Delta Air Lines Flight 1288

A scheduled departure for Atlanta

Delta Air Lines flight 1288 was preparing to leave Pensacola Regional Airport for Atlanta on 6 July 1996. Its McDonnell Douglas MD-88, N927DA, carried 137 passengers and five crew members. The passengers included two Delta employees travelling without charge, one in the cockpit jumpseat and one in an aft flight attendant jumpseat. The departure took place in daylight with visual weather conditions. The aircraft had two Pratt & Whitney JT8D-219 engines mounted beside the rear fuselage.

During the preflight inspection, which he began at 13:30, the first officer noticed two or three drops of oil on the nose of the left engine and two rivets missing from an outboard section of the left wing. He discussed both observations with the captain. The captain judged the aircraft airworthy and elected to depart without contacting maintenance. The engines started normally, and the crew reported no vibration while taxiing. The NTSB later found that the oil and missing rivets had no role in the fan-hub failure, but that Delta's instructions for reporting maintenance discrepancies required clarification.

At 14:23 the tower cleared the flight to take off on runway 17. The first officer was the pilot flying. He advanced the throttles and called for the autothrottles when engine pressure ratio reached 1.35. The throttles continued advancing automatically as the aircraft accelerated.

Failure at about 40 knots

At about 14:24 the crew heard a loud bang, and cockpit lighting and instrumentation went out. Passengers and flight attendants in the rear cabin experienced a blast-like concussion. The left engine's front compressor fan hub had fractured. High-energy pieces of the hub and fan assembly escaped the engine casing and struck the fuselage beside the aft passenger seats.

The captain took control, moved both throttles to idle and applied manual braking. He used neither reverse thrust nor the ground spoilers. The aircraft stopped on the runway about 1,350 feet from the threshold. The flight data recorder's last recorded airspeed was 39.75 knots, after the engines had increased to an engine pressure ratio of 1.9 over about ten seconds. Both flight recorders lost power at the failure.

The fragments caused extensive cabin and electrical damage. Investigators documented 16 holes, punctures or tears in the left fuselage skin and seven on the right side. Damage was concentrated near row 37. Of 154 wires in a bundle along the right side, 146 were severed, including four associated with the right generator's differential protection circuit. The left engine failure and the damage to the right electrical system interrupted normal aircraft power.

The first officer initially could reach neither the tower nor the flight attendants by radio or interphone. The pilots selected emergency power and declared an emergency at 14:25. Selected instruments and communications became available from the battery supply.

Different conditions at opposite ends of the cabin

In the aft cabin, flight attendants saw severe damage and injured passengers. Their attempts to contact the cockpit through the interphone were unsuccessful, and they began an evacuation after the aircraft stopped. Three passengers and an infant used the tailcone slide. An attendant opened the left rear service door and inflated its slide, then saw fire near the front of the left engine cowling. She blocked that exit and directed passengers forward.

Passengers opened all four overwing exits. One man told investigators that he had opened an exit while the aircraft was still moving, under pressure from other passengers. He later jumped from the leading edge of the left wing after seeing flames. About 30 passengers left through the tailcone or overwing exits. The NTSB found that the passenger safety card gave insufficiently specific guidance about when emergency exits should be opened.

The cockpit crew and forward flight attendant initially had little information about the aft cabin. With no cockpit fire indication, the captain instructed the forward attendant to keep passengers aboard. When the first officer walked aft, he saw open exits and heard engine noise. He returned to tell the captain to shut down the engines, and the captain moved both fuel control levers off. The NTSB found that the captain and aft flight attendants made appropriate decisions with the information each had. It also found that the captain shut down the engines promptly when he learned of the conditions in the rear.

Injuries and emergency response

The two passengers who died had been sitting in the window and aisle seats on the left side of row 37, beside the failed engine. A passenger in the aisle seat on the right side of the same row suffered serious head and other injuries from debris. Another passenger suffered a serious ankle fracture after jumping from the front of the left wing. Three passengers received minor injuries during evacuation. All five crew members survived without injury; 140 of the 142 occupants survived.

A short fire burned around the left engine cowling. Ground witnesses reported seeing flames for approximately 20 seconds. Firefighters arriving at 14:27 saw no flames or smoke, although one smelled smoke, and they applied extinguishing agent to the engine. Investigators found fire damage on the exterior of the cowling. There was no cabin fire.

The captain requested medical assistance at 14:27. Emergency medical technicians and firefighters reached the aircraft, and a treatment area was established beside the runway. The most seriously injured passenger was taken to a local hospital by helicopter at 14:42. Emergency personnel used the aft stairs to remove injured passengers. Other passengers remained aboard until suitable portable stairs arrived, approximately 25 to 30 minutes after the failure, and then travelled to the terminal by bus.

The hub's manufacturing and service history

The failed fan hub had been forged from titanium alloy by Ladish and machined, finished and inspected by Volvo Aero Corporation for Pratt & Whitney in January 1989. It held 34 fan blades. Twenty-four tie rods joined it to the next compressor disk, passing through holes in the hub rim. Volvo made each tie-rod hole by drilling, two boring operations and honing to the final diameter.

The hub entered service in April 1990. Delta removed it in January 1992 after foreign-object damage to fan blades and performed a visual inspection. It subsequently went into another engine. In October 1995 it received fluorescent penetrant and visual inspections during heavy maintenance at Delta's Atlanta facility. The engine fitted with this hub was installed on N927DA on 1 January 1996, following balancing and a test-cell run in which vibration measurements were within the manufacturer's limits.

At Pensacola the hub had accumulated 16,542 hours and 13,835 cycles, including 1,142 cycles since its October 1995 inspection. Its service-life limit was 20,000 cycles. Investigators recovered three major hub pieces. One was found 2,400 feet to the right of the runway centreline in an athletic field; another had penetrated the cabin and lodged on the right side above row 37. The fracture surfaces contained evidence of progressive fatigue followed by overstress separation.

Drilling damage beneath an accepted surface

Laboratory examination traced fatigue growth to two origins inside a tie-rod hole. At those origins, the wall surface was darker than surrounding metal, contained boring marks and showed small parallel ladder cracks. Beneath it, metallurgists found zones of altered, deformed and hardened microstructure. The evidence indicated intense local heating and deformation during manufacture.

Volvo conducted more than 300 drilling tests after the accident. Some tests deliberately promoted tool failure, chip accumulation and high local heat. One test hole developed features similar to the accident hub. The NTSB concluded that drill breakage or breakdown, localized loss of coolant and chip packing had created the altered metal and ladder cracks. About 12,887 fatigue striations were identified on the recovered fracture, roughly corresponding to the hub's flight cycles. The Board found that fatigue growth began almost immediately after the hub entered service.

The damage extended deeper than Pratt & Whitney had anticipated when accepting the drilling process. Subsequent boring and honing removed about 0.0185 inch radially, while the altered microstructure extended about 0.024 inch from the original drilled surface. Damaged material therefore remained in the finished hole. A manufacturing inspector had recorded marks in that hole, but the hub passed the blue etch anodize and visual inspections. The NTSB found that its appearance did not match the defect templates then available. After the accident, Pratt & Whitney added four templates illustrating additional abnormal conditions.

A detectable crack passes the last inspection

From the fracture evidence, the NTSB estimated that the crack at the October 1995 inspection extended about 0.46 inch across the aft face of the hub and 0.90 inch along the hole wall. Its combined surface length was about 1.36 inches. The Board found it large enough for detection. The FAA's review cited reliability data indicating detection probability and confidence greater than 95 percent for a visible crack of that size.

Fluorescent penetrant inspection uses a dye that enters surface-breaking cracks. After excess dye is removed, developer draws penetrant back to the surface, producing an indication under ultraviolet light. Investigators examined every part of that process at Delta, including cleaning and drying, developer coverage, control of development times, handling and the inspector's search. They found weaknesses that could reduce the visibility of indications or allow an inspected area to be missed. The Board could not identify a single reason for this crack's nondetection: inadequate cleaning or processing, inspector performance, or a combination could have caused it.

The inspector had completed the required training, passed vision examinations and was described as competent. The investigation examined how systematic visual search could be interrupted when a hub was repositioned or an indication required diagnosis. It also identified the potential effect of a low expectation of finding cracks. The hub had passed an inspection program in which the opportunities for a second check before failure were insufficient.

Final findings and safety recommendations

The NTSB adopted its final report on 13 January 1998. It identified the probable cause as the fan-hub fracture following Delta's failure to detect a detectable fatigue crack. The crack originated in altered microstructure created by Volvo's drilling process and missed during manufacture. The Board identified the lack of sufficient redundancy in the in-service inspection program as a contributing factor.

Recommendations to the FAA addressed manufacturing controls, the limitations of etch inspections, and common standards for preparing and inspecting critical rotating parts. The Board called for better tracking of development times and inspected areas, research into inspection duration, and proficiency testing using representative parts. It recommended inspection intervals that would give a crack missed on one occasion another inspection before it could grow to failure.

The report also recorded changes Delta had made during the investigation: daily hot-rinse temperature checks, more focused developer application, a status board for development times, and part-specific inspection technique sheets. Cabin recommendations called for independently powered evacuation alarms operable at every crew station and cockpit indications of open exits connected to emergency power. The Board also sought clearer operational guidance on maintenance irregularities before departure.