NATIONAL

8 PB2000-910403 NTSB/AAR-00/03 DCA96MA070

NATIONAL

TRANSPORTATION SAFETY BOARD

WASHINGTON, D.C. 20594

AIRCRAFT ACCIDENT REPORT

In-flight Breakup Over the Atlantic Ocean Trans World Airlines Flight 800 Boeing 747-131, N93119 Near East Moriches, New York July 17, 1996

〔图〕●●\n 〔图〕插图 6788G

Aircraft Accident Report

In-flight Breakup Over The Atlantic Ocean Trans World Airlines Flight 800 Boeing 747-131, N93119 Near East Moriches, New York July 17, 1996

NTSB/AAR-00/03

DCA96MA070

PB2000-910403

Notation 6788G

Adopted August 23, 2000

〔图〕8OALVs\n National Transportation Safety Board

490 L'Enfant Plaza, S.W.

Washington, D.C. 20594

iii

Aircraft Accident Report

Contents

Abbreviations $\ldots$ xi

Executive Summary $\ldots$ xvi

1. Factual Information .................................................................................................... 1

1. 1 History of Flight. $\ldots$ 1

1. 2 Injuries to Persons $\ldots$ 4

1. 3 Damage to Aircraft .................................................................................................................................... 4

1. 4 Other Damage $\ldots$ 4

1. 5 Personnel Information $\ldots$ 4

1. 5.1 The Captain (Left Front Seat) .................................................................................................................................... 4

1. 5.2 The Captain/Check Airman (Right Front Seat) 5

1. 5.3 The Flight Engineer (Right Aft/Flight Engineer's Seat) .................................... 5

1. 5.4 The Flight Engineer/Check Airman (Left Aft/Cockpit Jump Seat) 5

1. 6 Airplane Information $\ldots$ 6

1. 1 Boeing 747—General Description/Information $\ldots$ 7

1. 6.1.1 747-100 Wing Center Section and Center Wing Fuel Tank Description 12

1. 6.1.2 747-100 Air Conditioning Equipment Description $\ldots$ 16

1. 6.1.3 747-100 Electrical and Wiring Information 21

1. 6.1.3.1 747-100 Electrical Information. 21

1. 6.1.3.2 747 Wiring Information—General and Specific to Accident Airplane . . . . 22

1. 6.1.3.3 Separation of 747 Wire Circuits $\ldots$ 25

1. 6.2 747-100 Fuel System Description $\ldots$ 26

1. 6.2.1 Fuel Quantity Indication System Components and Wiring Information. 31

1. 6.2.2 747-100 Fuel Pump System Description. $\ldots$ 37

1. 6.2.3 747-100 Fuel Flow Indication Information $\ldots$ 42

1. 6.2.4 TWA Flight 800 Fueling Information $\ldots$ 42

1. 6.3 Maintenance Information $\ldots$ 43

1. 6.3.1 TWA's 747 Maintenance Inspection Program $\ldots$ 43

1. 6.3.1.1 TWA's General Inspection Policies $\ldots$ 45

1. 6.3.1.2 TWA and Boeing Wiring Inspection Guidance ... 46

1. 6.3.2 Accident Airplane's Maintenance Information. $\ldots$ 46

1. 6.3.2.1 Airworthiness Directives and Service Bulletins Applicable to 747 Fuel Pumps and Related Wiring 47

1. 6.3.2.2 Airworthiness Directives and Service Bulletins Applicable to 747 Structural Inspections $\ldots$ 48

1. 6.3.2.3 Maintenance Accomplished Near Fuel Quantity Indication System Wiring in the Accident Airplane. $\ldots$ 50

1. 6.3.2.4 TWA Flight 800 Predeparture Maintenance Information 54

1. 7 Meteorological Information $\ldots$ 56

1. 8 Aids to Navigation $\ldots$ 57

1. 9 Communications .................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................

1. 10 Airport Information $\ldots$ 58

1. 11 Flight Recorders. 58

1. 11.1 Cockpit Voice Recorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

1. 11.1.1 Sound Spectrum Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

1. 11.1.2 Cockpit Voice Recorder-Related Airplane Tests . . . . . . . . . . . . . . . . . . . . . . . . 60

1. 11.2 Flight Data Recorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

1. 12 Wreckage Recovery and Documentation Information . . . . . . . . . . . . . . . . . . . . . . . . . 62

1. 12.1 Wreckage Recovered from the Red Zone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

1. 12.2 Wreckage Recovered from the Yellow Zone. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

1. 12.3 Wreckage Recovered from the Green Zone. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

1. 12.4 Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

1. 12.5 Center Wing Fuel Tank Pumps and Other Components . . . . . . . . . . . . . . . . . . . . . 75

1. 12.6 Air Conditioning Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

1. 12.6.1 Brown Splatter Material on Air Conditioning Ducts . . . . . . . . . . . . . . . . . . . . . 78

1. 12.7 Recovered Electrical Components/Wiring. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

1. 13 Medical and Pathological Information. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

1. 14 Fire/Explosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

1. 15

References

Abstract

Sequence of eight primary radar returns recorded by the Islip, New York, radar site about the time of the accident .

References

x

Abbreviations

xii

Aircraft Accident Report

CDG Charles DeGaulle International Airport (Paris, France)

CDL Combustion Dynamics Limited

CEU central electronics unit

CFR Code of Federal Regulations

c.g. center of gravity

CHDO Certificate Holding District Office

CIA Central Intelligence Agency

CIT California Institute of Technology

CMR Christian Michelsen Research

CVRCockpit voice recorder

CWT center wing fuel tank

d.c. direct current

DERA Defense Evaluation and Research Agency (British)

DGAC Direction Generale de L'Aviation Civile

DNA deoxyribonucleic acid

DoD Department of Defense

DRI Desert Research Institute

EDS energy dispersive spectroscopy

E/Electrical/electronics

EME electromagnetic environment

EMI electromagnetic interference

EPR engine pressure ratio

F Fahrenheit

FAA Federal Aviation Administration

FARs Federal Aviation Regulations

FBI Federal Bureau of Investigation

FDR flight data recorder

FMEA failure modes and effects analysis

FOK Francis S. Gabreski Airport (Westhampton Beach, New York)

fps feet per second

FQIS fuel quantity indication system

FR Federal Register

FTHWG Fuel Tank Harmonization Working Group

Abbreviation

xiii

Aircraft Accident Report

GPS global positioning system

HF high frequency

Hg mercury

HIRF high-intensity radiated fields

HP High Performance

HPC high-pressure compressor

HSSM Hamilton Standard Service Manual

Hz hertz (cycles per second)

IAM International Association of Machinists, Aerospace Workers, and Flight Attendants

IOE initial operating experience

J joule

JAA Joint Aviation Authorities

JFK John F. Kennedy International Airport (New York, New York)

JSC Joint Spectrum Center (DoD)

kJ kilojoule

kW kilowatt

LaRC Langley Research Center

LBL left butt line

LFL lower flammability limit

LLS laser line-scanning

LPC low-pressure compressor

MAC mean aerodynamic chord

MHz megahertz

MIE minimum ignition energy

mil one-thousandth of an inch

mJ millijoule

mm millimeter

MO modification order

MSFC George C. Marshall Space Flight Center

MSIC Missile and Space Intelligence Center(Defense Intelligence Agency)

iv

Abbreviations

xv

Aircraft Accident Report

v

Factual Information

4

Aircraft Accident Report

Pieces of the airplane wreckage were discovered floating on and beneath the surface of the Atlantic Ocean about 8 miles south of East Moriches, New York. The main wreckage was found on the ocean floor, between 40° 37' 42" and 40° 40' 12" north latitude and 72° 40' 48" and 72° 35' 38" west longitude. The accident occurred in dusk lighting conditions.

1.2 Injuries to Persons

Table 1. Injury chart.

1.3 Damage to Aircraft

The airplane was destroyed by the explosion, breakup and impact forces, and fire. According to insurance company records, the airplane was valued at $11 million.

1.4 Other Damage

No structures on the ground were damaged.

1.5 Personnel Information

The Safety Board reviewed the flight crew's flight- and duty-time limits and rest records and found no evidence that they were not within the limits established by Federal regulations. The cabin crew comprised 14 flight attendants.

1.5.1 The Captain (Left Front Seat)

The captain, age 58, was hired by TWA on May 20, 1965. He held airline transport pilot (ATP) certificate No. 1453736 with multiengine land and instrument ratings. The captain's most recent FAA first-class airman medical certificate was issued on March 22, 1996, and contained the limitation that he possess glasses for near vision.

Factual Information

8

〔图〕69'6"8.S6L6.ZZ2°TYP231'4"H12'6"36'2"3225'6"78'11225'2"\n

Figure 1. Three views of the 747-100 airplane.

Factual Information

6

〔图〕STABODY STATIONS90STA1/2°STASTA1342.5140395STA1463.5WL762.7WL366.5WL199.8WL 91WL 0WL 14.8\n

Figure 2. Location of some body stations and waterlines in the 747.

Factual Information

10

〔图〕CREW ENTRYSTATOP OFSERVICE DOOR440LOWERSTARH SIDECREWDECK480OVERHEADFLOORSTASTAHATCHSTA760STA40590STA240STA464STASTA830S

〔图〕STA65STA13191460STA1295STA1140.5STASTASTA12411350148000|o1010010WL199.8PRESSUREDECKWINGMIDREARWINGBODYBOXSPARSPARGEARGEA

Figure 3a. Location of fuselage sections 41, 42, and 44.

〔图〕插图 〔图〕STASTASTASTA25982775.323602484WL366.5292.5STA2742STA2638STA2658STABILIZERACCESSHINGE POINTDOORSECTION 48\n

Figure 3b. Location of fuselage sections 46 and 48.

Factual Information

13

Aircraft Accident Report

〔图〕Floor BeamsRing ChordFront SpaRear SparVentSpanwiseOpeningBeam 1MidsparSpanwise Beam 2ForwardKeel BeamDryIntercoastalVen

Figure 4a. A cross-section of the 747-100 wing center section.

Factual Information

14

Aircraft Accident Report

〔图〕PARTIAL RIB-BBL OFLOOR BEAMUPPER PANEL(SKIN)REAR SPARBL 0.0FRONT SPARLOWER PANEL(SKIN)RBLFWDSPANWISE BEAM 1LBLMIDSPARBOD

Figure 4b. Another cross-section of the 747-100 wing center section.

Factual Information

15

Aircraft Accident Report

〔图〕NOSE RADOME FAIRINGDORSAL FIN FAIRINGWING TO FUSELAGEFILLET FAIRINGTAIL CONE FAIRING\n

Figure 5. The 747-100 aerodynamic fairings.

〔图〕STATEROOMZONE1ZONE 2RECIRC FANSZONE 3HEATERSZONE 4ZONE 2ZONE 2GASPER FANCOLD AIR PLENUMTRIMMWATERMASTERMSEPARATORTRIM销售↓

Figure 6. A schematic diagram of the 747-100's air conditioning system.

Factual Information

19

〔图〕RIGHT SIDEFORWARDFAN#3 PACKCONTROL VALVEBLEEDAIRBYPASSSECONDARY HEAT#1 PACKFROMCHECK VALVEEXCHANGERDUCTSCONTROL VALVEENG

Figure 7. Top and side views of the 747-100's wing center section and air conditioning system.

Factual Information

20

〔图〕Spanwise BeaRear SparDucts from PacksPort for Groundto PlenumSpanwise Beam 2Air SourceSpanwise Beam 3Mid SparKeel BeamBa

Figure 8. Underside view of the 747-100's wing center section and air conditioning system.

Factual Information

21

Aircraft Accident Report

〔图〕PLENUMLOCATION OFAIR CONDITIONINGAIR CONDITIONINGLONG DUCT“S"DUCT(TO COCKPITRIGHT-SIDEDUCTS TOPASSENGER CABIN(SAME ON LE

Figure 9. The air conditioning ducts around the 747-100's center wing fuel tank.

1.6.1.3 747-100 Electrical and Wiring Information

1.6.1.3.1 747-100 Electrical Information

The accident airplane's electrical system consists of four engine-driven generators (one per engine) and one APU-driven generator,$^{37}$ all of which are capable of providing 115/200-volt,$^{38}$ three-phase, 400-Hz alternating current (a.c.). These generators are driven by a constant-speed drive located on the engine/APU gearbox, as applicable. Two 36-ampere (amp)$^{39}$-hour nickel cadmium batteries provide 24-volt backup direct current (d.c.) power and APU start power and electrical bus $^{40}$ and switching logic for power

°°

Factual Information

24

Aircraft Accident Report

〔图〕BMS13-42A-typeWire with Double Insulation Layer and TopcoatConductorInsulation, Inner LayerInsulation, Outer Layer*Top C

Figure 10. Diagram of a BMS13-42A Poly-X wire.

wire were specified and recovered. For example, the fuel quantity indication system (FQIS) wiring used in the fuel tanks was polytetrafluoroethylene (PTFE), per MIL-W-16878, which is also known by the trade name $Teflon.^{53}$

〔图〕CENTER WING(INBOARD)FORWARD(INBOARD)TANKNO.2 MAINNO.3 MAINTANKTANK(OUTBOARD(OUTBOARD)NO.1 MAINTANKNO.4 MAINTANKNO.4 RESE

Figure 11. A schematic diagram of the 747-100 fuel tank arrangement.

〔图〕-□FLOAT VALVE(INBOARD)NO.2 MAIN TANKFLOAT DRAIN VALVECENTER WINGTANK(OUTBOARDNO.1 MAIN TANKCHECK VALVESURGE TANKDRAIN LI

Figure 12. A layout of a typical 747-100 center wing fuel tank and left wing fuel tank vent system.

Factual Information

30

Aircraft Accident Report

〔图〕MANUALSHUTOFF VALVEFUELINGRECEPTACLESCROSSFEED MANIFOLDCHECK VALVEMANUALDEFUELING VALVEMANIFOLD DRAINVALVES (2 PLACESMAN

Figure 13. A 747 pressure fueling flow diagram.

Factual Information

32

〔图〕M.OVKKFHL.RESETFUEL QUATITY0INDICATORTW1ONCEntde mod 14001-t([)-.--eu-.--~4......n.-4.I aooo回FUELING VALVEINDICATOR LIGH 〔图〕PRESSURE FUELING MANIFOLDBFUELINGCENTER WINGSTATIONREFUEL VALVETANKUNITVOLUMETRIC SHUTOFFCOMPENSATORREAR SPARREFUEL VALV

Figure 14. A diagram of the 747 pressure fueling system components.

Factual Information

35

Aircraft Accident Report

Note:

Style of wire attachments shown are typical and not specific to any probe or compensator.

〔图〕specific to any probecompensatorTERMINAL BLOCK0TERMINAL BLOCKKNURLEDSURFACESERIES 4&00号SUBSEQUENTCLAMPWIRE。BUNDLEASSEMBL

Figure 15. Diagrams of a fuel quntity probe and a compensator.

〔图〕TERMINAL0STRIP(T347FUELFQISQUANTITYWIRESPROBEFUELFUELQUANTITYQUANTITYPROBEPROBE9FQISWIRINGCONNECT ORFUELQUANTITYPROBEFUE

Figure 16. The fuel quantity probe and compensator locations in the 747's center wing fuel tank.

Factual Information

38

〔图〕Right wing FQIS & fuelflow #4 wire routes (ref.)Tank Unit (probeFragmentsRight wing fuetank wire route(ref.)12Fuel quant

Figure 17. A schematic diagram of the 747-100's fuel quantity indication system wiring, fuel flow wiring, cockpit voice recorder wiring, and other wiring information.

〔图〕FLIGHTENGINEERCWT REARSTATIONSPAR CONNECTIONHFOR FQISCCPGIM SHOWNIN OPPOSITEIK SHOWNIJ OPPOSITE\n

Figure 18. Locations of the wire raceways in the 747-100.

Factual Information

41

Aircraft Accident Report

〔图〕FUEL SYSTEMENGINEFUELSHUTOFF(TYPICALREPUEL/DEFUELMANIFOLDCENTER WINGCROSSFEEDSCAVENGENO.NO.1RESERVERESERVENO.3NO.2MAINMA

Figure 19. A schematic diagram of the fuel pump locations in the 747.

Factual Information

43

Aircraft Accident Report

JFK using the underwing pressure method. Table 2 shows TWA's fueling records, which indicate TWA flight 800's fuel loading when it prepared to depart JFK.

Table 2. TWA's fueling records, indicating TWA flight 800's fuel loading when it prepared to depart JFK.

a. The cockpit indicator has an analog scale and a digital (rotating drum) scale. When the digital display indicates 300 pounds, the analog needle would indicate near “0.”

On the basis of the airplane's dispatch fuel load information and performance data, the Safety Board calculated that the fuel on board the airplane at takeoff was about 176,600 pounds (consistent with TWA's dispatch records) and at the time of the accident was about 165,000 pounds.

1.6.3 Maintenance Information

1.6.3.1 TWA's 747 Maintenance Inspection Program

During its investigation of the TWA flight 800 accident, the Safety Board reviewed TWA's maintenance program and maintenance recordkeeping procedures and conducted a detailed examination of the accident airplane's maintenance records. The Board's review of TWA's FAA-accepted continuous airworthiness maintenance inspection program for its 747 fleet revealed that it included the following inspections and intervals:

Factual Information

48

Aircraft Accident Report

method of alleviating wire chafing that involved installing a double layer of Teflon sleeves $^{97}$ over the wiring where it was routed within the fuel tanks. TWA maintenance records indicated that the AD was accomplished on the accident airplane on June 6, 1979.$^{98}$

On August 3, 1995, Boeing issued ASB 747-28-A2194, “Fuel-Distribution-Fuel Boost and Override/Jettison Pumps-Inspection.” The ASB stated that operators had removed eight fuel pumps from service that had fuel leaks at the pump/wire bundle interface and recommended that all operators test the resistance on each 747 boost and jettison/override pump “at the next opportunity,” replacing those that did not pass the insulation check. The SB stated the following:

after a long time, water can get inside the potting of the wire terminal assembly and cause corrosion. The corrosion in the wire terminal assembly can cause arcing between the power pins and the pump case. The arcing causes thermal expansion of the materials inside the cap. This expansion causes failure of the cap attachment flange or the attaching screws and a subsequent fuel leak.

According to TWA maintenance personnel, at the time of the accident, ASB 747-28-A2194 had not been accomplished on the accident airplane because the FAA had subsequently issued Notice of Proposed Rulemaking (NPRM) 96-NM-57-AD, which proposed an AD that would “require repetitive visual inspections to detect discrepancies of the wire terminal assembly, electrical connector, and wire insulation on the fuel pump; and replacement of the fuel pump with a new fuel pump, if necessary” and “repetitive insulation resistance test of the fuel pump wiring.” The resultant AD (AD 97-03-17) became effective on March 14, 1997, and TWA’s corresponding MO (72F57) was finalized on April 1, 1997.

1.6.3.2.2 Airworthiness Directives and Service Bulletins Applicable to 747 Structural Inspections

Nose Section (Section 41)—Internal Inspections

On January 31, 1986, the FAA issued AD 86-03-51, which required structural inspection of the nose section (section 41) on certain 747 airplanes, including the accident airplane, “to prevent sudden decompression of the fuselage.” On February 14, 1986, Boeing issued SB 747-53A2265, “Fuselage-Nose Section 41-Body Frame Structure Inspections and Crack Repairs,” which provided additional information regarding inspection and repair of this area.$^{99}$ According to the SB, “numerous body frame and other internal structure cracks in the upper deck, main deck and lower lobe of the [nose section],

Factual Information

UPPER DECK

〔图〕COCKPIT9\n

LOWER DECK

〔图〕52010009201241400142.5R3R1VAVGALOLSGALALAVLAVEL3\n

LOWER DECK

〔图〕R5■LAVLAVLAVLAVGALGAL不GMDESTOLAVLAVLAV323536383437653NR8854833637384052535466L504\n

Figure 20. A diagram of the 747-100 interior.

Factual Information

57

Aircraft Accident Report

Table 3. Windsor aloft measured by a weather balloon launched from Upton, New York.

These numbers were used during the Safety Board's trajectory study.$^{105}$ Review of the meteorological data revealed no record of significant meteorological conditions in the area or at the time of the accident.

1.8 Aids to Navigation

No difficulties with the navigational aids were known or reported.

1.9 Communications

No difficulties with communications were known or reported.

Factual Information

64

Aircraft Accident Report

〔图〕LONG ISLAND SOUNDCalvertonLONG ISLANDShinnecockInletMorichesInletATLANTIC OCEAN2002040 Miles\n

Figure 21. The wreckage location relative to the airplane's flightpath, JFK, and Long Island.

〔图〕NKFOKBrookhavenShinnecockIniletMorichesInletGreen zone口Red zoneYellow zone1010Nautical Miles\n

Figure 22a. Map showing the locations of the red, yellow, and green zones.

Factual Information

67

〔图〕插图

Figure 22b. A 747, color-coded to indicate the debris fields from which corresponding wreckage was recovered.

Factual Information

8

〔图〕NTSB\n

Figure 22c. The accident airplane's flightpath, color-coded to indicate the debris fields from which corresponding wreckage was recovered.

Factual Information

75

Aircraft Accident Report

Additionally, the Board's examination of the APU revealed no evidence of uncontainment, case rupture, fire, penetration of an object from outside into the APU, or preimpact damage.$^{146}$

1.12.5 Center Wing Fuel Tank Pumps and Other Components

Both jettison/override pumps from the CWT were recovered, examined, and tested. The jettison/override pumps were impact damaged (especially at the aft ends, where they had been attached to the rear spar), but exhibited no evidence of electrical shorts, arcing, preimpact breaching of the housing, or discoloration of the pump shaft. (Figure 23 is a diagram of a jettison/override fuel pump.) Examination revealed that the jettison/override pumps' thermal fuses and flame arrestor tubes were intact and exhibited no evidence of thermal or flame exposure. The CWT jettison/override pump motors exhibited no evidence of heat-related discoloration; the motor windings were the same tan color as those from the wing fuel tanks. (A similar jettison/override pump was heated in a laboratory test until the thermal fuse melted at slightly less than 400°F, and the motor windings were subsequently found to have darkened substantially.)

The CWT jettison/override pump control switches in the recovered cockpit wreckage were found in the “Off” position. Investigators examined the possibility that any one or all of these switch positions might have been altered during the accident sequence. Unlike the scavenge pump switch (which was recovered damaged in the “Off” position), the jettison/override switches did not have a locking feature, and the recovered switch positions could not be positively related to a preimpact position. The jettison/override switch base damage was consistent with the toggle switch being in the position to which it was spring-loaded and not at the “Off” position. However, according to TWA procedures, with only unusable fuel remaining in the CWT, these pump switches should have been off at the time of the accident. Further, the CVR did not record any comments regarding activation of these pumps.

Although the general condition of the left CWT jettison/override pump was clean, two pieces of a clear, soft, rubberlike foreign material were found on the outside diameter of the motor/impeller housing. Infrared examination by the National Aeronautics and Space Administration (NASA) determined that the material was a silicone-based substance. Additionally, the right jettison/override pump exhibited evidence of rubbing between the impeller edges and adjacent housing surfaces. Overhaul records from Crane Company, Hydro-Aire, Inc. (the pump manufacturer), described similar damage in other pumps and attributed it to worn shaft bearings.$^{147}$

〔图〕OUTLET TOREARMANIFOLDHOUSINGWINGSPARFLAMEREPLACEABLESUPPRESSIONPUMP MOTERPASSAGEIMPELLER ASSEMBLYELECTRICALCONNECTOR者PUM

Figure 23. A diagram of a jettison/override fuel pump.

Factual Information

80

Aircraft Accident Report

〔图〕插图

Figure 24. A photograph showing wreckage recovered with tangled and damaged wires attached.

Some wiring from the upper left fuselage aft side of the cockpit was recovered in loose fragments, whereas other wiring from this area was recovered tangled with structural debris. The upper deck electronics rack (E11) was recovered as an independent piece of wreckage, with fragments of wire bundles attached. No evidence of fire damage or soot deposits was observed on the recovered forward fuselage electrical components or wiring.

The BMS13-42A Poly-X wire from the accident airplane was stiff and inflexible, whereas recovered Teflon wires (used in the accident airplane's fuel tanks) and samples of BMS13-42A Poly-X wire obtained from a retired 747 (N93117) were supple and pliable. Recovered wiring from the area near the CWT was tangled and exhibited fire damage and soot deposits. About half of the original FQIS wires (as shown in Boeing's 747-100 engineering and production records) were recovered and identified. Although no evidence of arcing was observed on any of the recovered components connected to the CWT FQIS (the fuel quantity indicator, totalizer gauge, AIDS, VSO, and left wing refueling station), examination of other FQIS wires and wires routed adjacent to the FQIS wires revealed possible evidence of arcing in several locations. For example, examination of the recovered wiring and wreckage from the right wing revealed evidence of arcing on

Factual Information

06

〔图〕插图

Figure 25. Radar data showing vehicle and/or object tracks within 10 nm of TWA flight 800 just before the accident.

Factual Information

91

〔图〕插图

Figure 26. Primary radar returns that appeared near the accident airplane after 2031:12.

Factual Information

92

Aircraft Accident Report

〔图〕插图

Figure 27. Sequence of eight primary radar returns recorded by the Islip, New York, radar site about the time of the accident.

Factual Information

97

Aircraft Accident Report

data). Further, the Board varied the timing of the forward fuselage separation, conducting simulations with nose departure times at the beginning, middle, and end of the 4-second interval indicated for possible forward fuselage separation by the trajectory study. Table 4 shows scenarios involved in the simulations.

Table 4. A time line of simulation scenarios.

The Safety Board's simulations indicated that the longitudinal response of the airplane to the loss of the forward fuselage (with no flight control input) was to climb to above 16,000 feet msl. However, the airplane's lateral response to the loss of the forward fuselage (with no flight control input) was less clear. Evaluation of primary radar return data (from radar sites at White Plains, Islip, and JFK) indicated that after the initial event, the main portion of the airplane may have turned north after the forward fuselage separated and then turned south toward the main wreckage recovery site in the green zone. However, there was considerable scatter in the radar data after the in-flight breakup, and the recorded radar tracks differed, such that data from the Islip and White Plains radar sites indicated a more pronounced turn to the north followed by a turn back south, while data from the JFK radar site indicated a much straighter path to the wreckage site.

Simulations were run using bank angle inputs to match the radar data. The slow and intermediate simulations that used a nose departure time at the beginning of this interval did not match the radar data, nor did any of the simulations that used a nose departure time at the end of this interval. The fast simulation that used a nose departure time at the beginning of the interval could provide close matches with data from all three radar sites with different bank angle inputs. Simulations with a nose departure time in the middle of the interval did not match the Islip and White Plains radar points; however, intermediate and fast simulations with a nose departure time in the middle of the interval matched the JFK radar data. (See figures 28a through 28e.) The Safety Board's simulations indicated that the maximum altitudes that might have been reached by the TWA flight 800 aft fuselage after separation of the nose section ranged between 15,537 and 16,678 feet msl.

In addition to the uncertainties caused by radar track variation, radar data scatter, and imprecise aerodynamic characteristics previously discussed, there are uncertainties in the timing and physics of the final breakup that affect the final seconds of the simulations. The timing of the wing tip and WCS failure were based largely on witness statements;$^{192}$ therefore, they are not precise to the second. Further, the change in aerodynamics caused by the wing tip failure at a high AOA is uncertain, as is the effective ballistic coefficient of the main body after the WCS failure.

Factual Information

98

Aircraft Accident Report

〔图〕插图

Figure 28a. Roll angle for nose off at 2031:15.2 cases.

〔图〕插图

Figure 28b. Map view of nose off at 2031:15.2 cases.

Factual Information

99

Aircraft Accident Report

〔图〕插图

Figure 28c. Altitude for nose off at 2031:15.2 cases.

〔图〕插图

Figure 28d. East position for nose off at 2031:15.2 cases.

Factual Information

100

Aircraft Accident Report

〔图〕插图

Figure 28e. North position for nose off at 2031:15.2 cases.

1.16.3 Sequencing Study Information

In December 1996, the Safety Board formed the Sequencing Group, which comprised Safety Board investigators and representatives from the parties to the investigation, to evaluate the sequence of the airplane's structural breakup and correlate and compare proposed accident scenarios with the structural observations. To this end, the Sequencing Group examined individual pieces of the recovered structure, two-dimensional reconstructions or layouts of sections of the airplane, and various-sized three-dimensional reconstructions of portions of the airplane.$^{193}$ The Sequencing Group also evaluated the relationship of the location of pieces of wreckage at the time of recovery$^{194}$ to differences in fire effects (soot deposits on surfaces and fractures, changes in electrical conductivity) noted on pieces that are normally mated or adjacent to each other. The following subsections discuss the two- and three-dimensional reconstructions of the recovered airplane wreckage, the Sequencing Group's examinations of the wreckage, and the accident airplane's breakup sequence identified by the Sequencing Group.

〔图〕R\n

Figure 29. A photograph of the right side of the large three-dimensional reconstruction, with the support scaffolding visible.

Factual Information

105

Aircraft Accident Report

〔图〕RightBowingWingFronSparPotablePotablewatorwaterbottlebottleCamo floor levdKeel beaFrames-siringersLF6A月145900RF32253800V

Figure 30. A schematic drawing of the lower portion of the fuselage skin in the area

Factual Information

107

Aircraft Accident Report

〔图〕Location of witnessmarks on theApproximateaft side of fronlocation of witnessspar stiffenersmarks on the afside of theup

Figure 31. Diagram showing the approximate shape of the upper skin panel stiffener and its location before being struck by SWB3 as the beam rotated forward.

Factual Information

116

Aircraft Accident Report

1.16.4.3 Metallurgical Examination of the Fuselage Joint Between Sections 41 and 42

The Safety Board was aware of concern about possible assembly problems with the fuselage joint between sections 41 and 42 (which extend forward and aft, respectively, from STA 520) on $747s^{212}$ and considered the possibility that a failure at this joint might have occurred, initiating the breakup sequence. This joint area was wholly contained within the forward fuselage section that the Sequencing Group identified as having separated from the airplane early in the breakup sequence.

Overall examination of the forward portion of the airplane showed that sections 41 and 42 exhibited uniform crushing damage that extended across the bottom of the fuselage from about 3 feet below the left-side main cabin windows (S-39L) to above the right-side main cabin windows (S-14R). This damage was consistent with this section of the airplane impacting the water intact in a right-wing-low attitude. The structures examination also revealed that portions of the joint between sections 41 and 42 were separated during the water impact, whereas some portions remained intact.$^{213}$ The portions of the STA 520 joint that separated were examined for evidence of preexisting fatigue or other preexisting damage. The Safety Board's examination of this joint revealed that all fractures were typical of overstress separations, with no evidence of fatigue, fretting, or significant corrosion.

1.16.4.4 Metallurgical Examination of the Forward Cargo Door

The Safety Board also considered the possibility that the forward cargo door (the forward edge of which is located several feet aft of STA 520 on the lower right side of the fuselage) separated from the accident airplane in flight and that this separation initiated the breakup sequence. The Board examined the pieces of the forward cargo door, which were recovered from the yellow zone. All eight of the latching cams at the bottom of the door were recovered attached to pieces of the lower end of the door and were in the latched position. Additionally, the latching cams and pieces of the cargo door remained attached to the pins along the lower door sill. The hinge at the top of the door was broken into several pieces, but the hinge pin still held the various pieces of the hinge together. There was no evidence to suggest that this hinge separated. The forward cargo door exhibited severe crushing deformation and fragmentation, very similar to damage observed on the adjacent fuselage structure.

Factual Information

120

Aircraft Accident Report

Table 5. 747-100 wing center section/center wing fuel tank SWBs and spars failure strengths.

1.16.5 Jet A Fuel Vapor and Fuel Tank Research and Testing

During its investigation of the TWA flight 800 accident, the Safety Board was hampered by the lack of available pertinent research data regarding the flammability and other characteristics of Jet A fuel and its behavior in airplane fuel tanks. In an attempt to develop data that would be useful in the investigation, the Board assembled a team that included the most renowned experts in the combustion and fuel-related fields and, with the help of these experts, developed and implemented a research program. The objectives of this research program were to (1) develop an understanding of the thermal and vapor environment within the 747 CWT; (2) determine the chemical and physical properties, flammability, and combustion behavior of Jet A fuel; and (3) develop computer models of the combustion process within the 747 CWT to assist in the determination of the ignition location. The researchers performed laboratory experiments, scale-model tests, and numerical simulations to examine the explosion of Jet A fuel/air vapor under conditions similar to those that existed within TWA flight 800's CWT at the time of the accident. Specifically, the researchers wanted to identify the flammability limits and combustion behavior of Jet A fuel and the propagation of flames through the compartments within the CWT structure. Figure 32 shows a flow chart of the fuel-related research conducted to support this investigation.

The FAA has certified commercial jet airplanes based on the assumption that fuel tanks are considered to always be flammable and has required manufacturers to demonstrate that any energy inside the tank would be below the MIE for Jet A vapors to ensure safe operation of the airplane. However, at the time of this accident, specific information about fuel tank flammability and Jet A thermochemistry was largely unavailable. The research conducted during this investigation enabled analysis of the flammability of aircraft fuel tanks and provided a scientific basis for evaluating methods of reducing fuel tank flammability, as identified in safety recommendations previously issued to the FAA.$^{225}$

〔图〕Fuel MassFuel1/4 scale CWTFlame-frontLoadingweatheringcombustionquenchingexperimentexperimentsEarly EventCode Developmen

Figure 32. A flow chart of the fuel-related research conducted to support this investigation.

°

°

Factual Information

127

Aircraft Accident Report

〔图〕ForwardpJPMO_Side ViewDry BayBay 1Bay 2Left Mid BayRight Mid BayLeft Aft BayRight Aft BayTop ViewForward3D View\n

Figure 33. Side view, top view, and three-dimensional view of wing center section/center wing fuel tank bays.

Factual Information

128

Aircraft Accident Report

Table 6. A summary of the maximum and minimum recorded village temperatures at 13,700 feet msl during the emulation flight test.

a. In this section, the term “bay 1” refers to the fuel bay between SWB3 and SWB2.

b. In this section, the term “bay 2” refers to the fuel bay between SWB2 and the mid spar.

As previously discussed, the CWT is configured within the airframe such that the air conditioning packs are located in an enclosed bay, directly below and very near $^{242}$ the bottom surface of the CWT.$^{243}$ Because of the high temperatures at which the air conditioning packs operate, the ambient air temperature within the pack bay increases when the air conditioning packs are operated. As a result of these elevated temperatures, heat in the pack bay is transferred into the CWT through its bottom surface. Some of this heat is transferred out of the CWT (for example, into the wing main fuel tanks or the main landing gear wheel well), and some of it remains in the CWT. This heat flow results in a significant increase in the temperature of the CWT fuel and ullage, which was shown in table 6. The temperature variations within the CWT are illustrative of this heat flow. In general, the hotter temperatures in the CWT ullage were located in the lower, center portions of the CWT (nearest the heat source), whereas cooler temperatures were measured around the sides and top of the CWT (where heat was transferred out of the CWT).

Examination of the temperatures measured in the surge tanks, vent stringers, and the No. 3 fuel tank during the TWA flight 800 emulation flight test indicated that these temperatures were about the same or slightly warmer than the outside air temperature during the ground portion of the flight test (airplane temperatures of 85° to 90° F versus

Factual Information

130

Aircraft Accident Report

Table 7. Fuel/air mass ratios and ullage temperatures measured at the bay 2 sampling port during the TWA flight emulation flight test.

During subsequent flight tests, investigators collected additional vapor samples from the CWT ullage. The fuel in the CWT was not consumed, nor was the CWT refueled during the series of flight tests in which vapor samples were collected. Examination of the series of vapor samples showed that the vapor composition gradually changed through a process known as fuel weathering. Subsequent flash point testing (conducted at CIT)$^{250}$ of liquid fuel samples collected at each flight test did not indicate a significant change in flash point (less than 3° F) as the fuel weathered during the series of flight tests.

1.16.5.1.2 Testing of Fuel Samples from TWA Flight 881 (Arriving at JFK from Athens, Greece)

On October 1, 1996, a fuel sample (about 50 milliliters) was obtained from the CWT of a TWA 747 that had just arrived at JFK from Athens, Greece, as TWA flight 881. Because TWA flight 881's CWT had been refueled in Athens and was believed to have encountered conditions (flight profile, thermal environment, and weathering) similar to those encountered by the accident airplane during its flight from Athens to JFK (before it departed as TWA flight 800), the Safety Board considered this fuel sample to be representative of the fuel in the accident airplane's CWT when the overpressure event occurred. This fuel sample was taken to Saybolt International's laboratories for standardized fuel testing. The flash point of this fuel was measured at 114°F, and it was noted that the sample contained an antistatic additive that resulted in the sample having 90 picosiemens/meter electrical conductivity.$^{251}$

Factual Information

146

Aircraft Accident Report

10 drops per minute. The experiment was allowed to run for up to 25 minutes or until a circuit breaker tripped. In some cases circuit breakers were reset and, the test continued.$^{302}$ Table 8 shows the results of the wet short-circuit tests conducted by Lectromec.

Table 8. Results of wet short-circuit tests conducted by Lectromec.

According to the Lectromec report, during these tests, scintillations were generally visible soon after the tests were started. During the tests in which flashing occurred, flashing events frequently repeated at various intervals (ranging from several flashes per second to more than 1 minute apart) throughout the test. Oscillogram measurements indicated that the peak electrical energy dissipation (about 90 J, with power peaks between 3 and 6 kW)$^{303}$ during these flashing events occurred during a test in which a wet-short was initiated by dripping lavatory fluid onto Poly-X (BMS13-42A) wiring. The estimated peak electrical energy dissipated before the circuit breakers tripped during the tests in which strong arcing was observed was about 2 kJ.

Factual Information

149

Aircraft Accident Report

shows a "trip-curve" chart for a typical circuit breaker (excerpted from Texas Instruments Klixon® specification), which depicts the relationship between the amount of current that can pass through a circuit breaker and the amount of time until the circuit breaker opens.

〔图〕插图

Figure 34. A trip-curve chart for a typical circuit breaker (excerpted from Texas Instruments Klixon® specification).

The USAF handbook also describes typical circuit breaker failure modes, including welding and erosion/wear of the electrical contacts, material transfer, overcurrent, and contamination. Further, the handbook states that “the mechanical structures have changed over many years; therefore, age is also a factor.”

Boeing Document D6-40359 describes a study, “Impact of Electrical Overloads on Wire and...Circuit Breakers Used on 707, 727, and 737 Airplanes,” dated February 2, 1972. The document stated that “early in 1971, [an air carrier] reported a failure of a...circuit breaker. As a result of this failure and subsequent investigations, there has been consideration given to the need for electrically checking all [similar] circuit breakers which are in service.” The document further stated that circuit breakers that have been in service “could not reasonably be expected to meet the overload current versus trip time calibration curves of a new breaker” and contained a description of how the bimetallic strips may lose integrity and degrade with exposure to high internal temperatures.

Subsequently, Boeing studied about 1,200 circuit breakers that had been removed from service during a fleetwide survey$^{306}$ and found that “initial results of on-board testing showed that 72 percent of all breakers tested satisfied the procurement specifications.” Further laboratory testing of the 28 percent that did not meet procurement specifications

Factual Information

151

Aircraft Accident Report

1.16.6.3 Aging Effects on Material Properties of Wires

To better understand the condition of wiring in aging transport-category airplanes, the Safety Board contracted with Raytheon Systems Company (Raytheon) to test aged wire samples to the specifications for newly manufactured wire of that type. The aged wire samples were similar in age and type to the wiring on TWA flight 800 and were obtained from three recently (not more than 6 months before the wire samples were obtained) retired airplanes—two 747s, one manufactured in 1970 and the other in 1973, and a Douglas DC-10 manufactured in 1973.$^{309}$ The wire samples ranged in length from 1 foot to more than 26 feet and in size from small, partial bundles with just a few wires to large bundles with many wires. The samples were taken from pressurized fuselage locations, most of which were relatively protected from traffic, exposure to light, or other potential environmental stresses. Figures 35a, 35b, and 35c show the sample origination locations for the 747 manufactured in 1970, the DC-10, and the 747 manufactured in 1973, respectively. Table 9 further describes the origin and potential environmental exposure of each sample.

Table 9. The origin and potential environmental exposures of test samples.

Factual Information

152

Aircraft Accident Report

〔图〕STASTASTASTASTA10001140124113501480STAFRONT SPAROVERHEADSTASTA90REAR SPAR700HATCH830STAMID SPARSTA520140TOP OFPILOTSFLOO

Figure 35a. Sample origination locations for the 747 manufactured in 1970.

〔图〕Sample 2Samples 4 & 8UPPER CONSTANT SECTIONSample 1 & 31305.250840,2501851.850103910791273495535575615652115911991239134

Figure 35b. Sample origination locations for the DC-10 manufactured in 1973.

Factual Information

153

Aircraft Accident Report

〔图〕Sample 7Samples 1,2,3,4STASTASTASTASTA1000114012411350148Sample 5STAdVdS LNOWSOVERHEADSTASTA90REAR SPARHATCH700830STAcva

Figure 35c. Sample origination locations for the 747 manufactured in 1973.

In addition to visual examinations, Raytheon tested the wire samples against the following performance requirements for new wire:

· Mechanical and electrical tests, including concentricity tests (to measure the thickness of the insulation layers and the overall sample geometry), dielectric tests (to determine the integrity of wire insulation), insulation resistance tests (to measure the electrical resistance that the insulation imparts between the conductor and ground), insulation tensile strength and elongation tests (to identify materials that have weakened because of thermal breakdown, UV degradation, or other mechanisms), and notch sensitivity tests (to measure the ability of a wire insulation to resist the propagation of a nick or cut through the insulation layers to the conductor).

· Thermal tests, including blocking tests (to ensure that the insulation does not adhere to itself after heating), low-temperature (cold-bend) tests (to ensure that the wire retains physical and electrical integrity through mechanical and electrical stress at the temperature extremes), shrinkage tests (to indicate whether the wire insulation is physically and dimensionally stable when exposed to short-term high temperature), thermal shock tests (to measure an insulation's ability to resist shrinkage and expansion following several thermal cycles to temperature extremes), and wrapback tests (to evaluate the wire's ability to withstand thermal stress while under mechanical stress).

Factual Information

158

Aircraft Accident Report

Figure 36. Chart showing the percent probability of failure of Kapton wiring in different locations in the P-3 airplane.

Factual Information

166

Aircraft Accident Report

· A combustible fuel/air mixture existed in all fuel tank air spaces as modeled.

· Scavenge pump switch was turned off at the time of the event.

Boeing's CWT ignition fault tree analysis report identified the ignition of the CWT as the top-level failure mode, and the impact of all lower-level elements that compose the tree was evaluated using mathematical probability relationships. The fault tree analysis was constructed downward from the top-level event by progressively determining the conditions required for each lower-level event. Eventually, the tree reached a level where an event had no preceding events; this event was termed a "basic element." Boeing's CWT fault tree analysis report identified 167 basic elements. According to Boeing, each basic element was assigned a failure rate and an exposure time (the maximum time presumed by Boeing that the condition could go undetected) based on either in-service data or engineering estimates. These failure rates and exposure times were combined to produce a probability of failure for each element, using the following formula: probability of failure = 1 x $e^{-(failure rate x exposure time)}$. Some of the basic elements identified in Boeing's fault tree analysis report are shown (with their failure rates, exposure times, and calculated event probability) in table 10.

Table 10. Several fault tree analysis basic elements, as identified in Boeing's fault tree analysis report, with their failure rates, exposure times, and calculated event probability.

Factual Information

167

Aircraft Accident Report

Boeing's fault tree analysis report contained a graphical representation of the fault tree and a table showing the failure rates and exposure times for the fault tree's basic elements. During its review of the fault tree analysis report, the Safety Board noted that, in some cases, the information in the graphical representation of the fault tree did not agree with the information in the table. For some elements, the disagreement was several orders of a magnitude. According to Boeing's graphical representation, the probability of an ignition event in the CWT (the top-level failure mode) was $8.45 \times 10^{-11}$. Although Boeing did not calculate the top-level failure mode probability using the table values, when Safety Board investigators substituted the table values into the appropriate positions in the fault tree and completed the calculations, the table values showed that the probability of an ignition event in the CWT was $1.46 \times 10^{-5}$. Table 11 shows the differences between the data presented by Boeing in a graph and a table for of the some basic elements of its fault tree analysis.

Table 11. Some of the basic elements of Boeing's fault tree analysis, showing the differences between the data presented in Boeing's graphical and tabular representations.

Factual Information

168

Aircraft Accident Report

Table 11. Some of the basic elements of Boeing's fault tree analysis, showing the differences between the data presented in Boeing's graphical and tabular representations.

$^{a}$. This event was listed in the fault tree two separate times with different values.

Because of these inconsistencies in Boeing's fault tree analysis report, because of other questions the Safety Board had regarding the detailed information used to construct the fault tree,$^{340}$ and because the fault tree did not reflect the findings of the ongoing investigation, on June 4, 1998, the Board requested that Boeing further review and revise the fault tree analysis report, as necessary. On July 29, 1998, Boeing responded and indicated the following, stating that although the Board was correct in some of its points:

we [Boeing] do not believe that revising the [fault tree analysis] by incorporating new data or changing the probability numbers will help identify new areas to inspect or help identify the cause of the accident. We believe it would be more productive to continue the various inspection and modification programs that are presently underway.

The Safety Board subsequently asked NASA's George C. Marshall Space Flight Center (MSFC) to review Boeing's fault tree analysis report and the Board's concerns. In a November 25, 1998, letter, NASA MSFC personnel stated the following:

Many of the probabilities, failure rates, and/or exposure times were much lower than would reasonably be expected. The probability of occurrence should be higher and/or exposure times should be longer on many of the basic events...Based on our evaluation of the tree and the information provided by the [Safety Board], the subject fault tree analysis quantification cannot stand up to peer review and should not be viewed as realistic. It should be noted that the logic of the tree could not be fully evaluated since we did not have access to the engineering drawings and schematics of the system. However, based on previous systems experience, we would expect the tree to be constructed differently.

NASA MSFC personnel further indicated that the exposure time numbers used in Boeing's fault tree analysis report appeared to be overly optimistic and that the use of more realistic numbers would result in an increased probability of a top-level failure.

Factual Information

171

Aircraft Accident Report

contained multiple explosions that were intentionally ignited within the pump motor cavity, preventing the ignition from transferring to the fuel tank. (Figure 37 is a diagram of a scavenge pump.) Because the Board became aware of instances in which the flame arrestor tube was broken or missing, investigators also conducted a series of tests to determine whether the scavenge pump motor housing would contain explosions with only one of the flame suppression system components (either the flame arrestor tube or the check valve) functioning. In all tests, the scavenge pump housing contained the explosions, and the explosive atmosphere surrounding the test pump did not ignite.

During its postaccident examination of other transport-category airplanes, the Safety Board observed that components in the electrical connector on a scavenge pump in a retired 747 were deteriorated. (Boeing ASBs 747-28A2194 and 747-28A0043, both dated January 18, 1996, “encouraged” operators to inspect for degradation in connectors.) Investigators observed that several scavenge pump electrical connectors had been manufactured of a material that degraded with long-term exposure to fuel. Boeing provided the Board with a service history for the 747 scavenge pump for the 10 years preceding the accident. When queried about the brevity of this history, Boeing indicated that most operators did not return the scavenge pumps to the manufacturer for overhaul.

〔图〕IMPELLERCHECK VALVEFORWARD(CWT SIDE)REAR SPAR(LANDING GEARBAY SIDEFLAME SUPPRESSIONPASSAGE (TUBEMOTOR\n

Figure 37. A diagram of a scavenge pump.

〔图〕COCKPIT ESCAPEFLIGHT ENGINEERDOOR (RIGHTSTATION (RIGHTSTA 920STA 100WATER BOTTLESON FORWARD SPARPRESSURE SEALS TOWING WI

Figure 38. Diagram showing the areas typically inspected on 747 airplanes.

〔图〕插图

Figure 39. A chart showing the 737 fuel pump conduit wear rate that was identified by the FAA during representative 500 wire inspections.

〔图〕73°45'74°0073°3073°1573°0072°45'72°307201572°00'LegendWitness situation41°1541°15'AirborneOn landSurfing, swimmingor on

Figure 40. Map showing the witness locations and the location of the accident site.

Factual Information

244

〔图〕Eastwindflight 507Shinnecook InletBeach Lane BridgeMoriches InletBoat witness(approximately)US Air flight*21TWA flight 8 〔图〕LegendWitness situationWitnessTWA flight 800*Last secondary returnDebris fieldsYellow zoneRed zoneGreen zone\n 〔图〕ZS\n

Figure 41. A map showing the debris zones and the positions of the five witnesses of special interest.

〔图〕73°15'73°1073°573°0072°4572°35'72°5572°5072°4072°3041°00Legend40°5540°55Line of sight features527694Witnesses649Line of

Figure 42. Map of sight lines plotted for 11 witnesses.

Factual Information

253

Aircraft Accident Report

〔图〕插图

Figure 43. Diagram of warhead detonation fragment pattern.

The Safety Board's computer simulations (using this missile performance data and taking into account the winds at the time of the accident) showed that the locations on the surface from which a missile launched at TWA flight 800 would result in selfdestruction within 40 feet of the airplane were extremely limited. Figure 44 shows airplane locations at which, if a high-performance missile were launched at the airplane, the missile would selfdestruct within 40 feet of the airplane. Note that, at the scale shown, the launch area that would result in a selfdestruction within 40 feet of the airplane is not as wide as the line showing the 15-second perimeter in this figure.

Simulations further showed that, because of the pattern in which missile fragments are dispersed after selfdestruction (see figure 43), launches from many of these positions would result in no fragments having a trajectory that would result in penetration of the CWT. For example, because of the nature of missile guidance, a launch from most positions on the perimeter shown in figure 44 would result in the missile catching the aircraft from behind. The area in front of the missile is not in the fragment dispersal area.

Factual Information

254

Aircraft Accident Report

〔图〕插图

Figure 44. Diagram showing airplane and missile launch site locations that could result in missile selfdestruct within 40 feet of the plane.

1.18.4.7.1 Missile Visibility Study

On the basis of performance data for a fairly high-performance, shoulder-launched missile, to collide with TWA flight 800, a hypothetical missile would have had to have been launched between about 4.5 nm and 11 nm from shore. Safety Board investigators consulted with missile experts in U.S. intelligence and defense communities to determine if the launch and flight signatures of a shoulder-fired missile would be detectable at those distances. Weapons systems experts indicated that there were no data available on which to base an answer to this question. Consequently, the Board conducted (and videotaped) missile visibility tests on April 28, 2000, at Eglin Air Force Base, Fort Walton Beach, Florida, to determine if observers could see and describe the characteristics of a missile launch.

The missiles were launched just after sunset on April 28, 2000, when the surface weather conditions and the position of the sun were similar to the conditions on Long Island at the time of the accident. Observer groups were positioned every 2 nm from the launch point out to 14 nm.$^{529}$ The observers were told where to look, and they listened to the launch countdowns via handheld radios. Although nine countdowns were

Analysis

268

Aircraft Accident Report

〔图〕e9P0\n

Figure 45. A graphical illustration of the equidistance tendency.

in the sky than it actually was, which explains why the streak of light may have been reported by some witnesses as having originated at the surface of the earth. It should be noted that almost all of the witnesses who reported that the streak of light originated at the surface viewed the streak of light on the far side of intervening terrain features, such as houses or the barrier island. These witnesses would have perceived the streak of light to be about the same distance as the intervening object because the elevation angle of the object would have been perceptually reduced.

Further, if an observer views an object for which few or no depth perception cues are available, the motion of the observer's head will cause the observer to perceive the object moving concurrently with the observer's head. At the time of the reported observations, 13 of the 38 witnesses who reported that the streak of light ascended vertically or nearly so were in vehicles (including boats) and one of the witnesses was paddling a surfboard. Other witnesses may also have been moving. The effect of the concurrent motion of the observer's head and the observed object would have caused the streak of light to appear to have been moving differently than it was actually moving, and this effect could have caused the streak of light to have appeared to rise vertically or nearly so. The closer that the streak of light appeared to a given witness, the greater the apparent motion of the streak of light to that witness.

Further, the airplane's relatively low elevation above the horizon$^{570}$ could have led many witnesses to believe that they were observing a flare or fireworks. (As discussed previously, because of the absence of depth perception cues, witnesses who saw the airplane in crippled flight could not be expected to provide good information about how far away it was or its direction of travel.)$^{571}$ As discussed in section 1.18.4.3, a witness's perception of events is driven by context cues, experience, and expectations. The fact that it was July (a month often associated with fireworks) and that several flares had been launched in the area during the week preceding the accident could have led witnesses to

Analysis

282

Aircraft Accident Report

〔图〕PressureTOTALCWTRefuelingIZERGAGEAIDSFUELPaneParalleUNITFLOWCVR Wires⑥LefWingWiring BehindGalley CVSOFUELBeneath FloorCM

Figure 46. Diagram of the center wing tank fuel quantity indication system wiring on the 747 and its common connectors.

Analysis

300

Aircraft Accident Report

compared with airplanes that have heat sources adjacent to fuel tanks (30 percent of the fleet operational time).$^{651}$

The Safety Board is pleased that the FAA agrees that the heat generated by the packs located under the CWT significantly increases the flammability exposure in the CWT. Further, the Safety Board concludes that the placement of heat-generating equipment under a fuel tank containing Jet A fuel can unnecessarily increase the amount of time that the airplane is operating with a flammable fuel/air mixture unless measures are in place to either (1) prevent the heat from entering the CWT or (2) eliminate the flammable vapors inside the CWT.

Although the Safety Board supports the FAA's efforts to address fuel tank flammability in new designs, the Board is concerned about the FAA's apparent premise that minimizing, rather than eliminating, fuel tank flammability is an acceptable goal. The ARAC working group (the FTHWG) established to advise the FAA in its rulemaking efforts concluded that the safety record of fuel tanks located in the wings (which the FTHWG calculated were flammable about 7 percent of the fleet operational time) was adequate and that if the same level of safety could be achieved in CWTs, the overall safety objective could be achieved. Thus, the FTHWG proposed limiting the exposure to flammable conditions in all fuel tanks to less than 7 percent of the expected fleet operational time.

FAA staff indicated during meetings with Safety Board staff on February 29, 2000, that the FAA did not intend to endorse the FTHWG's proposed 7 percent exposure criteria. However, the FAA's draft $guidance^{652}$ concerning compliance with the proposed new requirement to minimize fuel tank flammability reflects exposure criteria that are tied to the exposure of wing tanks; these criteria are reiterated in the FAA's May 17, 2000, letter to the Board describing its actions in response to several safety recommendations, including A-96-174 and -175. The draft guidance states that "the intent of the regulation is to require that the exposure to formation or presence of flammable vapors is equivalent to that of an unheated wing tank in the transport airplane being evaluated," which, as previously mentioned, the FTHWG calculated to be about 7 percent of the fleet operational time.

The Safety Board registered two concerns about such exposure criteria in its comments on NPRM 99-18. First, the Board stated that because the exposure criteria is based on a fleetwide average, it does not account for increased risks that may exist at specific locations, during certain time periods, or for certain flights.$^{653}$ Second, the Board

Appendixes on pages 313-416 are contained in a separate document

for download from http://www.ntsb.gov/investigations/summary/AAR0003.html