From the hangar to the runway

American Airlines flight 191, Chicago, 25 May 1979

Adrian Park analyses the deep roots of a disaster that happened forty years ago next month

I don’t usually speak this way but here goes: aircraft engineers are my heroes. For 25 years of flying—my flying—they’ve checked, tightened, maintained and replaced hundreds of thousands of parts—large and small—on aircraft I’ve flown over ocean, land, forest, desert and snow. And somewhat amazingly, they’ve done it with nary a mishap. That alone should be reason enough for the hero call but there’s another, and I’d like to illustrate it using philosophy: let me introduce you to the philosophical term ‘counterfactual’.

The Oxford Dictionary says a counterfactual is a philosophical adjective ‘expressing what has not happened or is not the case’. The philosopher David Hume put it this way:

‘We may define a cause to be an object, followed by another, and where all objects, similar to the first, are followed by objects similar to the second. Or in other words, where, if the first object had not been, the second never had existed …’

Putting aside the initial befuddlement, it’s just describing how the addition or the subtraction of one circumstance might change the whole course of events. You might well say what’s the point of that? Well think Einstein’s famous thought experiments. Thinking through in systematic ways what might happen because of the addition or subtraction of a circumstance opens up new trajectories of investigation and observation. Take the very famous, or infamous, case of American Airlines flight 191—a favourite for human factors case studies everywhere. Let’s begin with the factual—what did happen—and then move to the ‘counter’: what most certainly did not happen.

It was March 1979. The tail end of the famous ’70s. Think disco, the ‘fro, space opera (Star Wars) and the disturbingly widespread acceptance of flared pants. If ’70s pop culture birthed all that, then ’70s aviation birthed a very 70’s airliner: the DC-10. The DC-10 didn’t have two or even four engines: it had three. One under each wing and one looking like a Thunderbirds retro-fit melded into the tail-root empennage. It’s worth a google. So is the ‘fro. The design of the DC-10, like the 70s, was kind of bold and in your face: an almost adventurous aesthetic.

There was something else adventurous going on with the DC-10: underwing engine mount removal and replacement. In early 1978, McDonnell Douglas issued service bulletins calling for the replacement of the spherical bearings on the pylon mounts of each underwing engine. The service bulletin recommended the procedure ‘be accomplished during engine removal’. One can almost feel the face-palm moment as engineers read this line. Of course it should be accomplished during engine removal, the only way to get to these bearings and replace them was to remove the whole engine.

McDonnell Douglas helpfully advised that the 4900 kg engines should be removed from their pylon mounts and then, if necessary, the pylon mounts removed from the wing. Because of the maintenance hours involved in this process many operators including American Airlines decided to get a little enterprising. They’d heard their competitors over at United Airlines were removing the engine and the pylon mount in one go by using an overhead hoist. The American Airlines engineers decided to go one better: a forklift instead of a hoist. Thus, it came to pass that, by clever coordination of said forklift, forklift operator, maintenance crew, hand signals, voice signals and not a small amount of luck, the number of fuel, hydraulic and electrical disconnects was reduced from 79 to 27 and the total maintenance time reduced by 200 hours.

Despite the time-saving benefits, the aircraft manufacturer clearly had concerns with the procedure:

McDonnell Douglas would not encourage this procedure due to the element of risk involved in the re-mating of the pylon assembly to the wing attachment points.

Before too long the procedure was the done thing and many aircraft were subject to pylon-replacement à la forklift.

On 29 March 1979, at the American Airlines maintenance facility in Tulsa, Oklahoma, what should have been a curiosity had become common place: a turbofan engine—as big and as heavy as two Range Rovers—balanced on the tynes of a repurposed forklift underneath the left-hand wing of a DC-10, registration N110AA. Maintainers on work stands yelled or gestured alignment cues to the forklift operator below them. It was, very literally, a tall order. The movement required to get the mounts re-mated required precision adjustments in the order of millimetres. The best the operators could manage was centimetres. Adding to the difficulty was a faulty check valve on the forklift which meant a pressure bleed-off and, in turn, ‘jerking’ motions as the operator tried to correct for the drift-down of the forks.

At some point during the procedure 1500 kg of forklift force transferred itself through the carefully balanced engine-pylon-assembly to the aluminium of the wing mount surrounds. Without anyone noticing, the wing came out of the altercation with a large, 25 cm crack. After a few more hours of manoeuvring, gesturing and voice commands followed by a shift change, the engineers were done. They moved onto the next job. Investigations would later find at least nine other aircraft had been damaged in similar ways.

And here is where we enter the realm of four counterfactuals—four circumstances that didn’t happen that would have fundamentally avoided the impending tragedy. Firstly, what didn’t happen was the engineers learning from the numerous other incidents where the wing had been damaged. It’s tempting to say if just one engineer or one supervisor had made a professional nuisance of themselves and pointed out the risk to the structural integrity of the wing with a view to getting rid of the forklift then the crack, then the tragedy to follow would never have occurred. The facts are murkier in this case, however. An engineer at the Tulsa depot, Joe L. White, said he did raise concerns that were ignored. American Airlines denied this and fired White for alleged poor performance after the crash. It said he only raised concerns after the crash. White says an American Airlines lawyer tried to influence him to be quiet about what he knew before the crash. White sued the airline, won, had the judgement in his favour overturned, reached a settlement, lost his house, and became a lawyer. Perhaps the lesson here is to keep several copies of everything you write if you feel you have to blow the whistle.

So, the first two counterfactuals are where there either was a report, or that it was taken seriously.

The third counterfactual, the second ‘what didn’t happen’ was one of these engineers (perhaps after having their professional nagging rebutted) deciding to report to the regulator the previous forklift mishaps. After all there had been numerous other instances of damage on other aircraft and even, in a few instances, a ‘pistol-shot’ noise as the wing was punctured. The regulations demanded such instances be reported. If they had been reported, even by another company, and appropriate action had been taken by the regulator, then the crack in N110AA would never have happened and neither would the tragedy to follow.

The fourth is the service rep at McDonnell Douglas making a professional nuisance of himself and insisting to his bosses they really should do something about the unauthorised procedure. Instead, despite their expertise and their clear concern about the re-mating tolerances being at the mercy of an unwieldy forklift, they went with ‘McDonnell Douglas does not have the authority to either approve or disapprove the maintenance procedures of its customers’.

On 25 May 1979 at 1504, N110AA, operating as American Airlines flight 191, lined up for take-off at Chicago O’Hare International Airport. On board were 258 passengers and 13 crew. As the aircraft accelerated along the runway passengers were able to get a pilots’ view through the cockpit windshield from a newly fitted and rather innovative closed-circuit camera. As the pilots applied maximum power to the number one engine for take-off they also applied maximum stress to the weakened wing and its 25 cm crack. During rotation the wing integrity finally had enough: 5 tonnes of weight combined with accelerative and take-off forces tore the engine mounts clean away. With that, the engine went rogue. It didn’t just drop onto the runway, it went out fighting flipping up and over the wing and smashing down onto the runway behind. It was a whole new rendering of the term ‘engine loss’. Large sections of hydraulic and electrical lines went as well leaving the wing slats without hydraulic power and cockpit stall warning without electrical power. The DC-10, still able to climb on the number two and three engines, struggled into the air with the pilots fighting to keep the aircraft level and under control.

But then, at 300 feet, the bad situation turned horrible. The loss of hydraulics meant the slats on the left-hand wing had been retracting ever since the engine had departed the fix. Hydraulic pressure normally kept them out. With no hydraulic pressure the slats completely retracted just as the pilots needed them the most. This, in turn, increased the stall speed of the wing beyond the airspeed of the aircraft, because, in a vicious irony, the crew had followed standard operating procedure to the letter and slowed the aircraft to best-climb speed. Later simulator analysis showed the aircraft could have flown at higher speed. The stick shaker stall warning had also been disabled when the hydraulics were torn away, making the situation the equivalent of an aviation king hit. The pilots had no audio warning their left wing was stalled. The roll continued reaching a bank angle of 112 degrees which meant N110A was now more upside down than right side up—never a good thing for an airliner full of passengers.

Speaking of the poor passengers, in a very cruel twist, they got to watch their own demise. Through the CCTV their world would have tilted abruptly left, filled with ground and trailer park, and then abruptly ended. The DC-10 struck the ground and became a scattering conflagration of metal and fuel. The impact was so bad investigators would later state ‘little useful data was obtained from post-impact examination’. The biggest part to survive was the number 1 engine-pylon which was found off the right side of runway 32R.

Counterfactuals are really about the ‘if only’. If only an engineer had spoken up, or if only he had been taken seriously. If only someone had reported. If only the regulator had acted. If only the manufacturer had insisted. The passengers would not have viewed their world abruptly turn left, then upside down and then end in a black and white, CCTV-mediated fireball. And although passengers and crew would have had neither the awareness or the motivation to say it, the truth of the matter would be if any one of those engineers had spoken up, if anyone had done something, they would indeed have been heroes. They would have saved 271 people on board flight 191 and two people on the ground even though they would never have been thanked for it. A crew chief at the Tulsa depot, who killed himself two years later, might have lived out his natural lifespan.

Which brings me back to my intro: engineers are indeed, most of the time, my heroes. I’m pretty sure in 25 years of flying—my flying—behind a hangar door or over an open engine cowl or in a closed office, more than one engineer has indeed spoken up, reported, acted, insisted and/or professionally nagged. And as a result, my own aviation experience has been relatively unbroken which I’m deeply thankful for. These engineers might never have dragged a passenger from a burning aircraft or successfully handled an engines-out ditching but make no mistake, they’ve saved many a life. And that is why I think they are heroes.

Further Information

National Transportation Safety Board (NTSB) (1979). Aviation Accident Report AAR-79-17. Retrieved from https://https://www.ntsb.gov/investigations/accidentreports/pages/aar7917.aspx