Factors Analysis and Classification System (HFACS) has recently been developed to meet these needs. This system, which is based upon Reason’s (1990) model of latent and active failures (Shappell & Wiegmann, 1997a), encompasses all aspects of human error, including the conditions of operators and organizational failure.

Definition of Human Factors Human Factors discovers and applies information about human behavior, abilities, limitations, and other characteristics to the design of tools, machines, systems, tasks, jobs, and environments for productive, safe comfortable, and effective human use

人，與其所生活、工作、活動的環境間之互動，以及環境中各項事物對人類的影響 人為因素 人，與其所生活、工作、活動的環境間之互動，以及環境中各項事物對人類的影響

人類為什麼會犯錯？ 設計

人類為什麼會犯錯？ 工作環境

人類為什麼會犯錯？ 環境演變

為什麼會犯前面的錯誤？ 生理與心理的極限

人類對資訊的處理 認知與記憶

資訊的處理 訊息 資訊儲存 資訊的接收 資訊處理 與 決心下達 動作反應 結果

視覺特性

錯覺

方向性 人類視覺系統對方向性非常敏感

Display or Control Grouping

腦部活動的限制 注意力 記憶力

注意力 串聯而非並行 瓶頸 注意力超負荷 Load shedding Channelized attention 壓力與注意力

壓力與注意力 High Attention Low Low High Stress Level

壓力與表現 High Low Performance Efficiency Stress Level

請仔細觀察所播放的短片

發生了什麼事 ??

記憶力 不同之目的而有不同的記憶力系統 知覺的貯存（Sensory store） 短期記憶力（ Short term memory ） 長期記憶力（ Long term memory ）

知覺的貯存 訊息最初是被暫存於知覺上的貯存（ sensory store ） 兩種知覺上的貯存 影像的（ Iconic ）-- 眼睛所見 視覺上的訊息可被暫存約 0.5 至 1 秒 回聲的（ Echoic ）-- 耳朵所聽 聽覺上的訊息可被暫存約 2 至 8 秒

短期記憶力 特性 未不斷複習，很快就會忘記 記憶容量非常的小 Miller number

Short term memory practice

ASCNIKEVHSBMWIBMEVA ASC NIKE VHS BMW IBM EVA

長期記憶力 特性 三種貯存於長期記憶中的資訊 容量沒有限制 訊息可被永久貯存 不需控制腦部的活動來存取訊息 一般常識，對環境的瞭解 過去發生的事件 處理事物的知識

長期記憶力 根據經驗建立自我的心像模組（ mental models ） 預測事物未來走向 以心像模組中之相似情境來達到狀況警覺（ Pattern matched to elements in the mental model to achieve situation awareness ） 情境的認知可達到自動化（ Pattern-recognition sequence can become automaticity ） 有效的運用有限的注意力資源 自動化的情境認知可能對狀況警覺造成負面的影響

Human Factors Models

H S E L 人為因素模組 SHELL Model L : Liveware S : Software H : Hardware E : Environment

L SHELL Model Liveware Physical size and shape Physical needs Input characteristics Information processing Output characteristics Environmental tolerances

L SHELL Model Liveware - Liveware Leadership Crew cooperation Teamwork Personality interaction Staff/management relationship L

H L SHELL Model Liveware - Hardware Design of seats Design of displays Design of controls Equipment locations H L

S L SHELL Model Liveware -Software Procedures Manuals Checklists Symbology Computer programs S L

E L SHELL Model Liveware - Environment Adapting the human to the environment Helmets Flying suits Oxygen masks Anti-G suits Adapting the environment to match human requirements Pressurization Air-conditioning systems Soundproofing Everything affect human performance in the environment E L

人為因素模組 E H S L

Organizational Factors Reason Model Defenses Unsafe Acts Contributing Factors Organizational Factors Active Failures Latent Failures Latent Failures

飛航安全與人為因素調查

事故調查之目的 根據中華民國 飛航事故調查法 第五條 飛安會對於飛航事故之調查，旨在避免類似飛航事故之再發生，不以處分或追究責任為目的。

對於事件中發生的不安全行為或決定提出如何避免或降低的改善建議 調查程序 發覺人類的行為對事件可能造成的影響 找出環境中可能影響事故發生的因素 對於事件中發生的不安全行為或決定提出如何避免或降低的改善建議

事件發生後， 焦點通常在那？ Organizational Factors Unsafe Supervision Preconditions System failures are like dominos, with the failure of one “domino” effecting the toppling of the next. The end result is the accident or injury. When this happens, however, we often forget that the accident itself is the last “domino” in this sequence, and that many dominos fell well before the accident occurred. As a result, we tend to focus almost exclusively on the people responsible for front line operations (i.e., the aircrew). Unfortunately, this has lead accident pilots (if they survive the accident) to feel severely scrutinized, as if they are being placed under a microscope or interrogated for a crime. Unsafe Acts

失事案例 89年10月31日晚上11時17分 新加坡航空SQ006, B747-400 中正國際機場飛往洛杉磯

事件背景 起飛時誤入施工中跑道 飛機全毀、83人死亡

發生了什麼? WHAT? 發生原因 WHY? 如何避免再度發生??

為了避免類似事件再度發生，我們必需注意何處？ Organizational Failures Unsafe Supervision Preconditions Unsafe Acts Rather than scrutinizing the failure of a single system component, we must take a step back and look at the entire sequence of events that lead to the accident. A systems perspective requires that we examine blemishes or faults throughout the entire system. After all, it is often the failure of multiple components that combined together to produce an accident. Some people may raise the question, “Why stop at the organizational or even industry level?” Does the system’s boundary really end there? Presumably everything has a prior cause. Therefore, we could potentially trace the cause of an accident all the way back to the Big Bang. Stopping at the organizational level is just arbitrary. Theoretically this may be true. But we need to be practical. In seeking the reasons for an accident, we should search far enough back to identify factors that, if corrected, would render the system more tolerant to, or even prevent, subsequent encounters with conditions that produced the original accident. The people most concerned and best equipped to do this are those within the organization (Reason, 1990). Operating Environment

調查重點演變 操作者+ 操作者 操作者與機器介面 操作者+ 操作者與機器介面+ 操作者+ 操作者與操作環境+ 操作者與機器介面+ 組織與管理

失事案件討論

當時天氣 象神颱風 機場的天氣 中正機場南方360公里 最大風速75浬，陣風90浬 風向020，風速30浬，最大陣風61浬 能見度600公尺 雲高200呎 大雨

機場施工區

人為因素分析 為何三名駕駛員會將航機滑入05R跑道？ 為何會自05R跑道上起飛？ 分析 駕駛員和相關因素間的互動 駕駛員 – 駕駛員 駕駛員 – 航管 駕駛員 – 天氣 駕駛員 – 機場環境 訓練 – 駕駛員，航管人員 組織規定之程序及政策 民航監理單位之督導

駕駛員 正駕駛 CM-1 Male, age 41 副駕駛 CM-2 Male, age 36 加強組員 CM-3 Male, age 38 Total Flying Hours 11,235 hrs Total Command Hours on B747-400 2,017 hrs 副駕駛 CM-2 Male, age 36 Total Flying Hours 2,442 hrs Total Command Hours on B747-400 552 hrs  加強組員 CM-3 Male, age 38 Total Flying Hours 5,508 hrs Total Hours on B747-400 4,518 hrs  

跑道選擇 新航通常使用06跑道 正駕駛選擇05L跑道 CAT II 能見度限制較低 跑道較長 駕駛員二至三年未曾使用05L跑道

正駕駛的考量 正駕駛的主要考量是在於強烈的陣風和低 能見度 表示若超過限制就延後起飛 告訴自已要比平日更加小心 擔心天氣狀況會愈來愈糟

滑行 正駕駛跟 隨著綠色 的滑行道中心線燈 滑入05右跑道

機場配置

轉入跑道時 當正駕駛轉入跑道時 看到了跑道頭標線 沒注意到N1滑行道上直行的中心線燈 不記得看見任何跑道標誌或指示牌

轉入跑道時 當飛機自NP滑行道轉入 05R 正駕駛在滑行 副駕駛正在執行起飛前檢查表 第三組員在計算側風量

對正跑道時 正駕駛表示 看見跑道中心線燈 有八成把握看見跑道邊燈 副駕駛和第三組員表示 不記得看見跑道邊燈 沒看見任何跑道標誌及指示牌

起飛前 三位飛行員 都瞭解05R因施工而關閉，但可用於滑行 未看見任何施工警告標示 相信他們是在 05L

儀表顯示 駕駛艙內可供飛航組員參考的資訊 PVD PFD

PVD Para-Visual Display

PVD

PFD Indications Associated With Tuning the ILS Primary Flight Display PFD Indications Associated With Tuning the ILS Aligned with centerline Not Aligned with centerline Rising runway Localizer pointer and scale

機場設施 跑道關閉標誌 跑道警戒燈 跑道中心線標誌 N1滑行道中心線燈

Breakdown of a Productive System Inputs Typhoon Condition Unfamiliar With 05R Organizational Factors Latent Conditions PVD Procedure Taxi Procedure Unsafe Supervision Latent Conditions PVD Training Assess Runway Condition Preconditions for Unsafe Acts Active and Latent Conditions Discount PVD & PFD Loss of situational awareness As mentioned earlier, we have incorporated Reason’s (1990) model of how humans contribute to the breakdown of safe flight operations into our HFACS model. In this model, system failures are classified as either active or latent conditions. However, the exact nature of these failures or “holes” in the cheese have yet to identified and described. In this section, we provide a framework or taxonomy for identifying, classifying, and organizing active and latent failures with in the system. As previously stated, the framework is based upon the “The Taxonomy of Unsafe Operations” (Shappell & Wiegmann, 1997) which was developed for, and has recently been adopted by, the U.S. Navy/Marine Corps and U.S. Coast Guard for use in aviation accident investigation and analysis. The taxonomy describes four levels of failure within the system which include: (a) organizational factors, (b) unsafe supervisory practices, (c) unsafe conditions of operators, and (d) the unsafe acts operators commit. Each level is described in detail, beginning with the level most closely tied to the accident itself, unsafe acts. Unsafe Acts Active Conditions Right turn onto 05R Took Off From 05R Failed or Absent Defenses Accident & Injury Crashed on a Partially Closed Runway Adapted from Reason (1990)

調查結果 與可能肇因有關之調查結果 與風險有關之調查結果 其它調查結果 此類調查結果係屬已經顯示或幾乎可以確定為與本次事故發生有關之重要因素。其中包括：不安全作為、不安全狀況或造成本次事故之安全缺失等。 與風險有關之調查結果 此類調查結果係涉及飛航安全之風險因素，包括未直接導致本次事故發生之不安全作為、不安全條件及組織和整體性之安全缺失等，以及雖與本次事故無直接關連但對促進飛安有益之事項。 其它調查結果 此類調查結果係屬具有促進飛航安全、解決爭議或澄清疑慮之作用者。其中部份調查結果為大眾所關切，且見於國際調查報告之標準格式中，以作為資料分享、安全警示、教育及改善飛航安全之用。

Not to Blame But to Prevent