On January 19, 1999, at 0326 Alaska standard time (AST), a Boeing 747-100 airplane, N481EV, sustained minor damage when a fire occurred in the number four engine while taxiing from landing at the Fairbanks International Airport, Fairbanks, Alaska. The crew of three, and the one jumpseat passenger, were not injured. The airplane was operated by Evergreen International Airlines, Inc., of McMinnville, Oregon. The flight was operated under 14 CFR Part 121 as a supplemental air carrier international cargo flight. The flight departed Chitose, Japan, at 2159 AST, and an IFR flight plan was filed. Visual meteorological condition prevailed at the time of the incident. Interviews of the flight crew and ground personnel, conducted by the NTSB investigator-in-charge (IIC), and FAA inspectors, revealed that ground personnel observed flames coming from the number four engine after the airplane had landed, and that they notified the airport control tower via radio. Control tower personnel relayed this information to the cockpit crew. The crew then shut down the number four engine and activated the on-board fire suppression equipment. During an interview on January 20, the captain told the NTSB IIC that there were no cockpit indications of fire except for a rapidly rising exhuast gas temperature (EGT), nor any cockpit indications of the thrust reverser remaining deployed. Postincident inspection revealed soot located outside the engine cowling, and the number four thrust reverser pneumatic drive unit (PDU) was found in the thrust reverser deployed position. Internal inspection of the engine revealed no overtemperature damage, or fuel soot streaking through the turbine section. All observed evidence indicated that the fire was confined to the tailpipe area of the engine. Facsimile correspondence between Pratt and Whitney product support engineers and the operator on January 21, indicated that if the thrust reverser remained deployed during taxi, an engine surge with rising EGT condition and a tailpipe fire could result in the JT9D-7A engine. The engine fire indicating system in a JT9D-7A engine is a heat sensitive coil inside the engine cowling. The PDU, which was received in the stowed position, was disassembled under the guidance of an NTSB investigator at the AlliedSignal facility in Anniston, Alabama, on February 24, 1999. The PDU was found to have several assembly discrepancies as compared to component maintenance manual (CMM) 78-31-63, Rev 19. Some of the discrepancies were: shims were installed on the wrong side of the slinger, the oil seal was installed backwards, and shimming spring washers to the feedback gears were installed incorrectly. The stroke of the interlock mechanism was measured to be 1.230 inches. The specified value is 1.318 to 1.326 inches. According to AlliedSignal, this stroke measurement is required to allow the gear change mechanism to fully shift (from extend, to retract). Damage was evident to the gear change mechanism outer ratchet ring, and locking pawl assembly, to include a broken pawl. All parties present on February 24 agreed that the broken pawl would result in binding in the retract direction. Other discrepancies in the buildup are noted in the teardown report (attached) submitted by AlliedSignal. The NTSB investigator who attended the inspection on February 24 told the NTSB IIC that he concurred with all factual observations noted on that day. The PDU was overhauled 113 hours and 26 cycles prior to failure of the retract pawl on the gear change mechanism. The broken retract pawl and gears were inspected at the NTSB metallurgical laboratory and all fracture surfaces were indicative of overstress. The geographic locations of operation for the previous 26 cycles were all warm weather locations, except for three. On January 7, and January 13, the airplane landed at Anchorage, Alaska, where the temperature was -1 degree Fahrenheit. On January 13, the airplane also landed at Khabarovsk, Russia, where the temperature was -13 degrees Fahrenheit. The ambient surface temperature at Fairbanks when the incident occurred was -27 degrees Fahrenheit. An interview by the NTSB IIC with the chief inspector for the overhaul company revealed that other PDUs assembled by the same mechanic were recalled, and disassembled for inspection. No similar discrepancies were found with these PDUs. At the time of assembly of the failed PDU, the company had a policy of quality assurance spot checks at various steps during PDU buildup, as specified in the manufacturer's overhaul manual.