Aviation Investigation Report A98H0003
1.12.3 Examination of Recovered Electrical Wires and Components
- 184.108.40.206 - General
- 220.127.116.11 - Control and Tracking of Electrical Wires
- 18.104.22.168 - Wires and Cables with Electrical Arcing Damage
- 22.214.171.124 - Positioning of Wires and Bundles from Exhibit 1-4372
- 126.96.36.199 - Identification and Description of Arc-Damaged Wires from Exhibit 1-4372
- 188.8.131.52 - Examination of the Nine Arced Wires and Cables
- 184.108.40.206 - Examination of Identified Aircraft Systems Wires with Copper Melt
- 220.127.116.11 - Examination of Remaining Unidentified Wires
The aircraft wiring was severely damaged by the forces of impact. Additional mechanical damage could have occurred to some of the wiring during recovery operations. Many of the wire segments had been broken into segments between 10 and 100 cm (4 and 39 inches) long.
All of the recovered wires were examined, primarily to identify any with signs of melted copper. As the fire did not reach temperatures high enough to melt copper, any areas of melted copper would indicate that an electrical arcing event had occurred.
Wire segments that showed signs of soot or heat damage, or that could be identified as being from the B section of the aircraft, were segregated from the other wires. The condition of the wires, heat damaged or not heat damaged, was used to help define the boundaries and heat pattern of the fire. The wire examination also assisted in explaining the spread of the fire and the loss of aircraft systems.
Recovered electrical components that had been installed within the fire-damaged area were examined for internal failures; such failures can be a source of heat, and therefore have the potential to be a source of ignition. Examination of these components did not show any evidence of their being involved in the initiation of the fire.
Approximately 250 km (155 miles) of wire was installed in the aircraft. Although 98 per cent of the structural weight of the aircraft was recovered, it is estimated that a slightly lesser percentage of the wire was recovered. The examination of the recovered wire identified several wire segments that exhibited melted copper consistent with electrical arcing damage. Some of the arcing damage was too small to see without magnification. It cannot be assumed that all arced wires were recovered.
During the examination, any wire that was deemed to be of interest to the investigation was tagged, assigned an exhibit number, and entered into a wire database. The database increased to contain information on approximately 3 000 individual wire segments. Segments that exhibited melted copper, or showed signs of significant heating, were tracked through the same exhibit numbering system that was adopted to track pieces of wreckage. The wire segments that exhibited melted copper were further analyzed, which required that the melted sections be cut from the original segment. These individual melted copper wire sections were assigned new exhibit numbers and were tracked within the database. Where practical, the wire segments from the fire-damaged area that were identified as to their location in the aircraft were incorporated into the reconstruction mock-up (see Section 1.19.3).
When metal-to-metal contact occurs between an energized conductor and a source of electrical ground, or between an energized conductor and another conductor of different potential, a short-circuit is created. Typically, a short-circuit will result in an arcing event. Copper or aluminum wire that has arced will typically be identified by an area of resolidified metal (see Figure 20). After the arcing event, as the metal resolidifies from its molten state, it often forms a distinctive bead. Short-circuit arcing will normally trip the associated CB, and typically result in a relatively small amount of conductor being melted.
In an arcing event, copper is vapourized and expands to many thousands of times its volume when in a solid state. High temperatures and pressures are generated in the localized vicinity by this vapourization and by the electrical discharge (arc) as it passes through the surrounding air. If flammable materials are nearby, they can be ignited by the heat of the electrical discharge, by the heat from the gases emanating from the area, or by the molten globules of copper that are typically ejected from the arc site.
With aircraft voltages of 115 V AC, an arcing event without physical contact is very difficult to obtain or sustain. However, in areas where wire insulation is destroyed in a fire environment there is an increased likelihood that such an arc can be initiated and sustained, and that the conductor will melt over some length. This rapid progressive arcing phenomenon, known as arc tracking, depends on numerous variables, the most predominant being the type of insulation used on the conductor.
Among the thousands of wire segments examined from SR 111, areas of melted copper were found on 21 exhibits that consisted of 17 individual wire segments, and 4 cable assembly segments that each comprise 3 separate spirally laid wires (3 individual wires twisted together to make up the one cable). All except one of the conductors that exhibited arcing had either polyimide or ETFE-type insulations. The one exception was a conductor insulated with a modified XL-ETFE (see information about Exhibit 1-3029 in Section 18.104.22.168).
Polyimide insulation does not melt when exposed to elevated temperatures; instead, it will pyrolyze, or thermally degrade. Thermogravimetric analysis of polyimide insulation in air shows that it begins to dissociate at about 500°C (932°F) and will be completely dissociated by 650°C (1 202°F).
One of the characteristics of polyimide insulation is that under the correct conditions it can arc track, either wet or dry. (STI1-63 (video clip)) Arc tracking can occur when the polyimide insulation decomposes from heat produced by either an electrical arc or by a fire. The exposure to heat produces an electrically conductive and thermally stable carbon char. The resulting carbon deposit provides a current path to perpetuate the arc. This arcing can further cause the surrounding insulation to decompose, allowing the electrical discharge to propagate, or track, along the wire. In some cases, the current required to form an intermittent or sustained electrical discharge (arc tracking) is below the "current versus time" trip threshold for the associated CB. In these cases, an undetected fault condition, producing intense heat, will develop.
Frequently in an arc-tracking event, the heat from the initial localized electrical discharge will degrade the insulation on adjacent wires, causing a cascade of arcing and burning between multiple polyimide wires in a bundle. This is referred to as flashover. Flashovers can result in the catastrophic failure of entire wire bundles, resulting in the loss of power or signals to all equipment supplied by the affected wires.
Unlike polyimide insulation, ETFE insulation will melt and burn. It melts in the 260°C to 270°C (500°F to 518°F) range, and will burn with a flame when exposed to a fire at temperatures above 500°C (932°F). When the flame source is removed, ETFE insulation is self-extinguishing. (STI1-64 (video clip)) The FAA testing of ETFE has shown that it will not support wet or dry arc tracking. The thermal decomposition of ETFE does not result in the formation of a conductive carbon char.
Testing has shown that although the ETFE insulation will melt and volatilize relatively quickly in a fire environment, it can be difficult to initiate a conductor-to-conductor or conductor-to-ground arcing event in a fire environment. The testing, using a butane flame to melt ETFE insulation, showed that it could take 20 to 30 minutes for arcing to occur between adjacent conductors. This suggests that a small creeping flame on an insulation blanket would be unlikely to degrade ETFE insulation and result in an arcing event in the time it would take for the flame to consume the available insulation blanket material and move away.
This testing also showed that in some cases the associated CB would immediately trip with the first arcing event; however, in other cases several arcing events would occur before the CB tripped. Neither the time to arc nor the tripping of the CB is predictable in a fire environment. If the initial arcing event does not trip the CB, arcing can occur on the same conductor some distance apart as the fire propagates and affects other sites on the wire.
In addition to the 21 exhibits that displayed copper melt, a single bead of once-molten copper, 2 mm (0.08 inches) in diameter, was recovered. This bead was found trapped in the damaged cooling fins on top of an emergency lights battery pack, which, based upon heat damage, was determined to have been located above the forward cabin drop-ceiling just behind the cockpit door. (See Figure 27.) This bead was likely fractured off, most likely during the impact sequence, from the end of an arced wire in the vicinity of the battery pack. No further determinations could be made about this bead.
One of the copper melts was not a result of arcing damage, but was determined to be the direct result of a welding operation during the manufacturing process of the wire (see Section 22.214.171.124).
During the wreckage recovery, one bundle of entangled wires was recovered that yielded 9 of the 20 wire segments that were found to have arcing damage. Collectively, this bundle was designated Exhibit 1-4372 (see Figure 21).
The wire insulation was missing from some of the wire segments, having been burned off by the fire or damaged by the fire and then stripped away at the time of impact. It is also possible that some additional wire insulation damage took place during the wreckage recovery process.
Several of the wire segments in Exhibit 1-4372 were identified as belonging to one of three aircraft wire runs (FAC, FBC, and FDC) (see Figure 7). As installed, these three wire runs, along with wire runs AAG, ABG, and the IFEN wire bundle, ran parallel to each other on the right side of the fuselage over top of Galley 2, between the cockpit rear wall at STA 383, and the aft end of Galley 2 at approximately STA 420. Just forward of the aft cockpit wall, the IFEN wire bundle entered one of two spare 102-cm (40-inch) long conduits that were installed between STA 383 and STA 420 (see Figure 5).
At STA 420, wire runs FAA, FBA, and FDA were broken out of wire runs FAC and FBC and FDC respectively. At STA 420, wire runs FAA, FBA, and FDA, along with wire runs AAG and ABG were routed over the crown of the aircraft to the left side of the fuselage. At STA 420, wire runs FAC, FBC, and FDC dropped down to run across the top of the forward passenger cabin ceiling; this routing was chosen to avoid contact between the bundles and the R1 door in the open position.
The recovered segments of wire runs FBC and FDC were each approximately 2.5 m (100 inches) long and had two wire clamps, identified as the marriage clamp, still attaching them together. The two wire runs were accurately positioned in the reconstruction mock-up based on the positive identification of individual wires, and by identifying the installed position of the marriage clamp at approximately STA 427. When these cable segments were positioned in the reconstruction mock-up, the forward end was located near the cockpit rear wall, and the aft end was located near STA 475. Accurate positioning of the segments from wire runs FDC and FBC, based on the known location of the marriage clamp, allowed for improved accuracy when positioning other wires from the same area. Heat damage patterns were used in the positioning of other wires from the same area.
Of the nine wire or cable segments with areas of melted copper that were recovered from Exhibit 1-4372, four were positively identified as being segments of the IFEN PSU cables (1-3790, 1-3791, 1-3792, and 1-3793) (see Figure 22). This identification was based on remnants of coloured ETFE insulation remaining attached to the wires. On each of these four segments, one end was heavily matted, or crushed together, and the opposite end was fractured and frayed. Exhibits 1-3790 and 1-3792 both had distinctive regions where the tin coating was missing from the wire strands on all three wires. Exhibits 1-3791 and 1-3793 did not have a similar region of missing tin.
Of the remaining five segments, based on adhering remnants of ETFE insulation, four (1-3794, 1-3795, 1-3788, and 1-10503) were identified as segments of the 16 AWG IFEN 28 V DC control wire used to control the PSU outputs. The total length of these four segments was 97 cm (38 inches). The fifth (1-3796) was a segment of polyimide-film-insulated, nickel-coated 16 AWG wire. This wire segment could not be associated with a specific circuit; however, because it was nickel coated, it is known that it was not from the IFEN system.
The copper melt on Exhibit 1-3796 was unique in that it had an area of melted copper that encapsulated all of the outer wire strands over a distance of 2 cm (0.79 inches); however, the melted copper still had nickel-coated wire strands protruding at both ends, indicating that the heat was highly localized and confirming that an arcing event occurred.
When Exhibit 1-4372 was being untangled, Exhibit 1-3796 was removed from wire run FBC at a position corresponding to approximately 79 cm (31 inches) forward of the known location of the marriage clamp. Although this corresponds to a location aft of the cockpit wall, it cannot be confirmed that this is where this segment of wire had been installed.
Wire run FBC did not contain any 16 AWG wires; however, 16 AWG wires were contained in adjacent wire runs. FAC had 25, FDC had 2, and ABG had 1. In total, there were twenty-eight, 16 AWG wires routed, within these various wire runs, above the Galley 2 area. It could not be determined whether Exhibit 1-3796 originated from one of these cable runs, and if it did, which one.
To assess whether the four IFEN PSU cable segments (1-3790, 1-3791, 1-3792, and 1-3793) and four control wire segments (1-3794, 1-3795, 1-3788, and 1-10503) had been located inside the conduit, they were submitted to the RCMP Chemistry Division (Ottawa and Halifax) for analysis. The single control wire (1-10503) was not submitted; except for one small piece of ETFE insulation, it was bare. The nickel-coated 16 AWG segment was also submitted to determine whether any FEP material had transferred to it during the impact sequence; this was to determine the proximity of this wire segment to the conduit. Furthermore, a request was made to identify any other materials found on these exhibits. Samples of the FEP conduit, the nylon clamps used to support the conduit, tie-straps, and other materials were supplied for comparative analyses. See Table 12 for the results of this testing.
Exhibit 1-3791 was also submitted to the Canadian DND Quality Engineering Test Establishment for further chemical analysis of the debris entrapped in the matted end, adjacent to the area of melted copper identified as Exhibit 1-14723.
Each of the four PSU cables consisted of three individual wires (three electrical phases). It was not possible to determine which wire had been used for which phase; therefore, each wire was arbitrarily labelled either phase A, B, or C.
Table 12 details the results of the physical and chemical examinations on the nine wire and cable segments recovered from Exhibit 1-4372. The matted ends of the four PSU cables were straightened. The cables were then measured from the straightened ends to determine the overall length of each phase and the distance to each arc location. Figure 22 depicts the locations of the arcs and other distinguishing features on the cables, relative to each other.
Number of Wires/Gauge (AWG) Wire Length (cm)
|Melted Copper on Phase (ph) @ Distance from Straightened End (cm)||Chemical Analysis||Other Observations|
|FEP Present (cm)*||Nylon (cm)*|
ph A - 88
ph B - 112
ph C - 100
|ph A @ 11 (1-14746)
ph A @ 62 (1-12652)
ph B @ 77 (1-12651)
ph C @ 69 (1-11182)
|0 to 15,
ph A - 46
ph C - 3
ph C - 13
ph C - 46
91 to 102
|86 to 97||Tin coating missing between 58 and 89 cm. Small pieces of red silicon rubber material 10 cm from end. Yellow fibreglass insulation embedded in matted end.|
ph A - 115
ph B - 116
ph C - 117
|ph A @ 9 (1-14723)
ph A @ 62 (1-12653)
ph B @ 65 (1-12654)
|0 to 15,
ph B - 10
ph B - 37
ph B - 54
76 to 91
|91 to 107||1-14723 was within matted end of wires. Small pieces of red silicon rubber material and yellow fibreglass insulation embedded in matted end.|
ph A - 97
ph B - 97
ph C - 104
|ph A @ 22 (1-12668)
ph B @ 23 (1-12666)
ph B @ 23 (1-12667)
|15 to 30
ph A - 45
ph A - 48
ph C - 53
ph C - 98
91 to 104
|0 to 15||Tin coating missing between 58 and 89 cm. Small pieces of red silicon rubber material 13 cm from end.|
ph A - 102
ph B - 120
ph C - 120
|ph A @ 64 (1-12669)
ph A @ 66 (1-12670)
ph A @ 67 (1-12732)
|0 to 15,
ph B - 7
ph C - 37
ph C - 43
ph C - 54
45 to 61
|91 to 107||Small pieces of red silicon material 9 cm from end of longest wire at matted end.|
|End of wire - (1-11166)||9 (from melted end)||Frayed wire strands at opposite end of melted copper missing tin coating.|
|End of wire - (1-11167) and 2 cm from melted end (1-11168)||9 and 19 cm from melted end||Tin missing from outer wire strands over last 7 cm from melted end. Small pieces of red silicon rubber material noted between strands.|
|End of wire - (1-11173)||No FEP found||Tin missing from strands over 6 cm from melted end and again 10 cm farther along wire for 10 cm; tin was present over the last 6 cm.|
|End of wire - (1-11163)||Not submitted||Tin coating missing over 2.5 cm from melted end.|
|3 cm in from end of wire and 2 cm long
|No FEP found||Nickel-coated wire with a small piece of polyimide film trapped in wire strands.|
* The RCMP Halifax analysis reported FEP and nylon in ranges whereas the Ottawa analysis stated a distance from an end.
Four wire segments with regions of copper melt were positively identified from wire numbers marked on their insulation. Three of these four wires had a terminal lug connector still attached to one end, allowing the wires to be accurately positioned in the reconstruction mock-up.
All four of these wire segments originated from behind the overhead CB panel, which constitutes the upper part of the overhead switch panel (see Figure 23). The switch panels and CB panel are attached to a fibreglass enclosure, referred to as a "housing," within which terminal strips, bus bars, and other components are mounted. The wires are routed into and out of this housing through two holes located on the left and right sides of the aft end of the housing.
Exhibit 1-6976 consists of a segment of a 6 AWG nickel-coated, polyimide-insulated wire, approximately 102 cm (40 inches) long with the wire number B205-4-6 stamped on the insulation. This number identifies it as a segment of the left emergency DC bus feed wire. One end of the wire segment had two flag lugs attached that would have been connected to the left emergency DC bus bar located behind the overhead switch panel. The known location of the flag lug to bus bar connection allowed for this segment to be accurately located in the reconstruction mock-up. The end with the melted copper was located approximately 15 cm (6 inches) outside the right hole in the overhead CB panel housing. This placed it above the forward right side of the cockpit ceiling.
There was a 2-cm (0.8-inch) long section of resolidified copper located about 97 cm (38 inches) from the flag lugs. The nickel coating was missing from the conductor, for about 6 cm (2.4 inches) on either side of the melted area. The outer surface of the resolidified copper was smooth and did not exhibit any of the normal characteristics associated with an arcing event, such as surface porosity or irregular surface structure. Radiographs of this area showed it to be a solid mass, without voids. This was the only nickel-coated wire segment that exhibited melting of the nickel coating beyond the area of the resolidified copper. This melted copper was considered to have been the direct result of a welding operation during the manufacturing process of the wire.
Welding or "splicing" is done to avoid interruptions during the wire manufacturing process. Spliced joints are required to be flagged and removed prior to the wire being installed in an aircraft. If not removed, such a solid piece of copper conductor, especially larger gauge wires, such as the 6 AWG wire, can be susceptible to fatigue cracking. The left emergency DC bus feed wire was installed in the occurrence aircraft in a location where the wire run was relatively straight, and the vibration potential was low. These mitigating factors would reduce the risk of fatigue cracking; there was no sign of fatigue cracking in the melted area of Exhibit 1-6976.
Exhibit 1-3029 was identified by the wire number B205-1-10 as being a segment of the left emergency AC bus feed wire (see Figure 20). Although this wire was identified by the Boeing wiring database as a nickel-coated, polyimide-insulated wire, it was in fact an XL-ETFE (BXS7008) insulated tin-coated wire. The recovered segment was approximately 122 cm (48 inches) long. One end of the wire segment had two flag lugs attached, which would have connected it to the left emergency AC bus bar. The end opposite the flag lugs exhibited melted copper on the surface of the conductors. As installed, the end with the melted copper would be located approximately 15 cm (6 inches) outside of the right oval hole in the overhead switch panel housing, and in the immediate vicinity of the melted copper on the left emergency DC bus feed wire. The insulation remained intact on this wire, from the flag lug to within 17.5 cm (7 inches) of the melted copper at the other end.
Exhibit 1-1733 was identified by wire number as a segment of B203-974-24, a 24 AWG nickel-coated, polyimide-insulated wire that was identified as part of the Engine 2 fire detection loop "A" circuit. The recovered segment was approximately 39 cm (15 inches) long. One end of the wire segment was attached to a remnant of module block 46 that was originally attached to modular track S3-613, located on the right side of the overhead switch panel housing. The other end of the wire segment exhibited melted copper. The known location of the module block allowed it to be relocated in the reconstruction mock-up of the overhead housing. This placed the area of melted copper approximately 15 cm (6 inches) outside the right oval hole of the overhead switch panel housing.
Exhibit 1-12755 was identified by wire number as a segment of B203-189-22, a 22 AWG nickel-coated, polyimide-insulated wire that was part of the high-intensity lights (supplemental recognition) wingtip strobe lights. The recovered segment was approximately 81 cm (32 inches) long. This wire segment exhibited melted copper on one end, and frayed wires at the other. The frayed wire end did not appear to have had a terminal lug or pin attached to it; it appeared to be a segment from a longer wire. As installed, this wire was routed from the push button switch S1-9094, identified as HI INTENSITY LTS, located in the overhead switch panel, to plug P1-420 pin X located on the overhead disconnect panel behind the upper avionics CB panel. The area of melted copper could not be accurately positioned between the start and end points but the wire was routed in wire runs AMJ and AMK, which are routed out the oval hole in the right side of the overhead switch panel housing.
The remaining eight wires with arc damage could not be identified regarding their function, nor could their specific location within the aircraft be determined. Table 13 contains a description of these eight wires.
|Exhibit Number||Length (cm)/Wire Size/Wire Coating/Insulation New Exhibit Number||Melted Copper||Observations|
|1-4689||19 cm/10 AWG/tin coated/
no insulation remaining
|0.2 cm from end. Copper melt encompasses 8 to 9 wire strands 0.3 to 0.5 cm long and 1 to 2 mm wide. Numerous cavities and voids in melt.||Tin coating missing from wire strands. Both ends frayed. Polyimide film caught in strands but could not positively determine whether it is from this wire.|
|1-11897||16 cm/10 AWG/tin coated/
no insulation remaining
|Both ends melted.
Straight end of wire 14 cm from curved end.
6 cm from curved end.
|Wire embrittled from end to end.|
|1-12756||40 cm/18 AWG/tin coated/polyimide film
|Melted copper over 0.2 cm long, 0.8 cm in from one end and encompassing 8 to 10 wire strands. Numerous voids and craters in melt.||M81381/21 (polyimide insulated/tin coated).|
|1-11252||7 cm/24 AWG/nickel
|Melted copper at end extending 0.15 cm down wire. Small bead of copper fused to strands adjacent to melted end.||Wire embrittled from end to end. Nickel coating missing in some areas and charred or blackened material adhering to strands.|
|1-3713||9 mm/24 AWG/nickel
|Melted from end to end with no obvious coating remaining.||Numerous voids and holes through wire. No insulation remaining.|
|1-12809||25 cm/24 AWG/nickel
|1.2 cm from one end 0.2 cm long encompassing all strands with a small bead protrusion.||No insulation remaining.|
|1-3700||10 cm/20 AWG/nickel
|End of wire melted over 1 cm with a 1.5 mm bead on end.||No insulation remaining.|
|1-3718||28 cm/20 AWG/nickel
|End of wire melted over 0.6 cm ending in a flatted point.||No insulation remaining. Melt area blackened from smoke.|
 National Fire Protection Association (NFPA) 921 Guide for Fire and Explosion defines an arc as a high-temperature luminous electric discharge across a gap. Temperatures at the centre of an arc can range up to 5 000°C (9 032°F) or more.
 The NFPA 921 defines a bead as a round globule of resolidified metal at the end of the remains of an electrical conductor that was caused by arcing. The TSB notes that a bead may also form at any location along the copper conductor without necessarily causing the wire to separate into two halves. Copper arc damage also does not always form spherical beads but may take on any irregular shape and exhibit voids or cavities.
 Pyrolysis is the breaking apart of complex molecules into simpler units by the use of heat.
 A testing procedure in which changes in weight of a specimen are recorded as the specimen is heated in air or in a controlled atmosphere, such as nitrogen.
 Dissociation is a reaction involving the breakdown of chemical compounds.
 Not easily decomposed or otherwise modified chemically.
 American Society for Testing and Materials Designation B172-90, Standard Specification for Rope-Lay Stranded Copper Conductors Having Bunch-Stranded Members, for Electrical Conductors, Section 6 - Joints, Subsection 6.2 states bunch-stranded members or rope-stranded members forming the completed conductor may be joined as a unit by soldering, brazing, or welding. Subsection 6.3 states joints shall be so constructed and so disposed throughout the conductor so that the diameter or configuration of the completed conductor is not substantially affected, and so that the flexibility of the completed conductor is not adversely affected.
- Date modified :
- Date modified: