COMET Chapter Six Measuring Disaster

I invite you to witness the midnight blue and white de Havilland Comet settle onto the runway at Hatfield, Hertfordshire, England (above). It was Saturday, 10 April, 1954. The prodigal child had returned to the place of its birth.  

Across the tarmac, in the giant factory built to assemble the world's first jet powered passenger aircraft, the production lines were crowded with aluminum frames of the second generation, the Comet 2, being fitted with more powerful Rolls-Royce engines and a redesigned wing. 

But without paying homage to its offspring Comet G-ALYU – Yoke Uncle - taxied directly to the testing hanger.

The first generation was in sorry shape. After three years of service, of the nine Comet's leased to British Overseas Airways Corporation, only four were still flying. Yoke Yoke had disappeared near Stromboli just two days earlier. Yoke Victor had been lost north of Calcutta, India. Yoke Peter had exploded in mid-air near the island of Elba, Italy. Yoke Zebra had failed to get airborne off the runway at Rome (above) and was a total loss. And Yoke Uncle had been chosen to be the sacrificial lamb.

Workers who had assembled Yoke Uncle three years earlier presided over her demise. They removed the interior fittings – seats, carpeting, wall coverings – stripping Yoke Uncle to bare metal. To replace the seats and tables, heavy duty water plumbing was installed. 

And then on Friday, 7 May, 1954, Yoke Uncle took her last flight of sixty miles, from Hatfield to the Royal Aircraft Establishment at Farnborough. The fuel tanks were pumped dry. And, in a final indignation, Yoke Uncle's tail was sliced off. The engines were removed, to suffer their own autopsy, and the husk of Yoke Uncle was towed to the prepared site.

The engineers and scientists at the R.A.E. had always harbored suspicions about the de Havilland design. And the first pieces of Yoke Peter, dredged up from the bottom of the Tyrryenian Sea, and brought to Farnborough (above) had strengthened those concerns.

Six weeks ago, a month before Comet Yoke Yoke disappeared off of Naples, the engineers had begun preparing the site of Yoke Uncle's Calvary. A concrete foundation had been poured. Atop this a series of precast concrete forms (Jersey Barriers) were placed at regular intervals. Atop these were laid and welded 4' by 4' steel plates to form a flat surface. 

Yoke Uncle was then towed until it's tricycle gear straddled the construction stand. More plates were welded together until they fully enclosed Yoke Uncle, forming a water tight tank 112 feet long, by 20 feet wide by 16 feet high.

When the tank was finished, 250,000 gallons of water was pumped into shell of Yoke Uncle and the surrounding tank until a pressure of 8.25 pounds per square inch was achieved in both. This was the “test pressure” the Comet had been designed to withstand, the difference between the outside air pressure at 40,000 feet, and the internal “pressurized” cabin, set to replicate an altitude of 8,500 feet. 

Over two and a half minutes an addition 100 gallons of water was pumped into Yoke Uncle. Then the 100 gallons was pumped out again, completing a 5 minute cycle, which would represent a 3 hour flight, such as between London and Rome. This was repeated for 999 cycles. Every 1,000 cycles the internal pressure inside Yoke Uncle was increased an even further 33%, to 11 pounds per square inch pushing against the 8.25 lbs of the surrounding tank. 

At the same time the wings were being constantly flexed, because that was where Farnborough was convinced the problem with the Comet's skin would reveal itself. 

They were aging the Comet 40 times faster than she was expected to age while in service.

While still flying, Yoke Uncle had experienced 1,221 of these pressurization cycles. In the tank, it now suffered a further 1,836 cycles – for a total of 3,057 – when a minor fluctuation in the pressure gauges alerted the engineers. Performing due scientific diligence, the tank was drained and the skin of the Comet inspected. A tiny crack 2 millimeters long had formed near a rivet hole attaching the forward port escape hatch to the aluminum skin of Yoke Uncle. Still expecting a failure in the wings, the engineers patched the crack, refilled the tank, and resumed testing.

But as the summer progressed, the testing was repeatedly delayed as more tiny cracks appeared, radiating out from the windows. In their turn, they were all repaired. But with each new crack it became obvious the problem with the Comet was not in the wings, but in the thin aluminum composite skin of the pressure hull. And then, after 5,546 cycles there was an loud thud from within the tank, and the pressure gauges inside the shell of Yoke Uncle dropped abruptly. The pumps were turned off, and the tank drained again.

What the engineers and scientists found, startled them.  A 4.5 meter section – almost 15 feet long - of the cabin wall near number 7 window on the port side, had exploded (above). Had the cabin been pressurized with air, the effect would have been the equivalent of a 500 pound bomb going off. The water in the outer tank had suppressed the explosion, leaving the evidence intact. 

Had Yoke Uncle been still in service, the engineers calculated, this massive failure would have occurred after about 9,000 hours of flight.

Based on the metal fatigue on the Comet prototype, a failure of the hull was not expected to occur until 16,000 cycles. And as the design life of a Comet was to be only 10,000 cycles, design engineers had seen no cause for concern. But in service the hulls were failing after only 3,000 + cycles. Why? 

 Arnold Hall (above), head of the R.A.E. noticed that on a test bed the Comet hull prototype had been initially pushed to 2.5 times the anticipated internal pressure. And it had passed. He now suggested the de Havilland engineers had thus accidentally “fatigue proofed” the prototype's hull, locking the rivets and Redux adhesive together, strengthening them as in a trial by fire. The production line Comet's had not been subjected to this extreme pressurization, and thus  “cold” hulls had failed at 3,000 + cycles.

The separate  and independent examination of the recovered wreckage of Yoke Peter told the story. The first crack had ripped apart the front of the passenger cabin, and within less than a second killed everyone on board.  The rear of the fuselage and tail broke downwards. The nose and galley section were spun off, and gravitational forces broke the the outer lengths of the wings downwards. The center fuselage with the stubs of the wings caught fire as they corkscrewed down, but this quickly burned out.  Or so went the theories. 

To prove both of these theories, Eric Lewis Ripley, of the R.A.E., built 50 1/36 scale Comet models, 5 feet long with 3 foot wingspans, but designed to come apart just as Yoke Peter Comet was assumed to have. By tracking the fall of these debris on a hanger floor, the hired Italian fishing boats on Elba returned to the crash site, searching for a particular piece of the puzzle on a particular section of the ocean floor. After a couple of hours searching, the fishermen found it; Yoke Peter's Automatic Direction Finder windows (above).

The ADF windows were atop the aircraft (above), just forward of the wings. And they showed the “unmistakable fingerprint of fatigue..." 

 It fact, it was determined this had been the original crack, the exact spot where the pressurized cabin of Yoke Peter had first broken, by a joining of cracks from the aft lower corner of the forward escape hatch with one from the right-hand aft corner of the ADF window. Eric Ripley would later write a technical article on the issue, entitled, “Fractures talk their own language.”

https://www.youtube.com/watch?v=aHbyOqE0q3o

The evidence was now lining up. Yoke Uncle had first failed a 3,060 cycles, Yoke Peter at 1,290 cycles, but with repairs. Yoke Yoke after only 900 cycles. Yoke Victor at 3,050 cycles – all far below the 16,000 assumed by de Havilland's calculations. The cracks had all begun at the the windows because they were square. The right angles intensified the pressures at the corners to 450 pounds per square inch, far above the pressures designed to be accommodated. But it wasn't just the pressure. It was repeated stretching of the cabin, exacerbated by the short range of the Comet, requiring so many presserizations and de-presserizations.  

So there it was. The Comet's Achilles Heel was the square windows and the square Plexiglas ADF aperture,  antagonized by the aircraft's short range.  The only question left, was “What now?”

- 30 -