Skin Rupture Explodes Myth That Safety Is Under Control
Up on the top of the fuselage, tiny hairline cracks were creeping between the rivets; suddenly, the cracks connected and a 5-foot section blew open with the sound of a thunderclap as air in the fuselage rushed out. The emergency oxygen masks dropped down to the startled and frightened passengers.
The manner in which the cracks suddenly connected together, growing a quarter-inch or more in fractions of a second, is known in engineering parlance as the “fast fracture” phenomenon. It has been seen before, notably in an Aloha Airlines B737 in 1988, when cracks, observed outside the fuselage by passengers while boarding, connected and caused complete loss of structural integrity. The walls and roof of a section of fuselage ripped off. Stewardess Clarabelle Lansing, serving passengers at the time, was sucked out of the airplane. Her body was never found.
In the case of Southwest Airlines flight 812 on 1 April 2011, while the flight attendants were taking drink orders, the fuselage of the 15-year old B737-300 ripped open at a lap joint where fuselage panels are held together by three rows of rivets. Fortunately, no one was ejected from the airplane. Emergency oxygen masks dropped from their overhead compartments and the pilot began a rapid descent from 34,000 feet to a lower altitude where the ambient air is dense enough to sustain life. The 118 people aboard scrambled to don the masks. One passenger said, “People were dropping” (fainting). At the cruising altitude of 34,000 feet, the so-called “time of useful consciousness” is on the order of 10-20 seconds, so one does not have much time to fiddle with the unfamiliar oxygen mask.
It is not known if the emergency oxygen had been de-activated in the lavatories, or if anyone was in the lavatory at the time pressurization was lost. The airlines had until 4 April 2011 to remove or exhaust the lavatory oxygen canisters. (See Aviation Safety Journal, “ ‘Huge Favor’ to Airlines Ordering Emergency Oxygen Removed From Lavatories”)
The airplane made an emergency landing at Yuma, AZ, and passengers clapped and cheered when the captain emerged from the cockpit.
The Aloha accident set in motion the Federal Aviation Administration’s (FAA) supplementary structural inspection program (SSIP) for older jets. They were to be inspected for signs of fuselage skin cracking and patched up. Jet transports that had accumulated 75% of their design life were subject to the SSIP requirements, which meant more than 75,000 flights. This particular Southwest jet was well short of that threshold, which would have required about 20 years of hard airline service.
The SSIP clearly did not work – and did not apply – to Southwest Airlines flight 812. Investigators with the National Transportation Safety Board (NTSB) pored over the plane, using various eddy-current and other instruments to chart the extent of cracking.
Robert Sumwalt, NTSB Member on scene, said there was evidence of extensive cracking in the area of the rupture. A 9-foot section of the fuselage, covering the area of the rupture, was removed from the airplane on 3 April 2011 and sent to NTSB headquarters for detailed analysis. The rupture occurred in a different area than on the Aloha B737 but involved the same joint design: three rows of rivets, each row about ¾ inch from the next, and the rivets along each row about the same distance of ¾-inch apart from one another. As in the case of the Aloha B737, the failure began in the lowest row of rivets, where the stresses are highest. In this case, crack stopper straps halted the failure at the rib-stringer intersection.
NTSB investigators also conducted inspections of other portions of the lap joint along the fuselage and found evidence of additional cracks.
Southwest Airlines cancelled 600 flights to inspect its remaining B737-300 aircraft. There are 288 B737-300s in U.S. registry, of which 79 are operated by Southwest. As of this writing, 21 have been inspected and three of the planes were found with small, subsurface cracks.
“What we saw with Flight 812 was a new and unknown issue,” said Mike Van de Ven, Southwest’s executive vice president and chief operating officer.
The “fast fracture” phenomenon is hardly “new and unknown” for older aircraft with highcycles. By high cycles is meant take-offs and landings; during each flight the fuselage is inflated like a balloon, and the flexing is akin to the repeated bending of a paper clip until it breaks. The lap joint features three rows of rivets to contain the stress, but periodic inspection for signs of cracking is essential.
The airplane underwent overhaul in March 2010. During this so-called D-check, the cabin seats, sidewalls and other appurtenances were removed to facilitate inspection of the aluminum structure for cracking. The records of that D-check will be closely scrutinized by the NTSB, as the cracking found on the accident airplane was clearly extensive enough to predate that overhaul of a year ago. Were cracks discovered? Were they repaired?
According to the dictates of the SSIP, crack growth is slow enough such that they can be missed entirely during one teardown – as long as crack growth is detected and rectified at the next overhaul. These D-checks occur at approximately four year intervals. If the cracks were detected during the last D-check for the Southwest B737-300 but not corrected, were Southwest maintainers seduced by optimistic predictions of crack growth? If the airplane was stripped to bare metal and the cracks were not discovered, why not?
If only visual inspections were conducted – as opposed to more advanced non-destructive testing (NDT) techniques – it is possible but not plausible that cracking was not detected. Of interest, NTSB investigators employed NDT on the aircraft on the ground at Yuma and immediately concluded that widespread fatigue cracking was present. Under the controlled conditions of an overhaul, Southwest technicians should have been able to detect the cracking.
Southwest’s maintenance has been under a cloud before. In March 2008 the Federal Aviation Administration (FAA) proposed a fine of $10.2 million against Southwest for operating 59,000 flights with airplanes out of compliance with a structural airworthiness directive (AD).
On 3 April 2008 the House Transportation and Infrastructure Committee held hearings on the alleged “cozy relationship” between Southwest Airlines and other carriers with the FAA. Then committee chairman Rep. James Oberstar (D-MN) complained to top Southwest executives at the hearing, “Southwest Airlines allowed 117 planes to fly without AD compliance … the most egregious lapse of safety that I’ve seen in 23 years.”
Despite Oberstar’s concern, in March 2009 the FAA announced that the $10.2 million fine had been reduced to $7.5 million on condition that Southwest would rewrite all its maintenance manuals.
On a per flight basis, the FAA originally proposed a penalty of $172.00 for each non-compliant flight, negotiated down to $127.00 per flight – less than the price that a passenger pays for a ticket on the carrier. A token proposed penalty was reduced further based on Southwest meeting certain minor demands.
This penalty did not arrest the structural problems at Southwest. On 13 July 2009, the very next month after the reduced fine was announced, a Southwest B737-300 experienced a rupture in its ceiling, forcing an emergency landing. The NTSB removed the ruptured component and after due metallurgical analysis it was determined that the probable cause was “Fuselage skin failure due to preexisting fatigue at a chemically milled step.”
A chemically milled step is one where metal is removed during manufacturing to minimize aircraft weight.
Boeing issued a service bulletin (SB) covering inspection of the area where the rupture occurred (on the roof, near the tail) and the FAA published an AD making the inspection mandatory.
From 2009 through 2010 Southwest filed 6 Service Difficulty Reports (SDRs) with the FAA recounting depressurization events. Below is the one for July 2009 involving a B737-300 (the same model as this most recent pressurization emergency):
“Declared emergency/diversion. Pressurization event at 33,000 feet. Masks dropped, emergency descent … Found fuselage cracked between BS [body station] 827 to BS 847, STR [stringer]. Repaired skin per SWA [engineering order].”
On 28 December 2010 the FAA published AD 2010-25-06 concerning structural problems for early model B737s (including the -300 variant). The summary is instructive:
“This AD requires repetitive inspections for cracking of certain fuselage frames and stub beams, and corrective actions if necessary … This AD results from reports of the detection of fatigue cracks at certain frame section, in addition to stub beam cracking, caused by high flight cycle stresses from both pressurization and maneuver loads … This reduced structural integrity can increase the loading in the fuselage skin, which will accelerate skin crack growth and could result in rapid decompression of the fuselage.”
The inspections must be performed within 4,500 flight cycles, or within 9,000 cycles after the previous inspection. Southwest complained to the FAA that “the specified threshold will pose a significant burden [as] half of its model 737-300 and -500 fleet will require an out-of-sequence maintenance visit …”
At 4 flights per day, the 4,500 cycle threshold implies a 3-year time to accomplish the inspection.
Meanwhile, in an apparent effort to demonstrate it is on top of the situation, the FAA announced on 4 April that it will issue an emergency AD requiring structural inspections of older B737 using NDT. Randy Babbitt, head of the FAA, said, “This action is designed to detect cracking in a specific part of the aircraft that cannot be spotted with visual inspection.”
Why cracking on the top of the fuselage is not amenable to visual inspection was not explained.
From a B737-300 diagram, it appears that the skin panel which ruptured open begins around station 663 and continues past station 706 and ends before station 727. The rupture is close to, but not exactly correlated with, the location cited in AD 2010-25-06. There is a lap joint at that location. It was corrosion, a substandard repair, scribing or a similar scrape in the surface of the aluminum which cracked from pressurization cycles.
As one aviation industry insider remarked: “I would like to hang anybody I see scraping lap joints with metallic or non-approved scrapers.”
This observation is not presented to suggest that such practice was condoned at Southwest Airlines. Rather, it illustrates the sensitivity of the metal skin to unauthorized abuse when flight cycles are measured in the thousands.
Given the extreme danger presented by even a scribe line, allowing 4,500 flights – much less 9,000 – seems overly generous. Whatever “significant burden” imposed on Southwest, that mandate seems lax in the extreme. Consider the ongoing out-of-sequence inspections, the 600 cancelled flights (at least), the likely lawsuits from the passengers aboard Flight 812, and the cancelled bookings from a frightened public, not to mention the loss of confidence in FAA oversight, then the “cost” of this incident progresses to erosion of confidence in the safety of air travel. This lost confidence is not simply restored by quickly fixing this particular problem.