Seven months after a titanium blade snapped inside the engine of a Southwest Airlines jet, causing an emergency landing in Philadelphia and the death of one passenger, a federal investigation has zeroed in on two central questions:
Why did the blade break? And why weren't its shards contained by a protective sheath that surrounded the engine, preventing further damage to the plane?
At a hearing last week before the National Transportation Safety Board, expert witnesses stopped short of identifying the cause of these two failures on the 737, but their testimony disclosed that the April 17 accident was years in the making. The NTSB is in the midst of an investigation into how the incident happened and how to ensure it won't happen again. By examining the wear patterns in the failed blade from Flight 1380, forensic experts determined that the initial crack had been present nearly six years earlier — undetected during a 2012 overhaul and inspection.
Inspectors at the time were not required to use ultrasound or other advanced equipment to detect possible cracks — only their eyes, aided by fluorescent dye to highlight flaws. But the crack was already there, testified Mark Habedank, a lead engineer at engine maker CFM International, a joint venture of GE Aviation and Safran Aircraft Engines of France.
Metallurgical experts were able to trace the crack's history by looking at "striations" — estimating how fast the crack had grown over the thousands of flights, almost like reading tree rings. "If we look at the striation count and go backward, it appears that during that inspection, the size of the defect was about 1/16 of an inch," Habedank said.
The engine blades were designed to last at least 100,000 flights, but the crack began to form at just 20,000 flights, and the blade broke after 32,000, he said.
The engine used on the Southwest flight, a CFM56-7B, is one of the CFM56 series of engines manufactured by a partnership between General Electric and the French company Safran Aircraft Engines and used by about 300 different airlines worldwide. The findings of the NTSB investigation could have far-reaching consequences for the airline industry.
Even before April 17, CFM officials had determined that the roots on this type of fan blade suffered excessive friction in flight, and recommended that they be removed and lubricated every 3,000 flights. Since the accident, such blades must now be lubricated every 1,600 flights, as well as undergo crack detection with either ultrasound or an eddy-current device, which uses electricity to detect abnormalities.
Eight more blades have been removed from service as the result of these stepped-up inspections, Habedank testified, drawing surprise from John DeLisi, director of the NTSB's office of aviation safety.
"So it's great that the inspection technique is pulling those parts from service, but that's a lot of blades that are cracking," DeLisi said at the hearing.
Habedank and other witnesses stopped short of saying what caused the crack.
But Dawn DiMarco, a metallurgist with Robson Forensic in Philadelphia, posited that maintenance technicians may have been too aggressive in "grit-blasting" the fan blade during overhaul. Analysis of the blade showed at least one piece of grit lodged in the metal, and perhaps another such piece started the crack, she said after reviewing the documents at the request of the Inquirer and Daily News.
"They're getting pieces of the grit that are left behind, which is not good, because it is a source of contamination that could affect adhesion of the coating or serve as a site for crack initiation," she said,
Then there was the second failure: When the blade tore through the engine, its protective sheath couldn't contain the shrapnel.
Data collected by the NTSB described a fan blade break: Within 1/200 of a second, the fan blade fragments hit the fan case and cause a ripple, essentially a shock wave that emanates from the impact point. The disruption causes the airflow to surge. In the next two seconds the engine shuts down.
When the engine used on the Southwest plane was tested in the mid-1990s, engineers believed the casing would contain a broken fan blade, testified Stan Minabe, a project engineer with United Technologies Aerospace Systems, which made the case. The casings didn't, so engineers added a shield to contain any blade thrown from a rotor spinning at more than 5,000 rotations per minute. In tests, it was effective.
In the Philadelphia incident, though, the fan blade may have been thrown forward beyond the containment shield into the engine's inlet. Fragments of the inlet, a ring in front of the containment shield blades that ensures a smooth airflow into the engine, and the cowling, the metal shell that a plane's passengers see when they look out the window at an engine, broke off and hit the wing, fuselage, and one cabin window, causing the air in the cabin to rush out.
The FAA and Boeing have no standards for a window's ability to withstand impact, according to documents compiled by the NTSB, and there are no "directed inspections" for passenger windows.
Investigators saw similar damage to the inlet in a nearly identical mishap involving the same model engine on a Southwest flight near Pensacola, Fla., in 2016. It was a kind of damage experts had never seen, witnesses said.
"Fragments should not have traveled forward of the containment shield," Minabe said.
Victor Wicklund of the FAA's transport standards branch confirmed in the hearing that the damage seen in both the Philadelphia and Pensacola incidents was unlike anything seen before.
The inlet, meanwhile, is supposed to be able to withstand the impact of a fan blade to be certified for use.
Witnesses testified that modern diagnostic tools allow a much better understanding of the ripple effect that occurs when a fan blade breaks.
The Philadelphia and Pensacola incidents suggest the tests that engines undergo may not include all that can happen when something goes wrong in flight, DeLisi said.
"We are seeing on the Southwest events," he said, "that inlet retention is not assured even after a successful certification test. So isn't that a good opportunity to rethink what we're doing on the ground and how well it simulates that failure in flight?"
Addressing the problem would likely require engineers to consider changes in testing, design, and inspection, said Mark Ricklick, an aerospace engineering professor at Embry-Riddle University in Daytona Beach, Fla.
"The aerospace industry," he said, "it has to be safe, of course, but it also has to be light, and those two things counteract each other."
Technology that is not quite ready for use will likely be a longer term help in ensuring incidents like this don't happen again. Researchers are designing sensors that would be able to detect cracks in materials, or vibrations, pressure differences, or temperature variations that are outside the norm in the engine, giving advance warning of part failures that doesn't require manual inspections.
"The number of sensors on engines is continuously increasing," Ricklick said, "so I'd say [it's] a matter of time."
DeLisi told the hearing that manufacturers of engine components didn't agree how to interpret some of the fragments from the accident. He asked how much longer the investigation would need, addressing Minabe, the United Technologies engineer.
"There's so much data and the hardware is in such a state that it takes a lot of time in terms of understanding what you're looking at," Minabe said. "There are things that we might first look at and think of as one thing, and if you come back and look at it again, you can think of it a different way. This takes an enormous amount of reviewing of the data, and just making sure we get it right."