Figure: The accident aircraft flying over Okutama, from Aircraft Accident Investigation Commission Report, Photo 124.

Eddie Sez:

I started flying Boeing 747's the year of this mishap and the "why" was a mystery for a while but the community banded together and immediately thought they understood "how" to survive the circumstances. Of course they were wrong. So let's tackle the "how" before we get to the "why," because the "why" also explains the reason the "how" doesn't work. Confused? Good, now you are in the right mind-set to tackle this case study.

How

The flight controls on the Boeing 747 are hydraulically powered with no mechanical backup. The airplane had a great reputation and to the date of this mishap, there had never been a total failure of the flight control system. Each engine drives a separate hydraulic system, AND each control surface has two hydraulic sources, AND each axis has multiple surfaces, even the rudders. The two rudders were each powered by two hydraulic systems, a total of four systems. What we didn't realize was that all four lines had a common path to the tail.

A photo of the crippled airplane shows it was missing its tail and it became apparent the damage took enough hydraulic plumbing to render every flight control on the airplane useless. (Boeing has since installed fuse plugs upstream of this area so a similar rupture will not result in the same loss.)

We in the Boeing 747 world realized our four engines below swept wings gave us the ability to control roll and pitch with engines alone. You could vary left engine thrust against right engine thrust for roll as well as inboard thrust versus outboard thrust for pitch. So we smugly believed we would have done better than these hapless Japan Air Lines pilots, even though simulator test proved getting the airplane onto a runway was extraordinarily difficult.

But what we didn't realize is that without the vertical fin the airplane was highly susceptible to Dutch roll which is normally confronted by an electronic yaw damper or manually through the ailerons. It would be next to impossible to dampen Dutch roll with differential thrust. Add to all that, the aircraft so oscillating would also enter phugoid cycles, that is, it would pitch up and lose speed causing the nose to drop and gain speed, which then cause a pitch up and so on. This is next to impossible to dampen without an elevator. So these pilots did a very good job controlling the airplane for as long as they did.

Why

The airplane suffered a tail strike in 1978, damaging much of the airplane, including the aft pressure bulkhead. Japan Air Lines hired Boeing to make the repairs. One of the repairs didn't go as planned: a manufactured replacement section of the aft pressure bulkhead did not extend far enough to allow a double row of rivets. This could have been the result of other undetected deformations to the structure or just the way the airplane was supported structurally during the repair. A Boeing engineer devised a work around that involved a metal splice wide enough to join sections with a double row of rivets.

The actual repair, however, used a single row of rivets. This is believed to have weakened the connection by 70 percent.

The airplane flew for another seven years as the metal of the splice and joined sections started to crack and finally gave way in 1985.

When the aft bulkhead gave way, pressurized air entered the vertical fin causing the skin to rupture and peel away. This caused the top of the vertical fin to separate, taking both rudders with it and venting all four hydraulic systems. The flight controls lost all power and the airplane was momentarily controllable using differential thrust. But without the vertical fin the airplane's Dutch roll and phugoid oscillations made the airplane uncontrollable.

So I would contend that are not any flight lessons here at all, the ruptured bulkhead turned the airplane into a uncontrollable projectile. The lessons here are for mechanics to follow repair manuals to the letter, question any deviations from those manuals, and to show up for work well rested and alert. The lessons go deeper than that, see: Abnormal Procedures / Maintenance Malpractice.

What follows comes from the references shown below and my comments shown in blue.


Accident Report


Narrative

Seven Years Before

Figure: Instruction issued for correction and actual work of splice, from Aircraft Accident Investigation Commission Report, Attachment 1, figure 3.

[Aircraft Accident Investigation Commission Report, Attachment 1]

  • The aircraft, when landing at Osaka International Airport at approximately 1501 hours June 2, 1978 as JAL scheduled flight number 11S (Tokyo to Osaka), was substantially damaged by contact of its lower part of the aft fuselage with the runway, but no fire occurred.

  • The aircraft was inspected in accordance with part of the Hard Landing Inspection Phase 1 and 2 in JAL's Maintenance Regulations, and a more detailed inspection was conducted on the damaged portions. Major damage [included . . . the] aft pressure bulkhead dome (4 to 8 o'clock) webs [which were] deformed.

  • In the work to attach the new lower half of the aft pressure bulkhead, firstly, the lower half of the bulkhead was attached to the airframe side at the Y chord portion, then. to connect it to the upper half of the aft pressure bulkhead on the airframe side, rivet holes were bored along the web edge of the lower half of the bulkhead, making them match the existing rivet holes of the upper half bulkhead.

  • In the inspection by the inspector of the repair team after the completion of the work above, edge margin of less than value specified in the structural repair manual was found around the rivet holes for almost all area of L18 joint on the left side of the lower half of the aft pressure bulkhead.
  • There should have been enough metal overlapping each section to allow two rows of rivets, spaced an inch apart, to create a strong, two-row rivetted joint. For some reason there wasn't enough overlap. (The reasons are discussed below, under Analysis.)

  • To compensate for the discrepancy above, a rework disposition was given from the engineer of the repair team that a splice plate be inserted. as shown on the left side of Attached figure-3 of Attachment 1, between webs insufficient in edge margin.
  • The splice is the shaded metal. If you look on the left side of the figure, notice there are two rivets going through the upper dome and the splice, and the lower dome and the splice.

  • Irrespective of such rework instructions having been issued, in the actual repair work, a short splice plate and a filler were used as shown on the right side of the attached figure above, instead of a single splice plate, with the result that the connection for two bays between the 1st strap and the 3rd strap on the left side became a one-row riveted connection, instead of the two-row riveted web connection to be made between the upper half and the lower half of the aft pressure bulkhead. The splice plate and the filler were manufactured from the removed old bulkhead.
  • The work actually performed ended up with a narrower splice that has rivets going through the lower dome and splice in two rows (properly) but only once through the upper dome and the splice (improperly). A second splice is added that doesn't add any structural strengthening at all.

  • This work was carried out on June 26, and completion inspection by the inspector of the repair team on June 27, but the inspector could not find the above results.
  • We have no idea why this work was done this way; it could have been a mistake followed by an effort to disguise the mistake, or perhaps it was just a misunderstanding. Whatever the reason, the weaker joint went undetected. The airplane flew for the next eight years, each pressurization cycle theoretically weakening the joint until cracks developed and then finally gave way.

The Day of the Crash

[Aircraft Accident Investigation Commission Report, ¶2.1]


Analysis

Figure: Damage to aft pressure bulkhead, from Aircraft Accident Investigation Commission Report, Figure 32.

[Aircraft Accident Investigation Commission Report, ¶2.1.5]

[Aircraft Accident Investigation Commission Report, ¶3.1]

[Aircraft Accident Investigation Commission Report, ¶3.1.7]

[Aircraft Accident Investigation Commission Report, ¶3.2.2]


Probable Cause

[Aircraft Accident Investigation Commission Report, ¶4.2]


See Also:

Abnormal Procedures / Maintenance Malpractice

Technical / Fault Tolerance


References

(Japan) Aircraft Accident Investigation Commission, Ministry of Transport, Japan Air Lines Co., Ltd., Boeing 747 SR-100., JA8119, Gunma Prefecture, Japan, Tentative Translation from Original in Japanese, August 12, 1985