Accident Overview

Photo of Delta Air Lines McDonnell Douglas MD-88
Photo of Delta Air Lines McDonnell Douglas MD-88
Photo copyright Paul Robbins - used with permission

Accident Details

On July 6, 1996, at 1424 central daylight time, a McDonnell Douglas MD-88, N927DA, operated by Delta Air Lines Inc. as Flight 1288, experienced an engine failure during the initial part of its takeoff roll on runway 17 at Pensacola Regional Airport (PNS) in Pensacola, Florida. Flight 1288 was cleared for takeoff at 1423. The first officer, who was the pilot flying, advanced the throttles and called for the autothrottles to be set. The throttles were advancing in the autothrottle mode when the flightcrew heard a "loud bang," followed by the loss of cockpit lighting and instrumentation. Passengers and flight attendants in the rear of the cabin described experiencing a "concussion or blast-like sensation." The captain took control of the airplane and retarded both throttles to idle. He applied manual brakes and brought the airplane to a gradual stop on the runway. There were no cockpit indications or warnings of fire. The flight data recorder indicated that the airplane had reached a speed of about 40 knots when the left engine failed. Uncontained engine debris from the front compressor front fan hub of the No. 1 (left) engine penetrated the aft fuselage. Two passengers were killed, and two others were seriously injured.

Photo of damage to No. 1 engine and left fuselage
Photo of damage to No. 1 engine and left fuselage
- NTSB Docket Photo

Damage to Airplane and Engine

The aft fuselage and interior of the airplane in the vicinity of the No. 1 engine were substantially damaged by debris from the engine. A total of 16 holes, punctures, or tears were documented in the left fuselage skin. Several large holes and tears were found adjacent to seat row 37. Seven exit holes, punctures, and tears were documented in the right fuselage skin just forward of row 37. Most of the wires in the wire bundle located on the right side of the fuselage were severed. Of the 154 wires in the bundle, 146 had been severed. No evidence of penetrations existed below the floor level on either side of the fuselage. The cabin interior was substantially damaged near seat row 37, next to the left engine. Debris from the left engine's fan hub and fan blades had penetrated the left cabin wall and overhead bin vertically from the passenger window, through the overhead bin, and ceiling panel. Engine fan components had also pierced the side and ceiling of the right cabin wall. The No. 1 engine, a Pratt & Whitney JT8D-219 turbofan, was heavily damaged.

Airplane Engines

The airplane was equipped with two Pratt & Whitney JT8D-219 turbofan engines. The JT8D-200 series engine is an axial-flow front turbofan with a 14-stage split compressor, a 9-can combustion chamber, and a split, 4-stage reaction impulse turbine. The No. 1 (left) engine, S/N 726984, had a total operating time of 7,371.7 hours and 5,905 operating cycles since new. Delta was the original operator of the engine. The engine had been installed on the accident airplane on January 1, 1996 and had since then accumulated 1,528 hours and 1,142 cycles. It had been removed from another Delta airplane on December 21, 1995, following a report of "smoke in cabin." The problem had been identified as an oil leak in the compressor section, and a carbon seal was replaced.

Cross section of Pratt & Whitney's JT8D-200 series engine
Cross section of Pratt & Whitney's JT8D-200 series engine

Left Engine Compressor Fan Hub

Diagram of Pratt & Whitney JT8D-200 Series Engine Fan Hub
Diagram of Pratt & Whitney JT8D-200 Series Engine Fan Hub

The left engine's fan hub, S/N R32971, had a total time of 16,542 hours and 13,835 cycles at the time of the accident. At the time of the engine's installation on the accident airplane in January 1996, the hub had accumulated 12,693 cycles. The titanium fan hub was forged by Ladish Company in Milwaukee, Wisconsin, and machined, finished, and inspected for Pratt & Whitney by Volvo Aero Corporation in Trollhattan, Sweden, in January 1989, according to Pratt & Whitney records. The service life of this type of fan hub is limited to 20,000 cycles. The hub consisted of a disk forging that held 34 fan blades in dovetail (interlocking joint) slots. The aft end of the hub attached to the stage 1.5 disk with 24 tierods that passed through .5175-inch diameter tierod holes drilled in the hub web just inside of the dovetail slots. The 2.91-inch deep tierod holes were located around the circumference of the hub bore and alternated with 24 smaller diameter stress redistribution (SR) holes. The fan hub was forged from a titanium-based alloy containing 6 percent aluminum and 4 percent vanadium (Titanium 6AL-4V).

Volvo's Tierod Hole Drilling Process

Click here to view Delta Flight 1288 Accident Animation. (Note: It is recommended that this animation be viewed prior to reading the remainder of the Overview.) 

The tierod holes in the accident hub were created using a four-step process: the hole was drilled, bored in two steps, and then honed. Three tools were used to create tierod holes: a drill, a boring bar, and a hone. The 24 tierod and SR holes on the accident hub were drilled using a computer controlled coolant channel drill, which was designed to use coolant streams to flush titanium chips from the hole during a "one-pass" or single-plunge drilling process. The coolant channel drill used on the accident hub was a conventional pattern twist drill with tungsten carbide cutting-edge inserts. The .480 inch diameter drill had an internal conduit for coolant to flow down the drill core and enter the hole being drilled behind two carbide cutting edges. The coolant served as a lubricant and flushing agent to remove chips from the hole. The flushing was critical because titanium chips can be easily compacted in hole-drill interface areas, and can cause friction and elevated temperatures in holes.

Photo of Coolant Channel Drill (.480 inch diameter)
Photo of Coolant Channel Drill (.480 inch diameter)
Note coolant hole in drill tip

The previous drilling process used for the manufacture of fan hubs had used a standard drill that was removed every .20 inches during drilling. This is described as a "pecking process", the purpose being to purge chips from the hole. This is accomplished by removing the drill tool and applying high pressure coolant to the work interface. The fundamental reason for using a pecking hole drilling operation for the JT8D-200 Series fan hub is due to the disc having a large cross section of over three inches. This results in a very high hole length to diameter ratio of 6 (3 in / .5 in = 6) which has historically been shown as highly prone to chip congestion.

Subsequent to the drilling operation which drills the hole to a .480-inch diameter, the hole was enlarged by a boring operation. The first boring step enlarged the hole to .508 inch diameter, using the same type of spindle that held the drill. A second boring step enlarged the hole to .516 inch. The holes were then finished on a second machine that uses a boron nitride hone with a lubricant or honing oil, resulting in a finished diameter of .5175 inch. The difference between the diameter of the coolant channel drill and the finished hole was about .0375 inch. The total radial depth of the material removed after drilling was about .0188 inch.

Metallurgical Effects of Abusive Machining

Photo of fractured fan hub, fractured radially in two places
Photo of fractured fan hub, fractured radially in two places

The fractured components of the accident fan hub were examined in the National Transportation Safety Board's materials laboratory. The fan hub had fractured radially in two places. One of the radial fractures contained a fatigue crack that originated at two locations on the inboard side of a tierod hole.

Metallurgical examination of the surface of the hole wall revealed an area in which the surface finish was darker than the surrounding area at each fracture origin. The hole surface in the darker areas showed evidence of circumferential machining marks consistent with marks that would be left by the boring operation performed during the part's manufacture. There was no indication of honing in the darker areas. The remainder of the hole wall surface outside the darkened surface finish areas showed a cross-hatched pattern consistent with marks that would be left by the honing operation. Magnified examination of the hole wall in the darker areas also showed numerous small parallel surface cracks (ladder cracks) aligned with the longitudinal axis of the hole.

Magnified view of tierod hole wall
Magnified view of tierod hole wall.
View depicts numerous small parallel surface cracks
(ladder cracks) aligned with the longitudinal axis of the hole.

Magnified view of grain structure at edge of tierod hole
Magnified view of grain structure at edge of tierod hole.
Normal grain structure is present away from hole with
distorted grain structure located at the hole edge.

A scanning electron microscope (SEM) examination of the fracture face in the origin areas showed evidence of overstress to a depth of about 0.002 inch adjacent to the hole wall. The overstress fracture region was followed by an area about 0.006 inch deep that contained fracture features consistent with a fast-propagating fatigue crack. From a depth of 0.006 inch to the end of the fatigue region, striations were found consistent with a slower propagating fatigue crack. Approximately 13,000 fatigue striations were found in the fatigue fracture region, roughly equivalent to the number of the hub's flight cycles.

Cross section of fan hub depicting fatigue crack origins and crack propagation
Cross section of fan hub depicting fatigue crack origins and crack propagation

Metallurgical examination of the cross section of one of the fatigue origins showed three zones of altered microstructure adjacent to the hole wall surface corresponding to the darkened surface finish areas on the hole wall. The microstructural zone closest to the hole wall surface was about 0.002 inch deep (the same as the overstress depth). This zone was heavily layered with recrystallized alpha grains, indicating that the surface temperature had reached at least 1200°F, which is the minimum recrystallization temperature for titanium.

The second zone of altered microstructure was from 0.002 inch to 0.006 inch from the wall surface (about 0.004 inch thick). The microstructure in this zone consisted of heavily deformed alpha and beta grains elongated parallel to the surface. Below this area, to a depth of about 0.010 inch from the surface of the hole, was a third zone where the microstructure was distorted in a curved pattern, consistent with the metal having been deformed by bearing pressures from a rotating tool during the manufacturing process.

Pratt & Whitney Engineering Source Approvals

Illustration of coolant channel drill and chip congestion
Illustration of coolant channel drill and chip congestion
from Delta Flight 1288 Accident Animation

Pratt & Whitney Engineering Source Approval (ESA) is the manufacturing engineering requirement to verify that a component manufactured to a specific process and drawing requirement meets design integrity, and ultimately assures that no critical process is detrimental to achieving this integrity. ESA requirements may include: part cutups, metallurgical examinations, manufacturing sequence sheets, Critical Process technical plans, first article and selected process validation tests. ESA is a multi-disciplined function that requires a skill mix that includes Design Engineering, Manufacturing Engineering, Materials Engineering and Quality Assurance.

Pratt & Whitney records indicated approval of Volvo's request to use the coolant channel drill (rather than a standard drill) on February 11, 1988. The Pratt & Whitney Process Approval Record (PAR) noted that the request was characterized as an "insignificant change." The change was approved by the ESA function without metallurgical evaluation requirements because changes in drilling operations were classified as "insignificant" by Pratt & Whitney since subsequent material removal (in the boring and honing phases) had been accomplished to a depth of at least .010 inch. Pratt & Whitney indicated that generally, in the machining of titanium holes, if greater than ten thousandths were removed in subsequent operations, any potential adverse effects from the drilling process would be eliminated.

Special FAA Audit of Pratt & Whitney

Following the accident, the FAA conducted a "special quality system audit" at Pratt & Whitney from July 29 through August 2, 1996. The special audit noted that Pratt & Whitney's ESA function requires that a PAR be issued for "significant changes." The FAA audit report noted that "significant changes include new tooling, change in sequence of operations, a change in any process which could result in cracking, or change of process location within a plant."

Noting Volvo's request for the drill process change, the FAA audit stated that "several PARs were observed in which tooling was changed and/or operation sequence, and that these approvals were classified as insignificant." Subsequent to this audit, Pratt & Whitney changed this procedure, to require that all changes related to hole drilling be considered "significant" and reviewed according to requirements for that category of change.

Special FAA Audit of Volvo Flygmotor

Photo of Engine Inlet on Runway – NTSB Docket Photo
Photo of Engine Inlet on Runway – NTSB Docket Photo

Following the accident, the FAA conducted a "special quality system audit" at Volvo from August 13 through August 16, 1996. The audit issued a report that included the following issues:

  • At the time the accident hub was produced, the JT8D-219 fan hub scrap rate was approximately 11%. This scrap rate was considered relatively high within the turbine engine manufacturing industry.
  • Fan hub tiebolt hole drilling was considered an "insignificant" process, therefore no metallurgical evaluation was performed for machined feature microstructure. Metallurgical evaluation was performed for the boring and honing processes only.
  • An ESA review was conducted of all change activity associated with the tiebolt hole drilling process. The review identified 12 changes in tooling or processes from 1984 to 1996.
  • All 12 changes were classified as "insignificant" and thus never received a high level of engineering scrutiny.
  • A detailed review of the 12 changes revealed a drill process change to a coolant channel drill that resulted in substantial deposits, smearing and imbedding of titanium onto the drill.
  • It was concluded that the coolant channel drill process combined with localized loss of coolant, and chip packing overheated the material and created the altered microstructure and ladder cracking in the accident fan hub.
  • Per ESA requirements for tierod holes, Volvo was required to submit metallurgical samples of material machined with sharp tooling as well as samples machined just prior to tool resharpening. Neither Pratt & Whitney nor Volvo could provide evidence of metallurgical samples of material machined just prior to tool resharpening, or data for tool replacement/resharpening intervals. The ESA requirements for part sampling were intended to assure continuous manufacturing integrity. As tooling becomes worn, the sampling process is considered vital to confirmation of manufacturing quality and consistency.
  • Volvo manufacturing records indicated that there were Blue Etch Anodize inspection "indications" at the exact location of the two fatigue crack origins that were found during the post-accident metallurgical investigation. According to Volvo, the indications did not match any of the templates used by Volvo, at the time, to identify anomalies, and were therefore not considered blue etch indications. These two indications were strictly an observation made by the inspector regarding the hole surface. There was no notation of a Blue Etch Anodize indication of a defect in Volvo manufacturing records relating to the accident hub. In the subsequent visual inspection, the inspector used a sharp point stylus to check for geometric anomalies, and none were found. It was determined during the audit that a stylus would not have been effective in evaluating a blue etch indication. Investigators concluded that a geometric inspection method, such as using a stylus, would not be effective in identifying metallurgical anomalies, and that the fan hub should have been subjected to further metallurgical evaluation

Post-Manufacturing Inspection

Investigators identified several post manufacturing inspection processes for completed hubs that were performed at Volvo. These inspections included: dimensional and visual inspections, Fluorescent Penetrant Inspection (FPI) and Blue Etch Anodize (BEA) inspection. The dimensional inspection checks the location, concentricity, diameter, and perpendicularity of holes. The visual inspection examines the surface finish, and looks for evidence of residual machine marks. The FPI checks the surface of the material for physical defects such as cracks, voids, or metal porosity. The BEA inspection process, which is unique to titanium, visually inspects the surface after anodizing (the surface is electro-chemically oxidized), to check for anomalies associated with microstructure changes in the metal.

Photo of separated fan hub from accident
Photo of separated fan hub from accident
Note: hub failed through tierod holes

The investigation determined that the accident fan hub had two documented nonconformance notations, as it progressed through the manufacturing process:

  1. Following the drilling process, the drill operator noted two holes that contained "chatter marks". These "chatter marks" were no longer noted after completion of boring and honing.
  2. The BEA inspector further noted, during his inspection of the accident hub, manufacturing marks in a hole located 180 degrees relative to the serial number marking.

This was the same tierod hole analyzed by the safety board after the accident. Manufacturing records provided no further description of the accident hub manufacturing marks, or where they were located in the hole. Subsequent inspections determined that the fan hub met Pratt & Whitney's manufacturing criteria, and it was sent to Pratt & Whitney for installation.

The investigation determined that inspection notations had been made to alert visual inspectors to the surface condition. However, the indication did not match any of the templates used by Volvo to identify anomalies, and was not characterized as a blue etch indication, but was rather an observation made as to the surface condition. Finally, manufacturing records did not contain a BEA notation of a defect on the accident hub.

In-Service Inspections

The investigation established that, as a "safe life" or "life-limited" part, the fan hub was designed and certified to operate safely for its total design life of 20,000 cycles without inspection, unless it was removed from the engine. Once an engine entered operational service, Fluorescent Penetrant, and visual inspections were conducted on fan hubs at the Delta overhaul facility only if they were removed during engine overhaul or disassembly.

Fluorescent Penetrant Inspection of a turbine engine disk
Fluorescent Penetrant Inspection of a turbine engine disk

Following the accident, the FAA conducted a technical review of Delta's FPI process at its Atlanta, Georgia, maintenance facility.

In the report of the review team's findings, "Technical Review of Fluorescent Penetrant Process Delta Air Lines Inc.," it was concluded that, based on reliability data collected by the Nondestructive Testing Information Analysis Center (NTIAC), "a crack of this size (a total surface length of 1.36 inch on the accident hub) should be detectable with a probability of detection (POD) and confidence level both exceeding 95 percent." Data compiled by the NTIAC also indicated that the minimum reliable detection length for FPI is about 0.10 inch.

The findings of the FAA review included:

  1. "There is no assurance that the material received by the nondestructive inspection organization for FPI processing was clean enough for an adequate FPI.

  2. Engine part cleaning personnel receive on-the-job-training (OJT), with no formal classroom training. The team noted that sensitivity to the criticality of the engine components and the end purpose for which these components were being cleaned was not provided as part of the OJT (critical rotating versus static, general visual inspection versus nondestructive inspection).
  3. The solvent on the production floor the morning of August 14, 1996 was badly contaminated with fluorescing material.
  4. Visible trash and debris were under the transport rollers utilized on the FPI line. Since there are no protective covers over the tanks containing the FPI process materials, similar trash and debris is expected in the FPI material.
  5. The transport rings utilized for parts holding during the FPI process became easily contaminated with fluorescent material. One inspector was noted having a difficult time inspecting the inside of a hole because of the high fluorescent background from the transport ring visible through the hole. He tried shielding the ring from view with his glove, but it also was contaminated with fluorescent material.
  6. One inspector was noted touching the component to be inspected, and smearing the inspection area, before inspecting it.
  7. There appears to be no uniform way of handling and indexing components during evaluation in the inspection booth."
Detection of cracks under ultraviolet light using FPI (left), FPI of turbine engine blades (right)
Detection of cracks under ultraviolet light using Fluorescent Penetrant Inspection (left),
Fluorescent Penetrant Inspection of turbine engine blades (right)

The FAA report noted that during, and following the inspection team's on-site evaluation, Delta initiated positive and responsive actions to the team's recommendations. Personnel involved in inspections now receive enhanced training related to inspection preparation, especially of critical parts. Also, standards were developed with engine manufacturers for cleaning of parts prior to inspection.

Back to top