History of Flight
On May 16, 2010, about 2117 eastern daylight time, the flight crew of United Airlines flight 27, a Boeing 757-200, N510UA, declared an emergency because of a fire in the cockpit and diverted to Washington Dulles International Airport (IAD), Chantilly, Virginia. The flight was en route from John F. Kennedy International Airport (JFK), Queens, New York, to Los Angeles International Airport (LAX), Los Angeles, California. The airplane landed at IAD without incident. No evacuation was conducted, and none of the 7 crewmembers or the 105 passengers sustained injuries. The scheduled, domestic passenger flight was operating under 14 CFR Part 121 on an instrument flight rules flight plan. Night visual meteorological conditions prevailed.
The captain was the flying pilot. During post incident interviews, the captain stated that the flight departed JFK for LAX about 2030. The flight crew reported that the taxi, takeoff, and climbing phases were normal.
According to the captain, the first indication of a problem occurred shortly after the airplane leveled off at its cruise altitude of 36,000 feet. Both the captain and first officer heard a hissing sound and saw smoke emanating from below the glare shield. The captain stated that within a couple of seconds a fire started and he saw flames emanating from the top of the glare shield. The smoke was not thick but acrid. The captain got a burning sensation in his eyes and nose before they donned oxygen masks and smoke goggles. The captain transferred control to the first officer and told him to declare an emergency with ATC, which occurred at about 2136.
The captain got out of his seat because flames were in front of him and he needed to reach the Halon bottle, which was on the back wall behind his seat. As he reached for it, his oxygen mask and goggles were torn off because he moved beyond the hose reach. He retrieved the bottle, re-donned his mask and goggles, and discharged the bottle. The fire went out momentarily but re-ignited after a few seconds. His mask came off again when he opened the cockpit door to receive a second Halon bottle from a flight attendant. He discharged the second bottle and the fire went out. The first officer turned off all four window heat switches shortly after the fire re-ignited. The captain then returned to his seat and regained command.
The first officer informed the captain that ATC was asking where they wanted to go; the captain replied, “go to Washington Dulles Airport,” because it was about 50 miles away. The captain indicated he thought IAD was the best choice due to better maintenance and fire support.
The autopilot remained engaged until the airplane descended through about 10,000 feet, when the flight crew began to slow for landing. As the airplane descended through about 500 feet, a loud explosive bang was heard and the inner pane of the captain’s windshield cracked. The captain then transferred control to the first officer because his vision was impaired by the shattered windshield. The captain stated he could see through the window but it was “spider webbed.” The first officer landed and reduced speed; the captain then taxied off the runway.
Damage to Airplane
On May 16, 2010, an NTSB investigator visually examined the captain’s No. 1 windshield and the area near it. The windshield had fractured and its lower aft portion exhibited signs of high thermal damage and soot staining. Remnants of the J5 power terminal block, which had been mostly consumed by fire, were in this area.
Visual examination of the J5 power terminal block and its attachment hardware revealed that a lock washer was not present beneath the screw that connects the wiring harness terminal lug to the terminal block. Examination of the remaining terminal blocks on the captain’s windshield, as well as the three power terminal blocks and two terminal sensing blocks on the first officer’s windshield, revealed that no lock washers were present beneath any of the screws that attach the wiring harness terminal lugs to their respective terminal blocks.
Personnel Information
The captain, age 59, held a multi-engine land ATP certificate with type ratings in Boeing 737, 747-4, 757, and 767; Citation CE-500; Lockheed L-300; and Airbus A320. He held a first-class FAA medical certificate dated May 3, 2010. His most recent Boeing 757 flight proficiency check occurred in May 2009. He had flown 170, 60, and 7 hours in the 90 days, 30 days, and 24 hours before the incident.
The first officer, age 45, held a multi-engine land ATP certificate with type ratings in Boeing 757, 767, and A320. He held a first-class FAA medical certificate dated August 20, 2009. His most recent Boeing 757 flight proficiency check occurred in November 2009. He had flown 191, 79, and 7 hours in the 90 days, 30 days, and 24 hours before the incident.
Aircraft Information
All six cockpit windows have an electrically heated conductive layer for anti-icing and defogging. Two thin electrical conductors, called bus bars, are installed on the upper and lower edges to transmit current from power wires to the heating film.
The windshield heat system for the captain and first officer’s No. 1 windshield comprises three power terminal blocks: J1 at upper aft, J4 at upper forward, and J5 at lower aft. Two sensing terminal blocks (J2 and J3) are along the upper edge for sensing. Power is supplied by a window heat controller through a wiring harness with molded terminal connector lugs. Each terminal lug is connected to its terminal block with a screw and a lock washer.
The windshield is a bolted-edge design with internal and external aluminum retainers, using PPG’s Nesatron heating film. A polysulfide moisture seal protects interlayer materials. According to United’s maintenance records, the windshield had accumulated 12,065 hours and 2,325 cycles since installation.
A review of United Airlines maintenance records revealed that TIMCO Aviation Services replaced the Number 1 left-side window in January 2007 because its lower, forward, inboard corner exhibited delamination. The replacement window (part number 141T4801-49) was installed on January 29, 2007, in accordance with Boeing 757 AMM 56-11-01/401.
On May 15, 2010, the airplane diverted to McCarran International Airport (LAS) due to a report of an electrical odor in the cockpit. United maintenance inspected at LAS and found no defects or odors. The airplane was ferried to San Francisco International Airport (SFO) for additional maintenance. During the flight from LAS to SFO, no electrical odors were noted by the flight crew; only an unusual odor from forward galley ovens. At SFO, maintenance could not find the source and replaced both forward galley ovens as a precaution. The airplane returned to service on May 16, 2010.
During the first revenue flight from SFO to JFK after maintenance, the flight crew reported that the right No. 1 windshield lower outboard power connector appeared burnt and was hot to the touch. The captain of that flight stated he was aware of the previous reports and looked over the cockpit. He saw the lower left terminal block on his windshield was blackened or charred and too hot to keep his finger on. He submitted an information item via ACARS. Upon arrival at JFK, a maintenance technician met the captain. The technician stated he visually inspected the window and referred to United Airlines AMM 56-11-00/601, which indicated the window should be replaced due to bus bar discoloration. However, the limitations section of the United AMM stated, “When bus bar(s) show signs of blackening or burning, the condition is acceptable for continued service, although the window must be replaced within 50 flight-hours.” The technician and another lead technician decided to defer the maintenance write-up for 50 flight hours. The mechanic signed off and returned the airplane to service. A review of the Boeing AMM revealed it does not contain a limitations section and states to replace the windshield if there are any indications of arcing. On May 20, 2010, United Airlines revised AMM 56-11-00/601 from a customized version to a pure Boeing AMM.
Flight Recorders
The airplane was equipped with a Honeywell FDR in good condition; data were extracted normally. The data did not contain information regarding the windshield heating system.
The CVR was a Honeywell solid-state model sent to the NTSB laboratory. It was in good condition and audio was extracted without difficulty. None of the audio was pertinent to the incident, consistent with being overwritten by subsequent events.
Fire
On May 16, 2010, an NTSB investigator visually examined the captain’s No. 1 windshield (part number 141T4801-49, serial number 06039H6554) and its wiring harness before removal. Examination revealed relatively little fire damage except soot staining and surface charring in the area of the captain’s windshield. Soot staining moved up the aft portion of the window frame brace and below the glareshield. Slight charring existed on about a 1-square-inch area on top of the glareshield.
The lower aft portion of the windshield exhibited the most thermal damage and soot staining. Remnants of the J5 power terminal block, mostly consumed by fire, were found. The terminal connector lug was charred and melted; electrical contacts appeared oxidized and corroded consistent with high temperature exposure. The screw embedded in the terminal block also exhibited heavy oxidation. No additional fire damage evidence was found in the cockpit.
Test and Research
The windshield, wiring harness, and attachment hardware were shipped to PPG Aerospace in Huntsville, Alabama for detailed examinations on June 16, 2010, under NTSB supervision. Examinations revealed the upper 75 percent of the windshield’s inner pane had a dense network of cracks emanating from the lower, aft corner. Upper and lower bus bars were intact with very little discoloration. No evidence of moisture ingression or seal repairs was found. Soot was found on the inboard glass pane along the entire aft edge. Electrical resistance checks between each terminal and its braid wire were within specified limits. Except for the J5 connector, all terminal blocks and sensors remained intact. The J5 connector was extensively burned with only the aft edge remaining. Underneath the J5 block, small areas of charred and soft pliable sealant were present.
Examination of the wiring harness showed no evidence of chafing, rubbing, or repairs. The J5 terminal connector and lug had charring damage and the housing was cracked. A small area of bluing was on the lug surface mating with the screw head. Wire insulation was consumed for about 2 inches from the connector and charred for an additional 2 inches. Parts of the cap shoulders on J2 and J3 connectors were fractured. The J5 screw measured 0.75 inches and exhibited a small area of bluing on its head lower surface. The lug thread depth measured 0.410 inches (specified minimum 0.38 inches). No evidence of cross threading was found.
As noted, GSO replaced the No. 1 left-side window on January 29, 2007, per Boeing 757 AMM 56-11-01/401. A review of United Airlines’ and Boeing’s versions of AMM 56-11-01/401 revealed both stated to connect electrical wires to power terminals and tighten screws to 25-30 inch-pounds; however, neither manual contained a step to install a lock washer between the screw and terminal. After the incident, Boeing released a temporary revision (TR56-1004) to add specific instructions and graphics for lock washer installation. United incorporated this information.
Boeing 757 IPC 30-41-52-03, Revision 81, contains a figure depicting a lock washer between the screw and terminal block for each power and sensing terminal, but part numbers were not provided. On June 3, 2008, Boeing released MOM 1-855016051 regarding elimination of wire bundle breakdown data from IPCs for multiple aircraft models, stating that data would be removed over 12 months.
Additional Information
The Safety Board has investigated several similar smoke/fire events at windshield heat terminal connections since 2004. In June 2010, Boeing informed the Board that from January 2001 through May 2010, they received 35 reported smoke/flame events associated with screw and lug type power terminal connections on Boeing 747, 757, 767, and 777 No. 1 windshields. In some cases, the initiating cause was determined to be cross-threaded screw at the J5 terminal block; in other cases, undetermined. The TSB of Canada investigated a similar incident on a Boeing 777 operated by Air France in 2002 (report A02C0227).
Another windshield power terminal overheating event occurred on July 27, 2010, involving Delta Airlines flight 1553, a Boeing 757-200, N649DL, which diverted to Spokane due to an electrical odor. No injuries occurred. Examination of the J5 terminal block revealed a lock washer was not present beneath the screw.
In 2004, Boeing redesigned the No. 1 windshield terminal block installation with a pin and socket connector instead of the lug connector, and incorporated it into new production airplanes in mid-2004. Boeing planned service bulletins for retrofit, and the FAA planned to issue airworthiness directives. Boeing redesigned the 757 windshield in 2005 to incorporate a pin-socket connection. Service Bulletins 757-30-0019 and 757-30-0020 provide for inspections or windshield replacement. At the time of the incident, United had not performed the actions specified in SB 757-30-0019 on the incident airplane, nor were they required to do so by the FAA.
In September 2007, the NTSB issued Safety Recommendation A-07-50, asking the FAA to issue an AD to replace windshield heat terminal blocks on all Boeing 747, 757, 767, and 777 airplanes. This recommendation resulted from two incidents: an electrical fire on an Air Greenland 757-200 in January 2004, and an incident involving American Airlines flight 1477 on January 25, 2004. In April 2008, the FAA issued an NPRM to make inspection of J5 terminal blocks mandatory and windshield replacement a terminating action. The NTSB asked that inspections of J1 and J4 terminals be included. Before this incident, the NTSB made several inquiries about the status of the pending AD.