Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Airplane Accidents and Fires Thomas Me Murray* INTRODUCTION Airplane fires prompt fear and concern for anyone associated with air transport. Those associated include He passengers and flight and cabin crews, as well as those who must eventually provide rescue functions and extinguish an airplane fire. Fortunately, the aviation community experiences few fires while in flight. Furthermore, those fires that do occur on the ground are typically preceded by another event: a hard landing ruptures a fuel tank; a long landing prompts a runway overrun; a premature landing causes excessive airplane damage; a controlled flight into terrain shatters an airplane upon impact; etc. Due to the frequency of and often repeated- preceding events, Boeing believes that the most effective safety strategy available today is to PREVENT ACCIDENTS. Many airplane accidents are preventable using current technology and preferred cockpit procedures. For instance, controlled-flight-into-terra~n accidents are preventable through technology, the ground proximity warning systems, and an appropriate response to the warning. _ . . . . , , _ . · From reviewing multiple accident records, Boeing as well as others in the commercial airplane transportation industry have determined that rarely is an accident the result of only one event. Rather, most, if not all, accidents occur through a chain of events. For instance, the hard landing is often preceded by an unstabilized approach; the unstabilized approach may occur due to air traffic control instructions, weather, or other traffic; or cockpit procedures developed by the airline may be inadequate for some circumstances. By focusing on these chains of events for accidents and developing accident prevention strategies applicable to venous links in the chain, Boeing and members of the commercial airplane transportation industry have found, and continue to find, multiple opportunities for interrup~ir~g the chain of events leading to an accident (see Figure I). Fire certainly fits within the realm of a multiple-cause process. The only place on the airplane where fire is planned is within the combustion chambers of the engine. Otherwise, any other fire event on the airplane is unplanned and requires at least one "event or cause" to precede the fire. The information in this review of aviation accident data with associated fires is broken into three parts. First, there is an examination of the commercial jet transport accidents from 1959 through 1993. This information is provided to offer a perspective regarding aviation accidents from a variety of causes. Then, the second part reviews how fire has influenced the accident record. This data is presented in a format similar to the overall commercial jet transport statistics to help individuals compare the frequency of f~re-associated events with the variety of other causes of accidents. Fire details are also provided to enhance understanding of airplane fire-related accidents. Finally, the third part examines some specific fire scenarios to illustrate *Airplane Safety Engineering, Boeing Commercial Airplane Group, Seattle, Washington. 7
8 Improved Fire- arm Smoke-Resistant Materials the range of events that occur, since looking at the number of events often fails to illuminate the many parameters associated with an aspect of aviation accident data, such as fire-related events. Moreover, these specific fire scenarios will also highlight the mul~ciple-cause process and examine possible intervention opportunities where the outcome of the event might have changed considerably through alternative actions. In addition, these specific events have resulted in a range of outcomes despite some similar scenarios. Prevent accidents "Remove the links in the accident chain" Con ~ ~ Hull loss accidents Anr~u.J rates, accidents p r mililon at departures Conclu~lon-The accident rate has not decilned significantly In the past 20 years FIGURE 1 Accident chain concept. Paradigm shift Former view focused on the single probable cause of an accident · Current view: Examines the entire chain of events leading to an accident Promotes multiple intervention strategies to remove the links in the accident chain COMMERCIAL JET TRANSPORT ACCIDENT STATISTICS Background for Accident Data-CoHection Processes Boeing has collected accident data since 1959 for commercial jet operations with certified jet aircraft greater than 60,000 Ib maximum gross weight. This data-collection effort has included not only Boeing airplane models but also data for a variety of competitor airplanes. Boeing believes that it must improve airplane safety using all available data sources; therefore, all catastrophic airplane accidents are reviewed.
Thomas M. Murrary 9 For the accident information presented in this review, there have been some data exclusions. First, data for sabotage, military action/operation, and non-operational events are excluded. Non-operational events include those where the airplane is not in service; the airplane may be in a hangar for maintenance, on a ramp for overnight storage, etc. Relevant data includes passenger operations; all-cargo operations; and test, training, demonstration, and ferry flights. Furthermore, worldwide flight operations are included. However, accident data for airplane models from the counties of the former Soviet Union are excluded due to the lack of a complete accident record as well as a different and not well understood airplane certification process as compared with Federal Aviation Administration (FAA), Civil Aviation Authority, or loins Aviation Authority certification processes. To determine the flight time and cycles, data is collected by both the aircraft manufacturers and the engine companies. This data is gathered from operators' aircraft and engine logs. Accident data are obtained, when available, from government accident reports. In addition, information is solicited from operators, manufacturers (both for the completed airplane product as well as for various systems and components), and various government ant} private information services. The accident selection criteria essentially corresponds to the U.S. National Transportation Safety Board's accident definition. However, events are excluded that involve nonfatal injuries resulting from maneuvering, atmospheric turbulence, loose objects, boarding or disembarking, or airplane servicing activities. To fully understand the accident data that Boeing collects, it is important that the following definitions be relayed: Aircraft accident means an occurrence associated with the operation of an aircraft that takes place between the time any person boards the aircraft with the intention of flight until such time as all such persons have disembarked, in which any person suffers death or serious injury as a result of being in or upon the aircraft or by direct contact with the aircraft or anything attached thereto, or the aircraft receives substantial damage. Serious injury means any injury that (~) requires hospitalization for more than 48 hours, commencing within 7 days from the date of the injury received; (2) results in a fracture of any bone (except simple fracture of fingers, toes, or nose); (3) involves lacerations that cause severe hemorrhages or nerve, muscle, or tendon damage; (4) involves injury to any internal organ; or (5) involves second- or third- degree burns affecting more than 5 percent of the body surface. Fatal injury is defined as an injury that results in death within 30 clays of the accident. · A hull-Ioss means damage due to an accident that is too extensive to repair, or that, for economic reasons, was not repaired, so that the aircraft was not returned to service. A survivable accident is one in which the fuselage remains relatively intact, the crash forces do not exceed the limits of human tolerance, there are adequate occupant restraints, and there are sufficient escape provisions.
10 Improved Fire- arm Smoke-Resistant Materials Boeing has accident data for 9 airplane manufacturers, which include data on 28 different mode} types. As of the end of 1993, there were ~ 1,433 aircraft in service; 6,465 of these aircraft were Boeing models. From 1959 through 1993, the commercial jet transport industry had accumulated 406 million flight hours (237 million on Boeing aircraft) and 270 million departures (149 million on Boeing aircraft). Figure 2 provides visual information for both the number of airplanes in service and the annual departures since 1964. Aviation Accident Data for HuN Losses and Fatalities Hull-Loss Data The aviation accident record, as measured by the number of hull losses each year, suggests that the aviation industry possesses an enviable safely record compared with other transportation modes. Overall, the worldwide commercial jet fleet has a hull-Ioss accident rate of approximately two accidents per million departures. Figure 3 displays the worldwide commercial jet fleet hull-Ioss accident rate per million departures for all airplane models. The to~ number of hull losses from 1959 through 1993 was 512. From 1984 through 1993, 171 hull losses occurred. Further specifics regarding the hull losses can be seen in Table I. 1 2,000 1 0,000 Aircraft 8,000 6,000 4,000 2,000 14 12 10 Annual departures 8 (Millions) 6 A 1 1 ,433 - - - o 65 67 69 71 73 Is 77 79 81 83 85 87 89 91 93 13.86 : 65 67 69 71 73 75 FIGURE 2 Jet aircraft in service and annual departures. 77 79 81 83 85 87 89 91 93
Thomas M. Murray 50 40 30 Annual rates accidents per million departures 20 10 o \ 59 61 63 65 67 69 71 73 75 77 Year FIGURE 3 Hull-loss accidents (excludes sabotage and military action). TABLE 1 Hull-Loss Accidents 11 79 81 83 85 87 89 91 93 1959-1993 1984-1993 512 hull losses U.S. operators · 97 during passenger operations · 27 during all cargo operations · 16 during test, training, demonstration, or ferry Non-U.S. operators · 295 during passenger operations · 43 during all cargo operations · 34 during test, training, demonstration, or ferry 171 hull losses U.S. operators · 23 during passenger operations · 8 during all Argo operations · 3 during test, training, demonstration, or ferry Non-U.S. operators · 103 during passenger operations · 24 during all-cargo operations · 10 during test, training, demonstration, or ferry . From the hullL-Ioss accidents chart (Figure 3), it appears that the commercial airplane transportation industry has significantly reduced the accident rate from its initially high level to a relatively low rate. However, despite the low accident rate, THE COMMERCIAL AIRPLANE TRANSPORTATION INDUSTRY MUST CONTINUE TO IMPROVE. Indeed, the Boeing goal, and now FAA-industry goal, is zero accidents. Consequently, Boeing has promoted the concept of "preventing accidents" as noted in the introduction.
12 Improved Fire- arm Smoke-Resistant Materials Figure 4 shows why the current hull-Ioss accident rate is unacceptable. If the aircraft fleet continues to expand (and there are a number of reasons for expansion to continue: increased traffic in Asia, South America, arid counties of the former Soviet Union; increased global business processes; increased leisure travel; etc.), then it is possible that the aviation industry could experience a hull-Ioss every week by the year 2010. Clearly, this possibility is unacceptable for the aviation industry. 40 35 30 25 Annual hull 20 losses 15 10 5 o 2526 - 121 1 22 ~1 1 1515 14 20, l 1 990 21 ,? .. ~ ~ .. a ~ ~ In. w~.~, ~. ~ Projected hulilosses per year 1 995 Year 1 985 2000 200s 2010 FIGURE 4 Projected hull-loss accidents based on accident rates for past 10 years and expected fleet growth. Furthermore, some recent trends, as shown by a 5-year moving average of the hull-loss accident rate, suggest that those in the aviation industry must remain vigilant in the pursuit of zero accidents due to an apparent increasing hull-loss accident trend. (See Figure 5.) From Figure 6, it appears that most of the newer-generation commercial jet transports are contributing favorably to the lower hull-loss accident rate. However, the overall accident rate is still I.9 hull-loss accidents per million departures. This airplane generation assessment is further detailed by Figure 7. In this figure the hull-loss rates for various airplane generations, except for those airplanes from the first generation of commercial jet transports, are noted. The aviation industry is not without guides or direction as to the types of accidents that occur and possible remedies. As Figure ~ shows, there are specific phases of flight where the majority of accidents occur. For instance, final approach and landing phases of flight account for more than 40 percent of the hull-loss accidents. On the other hand, these two flight phases represent only 4 percent of the total flight time. Due to this data and other indicators used by the airlines, many airlines have instituted specific final approach criteria. For instance, Boeing
Thomas M. Murray 4 3.5 3 2.5 Annual rates, accidents per million 2- departures 1.5 1 0.5 ~L ~5 ear movie avera e: a/1 1 T K'' ~ . J _ _ 85 / T 83 7:T 7: ! 7g 81 o 76 78 80 82 84 8 7 8 3 91 93 86 88 90 92 YEAR 13 FIGURE 5 Five-year moving average of hull-loss accident rates (excludes sabotage, military action, and former Soviet Union airplane models). Airplane Tvpe ~9.631 707D72O8 ~9.55 8801990 = 5.96 Tvic7e21n70 ~ 3.87 1 i_3.31 1 55.00 BAC 111 _ .1.18 _2,37 737-100/200 ~ 3|82 747-100/200/300 i= 1.11 1, MD80 =~0.98 767 oo.29 BAG 146 _0.6 0.95 737 300/400F500 1.17 A320/A321 2 50 7474008 ~1~86 MA0~o 0 00 A330 0.00 Overall Rate 1.90 0 1 2 3 4 5 6 7 8 9 10 Accidents per million departures FIGURE 6 Hull-loss accident rates per generation of aircraft (excludes sabotage and military action).
14 Improved Fire- arm Smoke-Resistant Materials 20 15 Annual rates, accidents 1 0 per million departures o Second Generation .~ . Second generation .......... 727 Trident VC-10 BAC 111 DC-9 737-1 00/200 F28 DC-10 L-1 011 A300 , W i ~ /, Ah/ 59 61 63 65 67 69 71 73 75 77 79 81 83 85 Year , Wide body New (early) types 747-1 00/200/300 &1D80 MD1 1 737-300/400/500 747-400 757 767 A310 A320/A321 A330 A340 BAe 146 F100 \ ~ . New Types _____ j\ ~l ~ 87 89 91 93 FIGURE 7 Hull-loss accident rates by generic group of aircraft (excludes sabotage and military action). Exposure percentage based on an average flight duration of 1.6 hours Percentage of accidents Load, | Takeoff taxi | 14.3% unload 2.0% "" 1 54.5% ~. ~ Initial climb 1 0.5° Cruise 4.5°/O Climb Cruise Descent Initial Final | Landing 7.0% 4.5% 7.2% approach approach 18.4% 11 .5% 24.6%1 / Flaps retracted n \ ~ /, = . _ ~ \~ Nav Outer fix marker 1% 1 1% 1 13% 1 60% 1 10% 1 11% 1 3%1 1% Exposure, percentage of flight time FIGURE 8 Hull-loss accident rates for specific phases of flight (excludes sabotage and military action). recommends that a stabilized approach be achieved by 1,000 feet. Most airlines require a stabilized approach by at least 500 feet. If the airplane is not stabilized by 500 feet, most airlines that require the airplane to circle for another approach go-around. One airline that uses the ~ ,000-feet stabilized approach guideline as a requirement has never had an approach accident.
Thomas M. Murray 15 Take-off and initial climb also represent an opportunity for improvement. Some 25 percent of the hull-Ioss accidents occur during this phase of flight. Some actions that have occurred include reinforcement of appropriate procedures for refused take-off decisions. For instance, Boeing airplanes purposely disallow some messages from getting to the flight deck during the take-off phase to prevent the pilots from being distracted by airplane functions that do not affect the flying capability of the airplane. As shown in Figure 9, the non-U.S. operators still have a higher accident rate than the U.S. operators. 30 20 Annual rates, accidents per million departures 10 o T:~ 'at 'at Non-U.S operators L . ,,, , ,. 81 83 85 87 89 91 93 U.S. operators ~ . ,,# ,, _ _ _ _ 59 61 63 65 67 69 71 73 75 77 79 Year FIGURE 9 Hull-loss accident rates, U.S. and non-U.S. (excludes sabotage and military action). Fatality Data Hull-Ioss data is not the only measure of airplane safety processes. Another measure used by Boeing and many airlines is the number of fad accidents that occur. This measure is useful, since many hull-Ioss accidents do not result in fatalities. Table 2 shows a breakdown of the 398 fad accidents that occurred from 1959 through 1993. Figure 10 shows that the fatality rate has declined from its initial high rate. However, the number of fatuities per year does not show a similar decline. As shown in Figure ~ I, there are particular events that have prompted the most fatalities. Controlled flight into terrain (CFIT) remains a dominant cause of fatuities despite the
16 TABLE 2 Brealcdown of Fatal Accidents Improved Fire- aru] Smoke-Resistant Materials 1959-1993 1984-1993 398 fatal accidents Passenger operation · 319 fatal accidents · 18,956 fatalities* All~argo operation · 44 fatal accidents · 174 fatalities* Test, training, demonstration, and positioning · 35 fatal accidents · 168 fatalities* 120 fatal accidents Passenger operation · 96 fatal accidents · 5,397 fatalities* All cargo operations · 18 fatal accidents · 76 fatalities* Test, training, demonstration, and positioning · 6 fatal accidents · 53 fatalities* *Onboard fatalities only. introduction of Ground Proximity Warning Systems, improved crew procedures, and continuous · ~ training. Despite the low accident rates, any fatalities are unacceptable to all aviation participants. The aviation industry must continue to find ways of interrupting the accident chain to PREVENT ACCIDENTS. COMMERCIAL JET TRANSPORT DRE STATISTICS Recall that Figure 3 shows the worldwide hull-Ioss accident rate. Similar information for fire-related hull-Ioss accidents follows. The fire-related hull-Ioss accidents were divided into two categories: those hull-Ioss events where the fire was the primary cause of the hull-Ioss and those events where some other event preceded the fire, thus fire was a secondary event. "Primary f~re- related hull-Ioss events" include cargo compartment fires, lavatory fires, fuel tank fires, engine fires, engine burst, etc. For primary fire events, the events were screened with the following criteria: "If the fire had not occurred, would a hull-Ioss not have resulted?" If the answer to this question was yes, then fire was the primary cause. (Engine burst was included in the primary category despite the screening criteria, since flight crews have minimal control of the engine burst phenomenon.) "Secondary fires" include such events as a hard landing where the landing gear subsequently punctures the fuel tank, CFIT with resulting fire, refused take-off events where the wings are sheared after the airplane goes off the end of the runway, etc. Secondary events include all the events not placed into the primary category. Given these two categories, the f~re-initiated hull-Ioss accident rate is 0. ~ per million departures. The non-fire-initiated hull-Ioss accident with fire rate is 0.7 per million departures. Figure 12 displays this data along with the overall accident rate. The f~re-caused hull-Ioss accident rate, or primary category, is plotted in Figure 13 for 1959 through 1993.
Thomas M. Murray 20 Annual rates Accidents 1 0 per million departures o 15 Annual 1 0 fatalities,** hundreds 5 o 17 _ __ ~ ~ ~: . _ L/ \/\ . ~ _ ___Wi _/ ~ ~ ~ l 1 ** Onboard fatalities only _ _ 1 , ~ T 59 61 63 65 67 69 71 73 75 77 79 81 Year 83 85 87 89 91 93 FIGURE 10 Fatal accidents, annual rates, and annual fatalities (excludes sabotage and military actions). 2,000 1 ,500 Fatalities 1 ,000 500 o Number of fatal accidents (76 total) 28 r 1 88:` ,~ ~306 3,513 total fatalities 1 993 ~ ~4 357 ,57 278 157 ~ 43 15 Loss of Loss of Airframe Mid-Air Ice/snow Fuel Loss of Runway Other control Control collision exhaustion control incursion (airplane (crew) (weather) caused) 10 14 4 1 4 FIGURE 11 Worldwide airline fatalities (excludes sabotage and military action). 2 3 5
18 FIGURE 12 Overall hull-loss accident rates and fire. 4.5 3.5 3 Annual rates, accidents 2 5 per mililon departures 2 1.5 1 0.5 o FIGURE 13 Fire caused hull-loss accident rates. Improved Fire- arm Smoke-Resistant Materials Overall Hull Loss Accident Rate 1.9 Accidents per million departures ... - .~..?. '.. - . .~._ ~ · Fire Initiated Hull Loss Accident Rate per million departures · Non-Fire Initiated Hull Loss with Fire R ~, - , i# ~- . ~Accident Rate , ....................................................... ; H.ar.d...l~a.ndiri~ it: : ................ . . ~. ~ . ~ '.. Staiiii~g.s of Control ~ ................................................................................... ~ .................. .............. . .. a : CoII.iSton . ~ ............................................. ..... . hi' ~ _ E ~ ~ .7 per million departures - / - /1 1 1 ~ I i 1 t 1~ 1 1 ~1 1 ~ I I I I I I I I I ~ Q _ CatA" ~ A ~ Al ~ O - Cod Cal) ~ if) ~0 1~ a) Ch O - N All ~ ~ `0 ~ ~ 0~ 0 - Call CO try ~ ~ 0% 0~0% 0~ 0~ ~ ~ 0~ 0~ 0~ ~ Cat ~ ~ 0~ ~ ~ Ch _ __ ___ ___________ _________ ______ ___ Year Of all of the hull-loss accidents with fire, only ~ ~ percent of these events were initiated by fire or in the "primary" category. Indeed, the remaining 89 percent of these hull-loss accidents with fire were preceded by some other significant event, as shown in Figure 14 with details on the contributors to both the fire-initiated and non-fire-initiated hull-loss accidents.
03 ~ ~- - ~ ~- . CD. 00 IS - ~ ~- . - ~ ~- ~ 1 c] of CD A A 93 o ~- . CD =e ~ . 3 3 o ~ ~ o 'A ~ i o P) lo. =. ~ ~ ~ ~ L~.e B. m ~ it ~ ~ ~ ~-· 5" ~ · ~ g o (D to ~ I-' ~ {D ~ (p =`
20 Annual Fatatlies, 3 hundreds 2 FIGURE 16 Annual fire-smoke~aused fatalities. Improved Fire- arm Smoke-Resistant Materials ~ T 5 4 ^L~ ~IVI I IV'I I I lVIVi l\ ldlL - 5; ~ ~ ~ $ ,; ~ :8 7; $ $ ° ~ '~ ~ ~ )Q i~ ~ ~ ~ ~ a' '~ ~ ~ ~' ~ ~ 0 8; ~ ~ ~ 8: ~ ~ ~ ~ on on ~ on on on ~ on ~ ~ on __________ __________________ Year SELECTED lilRE SCENARIOS Although the statistics provide very detailed information regarding fire events, some sample scenarios also reveal the diverse nature of the fire event. Therefore, summaries of four different hull-Ioss accidents where fire occurred are provided. Event sequences are provided for each accident to help readers focus on the chain of events preceding the accident. The key message from these example accidents is that most, if not all, airplane accidents are a chain of events. Therefore, these chains contain multiple opportunities to interrupt the accident scenario. Example: 727, Salt Lake City The airplane had originally departed New York and landed in Denver (see Figure 17~. The captain had flown that segment. The first officer was flying the Denver to Sail Lake City segment. The airplane arrived within the Salt Intake approach area south of Provo, Utah. Although it was dark, it was typical Utah weather: VFR (visual flight rules) to Il,000 feet Normally, the captain preferred to descend prior to Provo, but that night the descent was delayed momentarily. Eventually, the first officer started the descent. To meet the threshold appropriately, the descent was motivated such that the airplane showed descent rates greater than 2,000 feet/minute. Company policy was to descend at 600 to 800 feet/minute. Prior to the threshold, the first officer moved his hand to the throttles to arrest the descent. The captain restrained the first officer. The flight engineer reported later that he, too, felt that the descent needed to be arrested. The descent continued. The airplane landed 335 feet short of the runway with a descent rate of 2,300 feet/minute. Despite the hard impact, the airplane remained
Thomas M. Murray 21 relatively intact. Although He impact injuries were not excessive, many individuals were killed by fire-related agents: smoke, fire, and lack of oxygen. Again, there were multiple opportunities for altering the chain of events. First, the air traffic control center needed greater understanding of the preferred airplane descent profiles. Although airplanes can lose altitude very quickly, accelerated descents not only affect safety but also affect airplane engine life. Next, the cockpit communications environment needed to utilize the capability and recognize the concerns of all the flight crew members. One the outcomes of this accident was to develop and promote the concept of cockpit resource management: all flight crew members offer input while still allowing the captain to maintain his or her role as final decision maker. Accident profile · November ~ 1, 1965 · 727 mode} · Salt Lake City Chain of events 1 . . · Fatalities: 43 · 85 passengers · 6 crew First officer flying 1 - -- ---- - - 1-- ----- Captain holds throttle · 1 Departs Arrive for Airplane Denver SLC approach lands short · Change ~ Procedures training for flighicrews FIGURE 17 Lessons learned, 727 Salt Bake City accident. NOTE: Timelmes not to scale. Example: DC-9, Cincinnati The airplane had departed Dallas and was enroute to Toronto (see Figure 181. At 1851 the lavatory breakers popped on the flight deck. The flight crew cried multiple times to reset the breakers. There was no communication between the flight crew and the cabin crew. At 1900, there was a strong odor noticed in the cabin by both the passengers and the cabin crew. Within a minute or so, the cabin crew recognized that there was a fire in the aft lavatory on the left side of the airplane. Multiple CO2 fire extinguishers were deployed to arrest the fire in the lavatory. At 1902, the first officer attempted to view the aft lavatory, but was stopped by smoke midway through the airplane. By 1904:07, the first officer returned to the flight deck and informed the captain of the situation. Within seconds, the cabin appeared clear. By 1905, the captain asked the first officer to return to the aft lavatory using the captain's smoke goggles to get a first-hand appraisal of the situation. Within 30 seconds, the airplane electrical system showed anomalies
22 Imp roved Fire- arut Smoke-Resistant Materials to the captain. The first officer returned to the cockpit where the captain was attempting to diagnose the electrical system problems. By 1907:41, the master warning light illuminated. At 1908: 12, the captain declared an emergency descent and asked for the nearest airport. During the venous attempts by the flight crew to understand the extent of the fire in the aft lavatory, the airplane had passed Standiford Field where the airplane could have landed. The airplane arrived on the ground at 1920:09 at Cincinnati International Airport. After this event, at least two specific changes and procedure reinforcements were made to respond to airplane fire events. First, all lavatories have smoke detection systems to alert crews of problems. Second, for any event where fire is suspected, flight and cabin crews must communicate immediately, and the flight crews will seek the nearest airport. · Accident profile · June 2, 1983 · DC-9 · Cincinnati International Airport · Chain of events Strong odor in cabin Fatalities: 23 · 41 passengers ~ 5 crew Electrical Master Flightcrew system warning informed abnormalities light 1 1 11 . I I I 1 1 1 rid 11 1 1 1 7 1 Lavatory Flight attendants Cabin Electrical Descent breakers pop recognize fire appears problem declared clear diagnosis · Changes · Smoke and fire detection system · New fire procedures FIGURE 18 Lessons learned, DC-9 Cincinnati event. NOTE: Timelines not to scale. Example: 737-200, Calgary and Manchester The Calgary and Manchester events involving 737-200s were very similar; however, their outcomes were remarkably different (see Figure 19~. In both events, the engine experienced an unconfined engine burst phenomenon. The wing fuel tanks were impacted. Fuel then spilled from the wing and ignited. The pilots, in both cases, were unaware of the extent of the fire. After becoming aware of the fire, the pilots chose to continue the taxi to exit the runway. Emergency evacuation was then initiated. However, the difference between the two accidents was the direction of the wind as related to the final stop for the airplane. For the Calgary case, the airplane was almost lined up with the wind. Thus, the fire did not penetrate the fuselage as quickly as in the Manchester case. Some of the industry responses to these events included engine improvements (blade changes because of the Calgary accident and burner can improvements because of the
Thomas M. Murray 23 Manchester accident); fuel door improvements; reinforcement of known procedures for communications between the flight and cabin crews and for information flow from the tower to the flight crew; arid reinforcement of current procedures by flight crews for fire events. One of the procedures that pilots were reminded of was the need to stop the airplane immediately. This action prompts airplane occupants to leave the airplane faster, promotes fuel runoff away from the airplane when fuel spills occur, and favors the wind direction. · Accident profiles · March 22, 1984 · 737-200 · Calgary, Alberta · Fatalities: None · ~ 14 passengers · 5 crew · Chain of events Calgary~ Takeoff Loud initiated bang Manchester · August 22, 1985 · 737-200 · Manchester International Airport · Fatalities: 55 · 131 passengers · 6 crew Left engine parameters Tower confirms Emergency evacuation fire, 62 sec initiated, 1 min 55 sec Flight attendant identifies fire, 45 sec Left engine fire warning Taxi Passengers continues escape Airplane Emergency taxis evacuation Takeoff Thud initiated Tower confirms Airplane parked Multiple fire in crosswind fatalities FIGURE 19 Lessons learned, 737 Calgary and Manchester events. NOTE: Timelines not to scale. SUMMARY Airplane fires, whether in-flight or on the ground, prompt not only apprehension in the flying public but also can cause major devastation, both to life and property. However, despite the enormity of the event, the commercial jet transport industry experiences such events very infrequency. Furthermore, most fire-related events are preceded by events where intervention at multiple points might prevent the accident. Therefore, the commercial jet transport industry must retain its vigilance for preventing airplane accidents by examining accident event chains, understanding the various links in the chain, and then attacking multiple links to prevent the accidents.
