Multiple Failures of Oversight Cited as Cause of In-flight Engine Fire

Air Safety Week, Dec 6, 1999

Safety Board calls for functional testing of emergency evacuation systems, to include door opening mechanisms and slide/raft deployment

The use of a part six thousands of an inch too small for the job set the stage for a fuel line leak that caused an in-flight engine fire some 33 months later. The case of the July 9, 1998 engine fire on American Airlines [AMR] Flight 574 shortly after takeoff from San Juan, Puerto Rico, illustrates the extremely fine tolerances under which the airline industry operates to achieve what in many respects is a remarkable safety record. At the same time, the event illustrates the cascading effects of a seemingly innocuous .006 inch deviation which, over time, led to a fuel line leak that erupted into an engine fire. The subsequent investigation of this event by the National Transportation Safety Board (NTSB) brought to light safety issues ranging from maintenance to training, inter-company communications, cockpit procedures, emergency evacuation, and to the design of engine fire warning systems. As Member George Black remarked at the Board's recent hearing on the case, "This is another example of the usefulness of accident investigations. You start with an engine fire and end with the emergency doors."

First alert, a muffled thud

About 88 seconds after an other wise routine takeoff, the crew of the American Airlines A300 jet with seven flight attendants and 215 passengers aboard heard a sound like a muffled thud. Crews on previous flights reported hearing thumps during landing gear retraction, so Capt. Ken Graham and First Officer Tom Szot at first thought they were hearing a manifestation of the same problem. Within seconds, however, the left engine fire warning light illuminated and the fire checklist appeared on the airplane's electronic centralized aircraft monitor (ECAM). The crew cut power, disengaged the autothrottles, began to turn back to the airport and declared an emergency.

The "thud" Graham and Szot heard was the ignition of the leaking fuel in the number 1 engine, according to Jeff Kennedy, the NTSB investigator-in-charge for this incident.

The conditions for this fire had been set nearly three years before, when the Number 1 engine was overhauled by MTU in Hanover, Germany, which was performing the work on contract for American Airlines. The MTU technicians used an adapter bolt that was too small "to hold the fuel lines in place," according to Jean-Pierre Scarfo, power plants group chairman for the investigation.

In 1997, American Airlines repaired the engine in-house, found the problem, advised all its repair facilities and advised GE Aircraft Engines (GEAC). However, according to Scarfo, GEAC did not tell other operators about the mis-named part. "They assumed repair stations would make the same discovery," he said. This remark prompted a strong reaction from Board members. "I find that remarkable," declared Vice Chairman Robert Francis.

"Frightening is the word," muttered Board Member George Black.

Melted copper

Despite the 1997 repair, a fire resulted on Flight 574, causing the cockpit light to illuminate. The fierce fire also burned through electrical wiring near the engine. The engine features two redundant fire warning "loops," designated "A" and "B." When sensors detect temperatures in excess of 500 degrees F, a fire alarm is triggered. However, The signals from Loop A and B must agree to generate a fire alarm. In this case, the signal carried by Loop B was broken. "With this loss, the system thought it was a loop fault and de-activated the fire warning system," according to Scarfo.

At this point, the crew had executed the first action on the ECAM checklist, pulling the throttle to idle.

The A300 emergency procedures manual states, "The fire warning may be caused by a hot bleed air leak. In this case, the fire warning may disappear after setting the throttle lever to idle."

Had the crew continued with the ECAM checklist, they would have found the faulty loop.

Rather, when the fire light went out, Capt. Graham ordered First Officer Szot to go to the abbreviated checklist on a plastic-covered card, which called for them to set the fuel lever to the off position. The fire warning light went off, and they did not see the need to execute step 3, pulling the fire handle to discharge Halon into the engine.

According to Yves Benoit, director of flight safety for Airbus Industrie, the logic has since been changed. Now, he said, "When a loop is declared faulty, it is not considered as part of the logic."

"The remaining loop will keep the (fire warning) light lit," he said in a telephone interview. Also, in April of 1999 Airbus advised operators via a change in the Flight Crew Operating Manual (FCOM) to wait 15 seconds after retarding the throttle before cutting the fuel switch to assess if the problem is a blown bleed air duct or a continuing engine fire.

Training and documentation issues

There are also training implications in this case. Capt. Graham told investigators that in training "the fire lights don't go out," thereby implying that crews are not presented with a case where the fire light does go out.