Reprise: Night of the Comet

The first edition of Aviation Safety Digest in 1953 reported on the runway overrun of a de Havilland Comet 1 in Rome, the previous October. Now Flight Safety Australia reanalyses this crash, with the help of the pilot who flew the fateful aircraft on its previous flight.

The scene has been replicated countless times since but, this take-off, on 26 October 1952, was one of the first few hundred of the commercial jet age. To anyone watching at Rome Ciampino airport the streamlined de Havilland Comet would have seemed a vision splendid of a glorious and imminent future.

But this time it went wrong. As quoted in the first issue of Aviation Safety Digest, Air Commodore Sir Vernon Brown’s official report on the crash says, ‘the speed … failed to build up and after becoming airborne for a few seconds, the captain, acting under the impression that there was a lack of engine thrust, decided to abandon the take-off and throttled back the engines. At the same time, the aircraft hit an obstruction a short distance beyond the end of the runway and finally came to stop near the airport boundary.’

Sir Vernon described how ‘the brakes were released, and the aircraft made a normal acceleration. At an IAS of 75–80 knots, the nose wheel was lifted from the runway and a slight tendency to swing to starboard was corrected. At an IAS of 112 knots the captain lifted the aircraft from the ground by a positive backward movement of the control column and when he considered that the aircraft had reached a safe height he called for “undercarriage up”. At about the same instant the port wing dropped rather violently, and the aircraft swung to port; the controls gave normal response and lateral level was regained.’

‘At this point the captain realised that the aircraft’s speed was not building up, although he made no reference to the ASI. A pronounced buffeting was felt which he associated with the onset of a stall and in spite of two corrective movements of the control column the buffeting continued. Before the first officer had time to select undercarriage up, the aircraft came down on its main landing wheels and bounced. It was now evident to the captain that the aircraft’s speed was not increasing, and he was convinced that there was a considerable loss of engine thrust. He was also aware that the aircraft was rapidly approaching the end of the runway and a decision to abandon the take-off was made. The undercarriage struck a mound of earth as he was closing the throttles and the aircraft slid for some 270 yards (250 metres) over rough ground. The main undercarriages were wrenched off and considerable damage resulted; a large spillage of fuel occurred but fire did not break out.’

Peter Duffey was the 27-year-old Comet first officer at the time of the crash and had been on the crew flying the crashed aircraft from Britain. Now an undiminished 94-year-old, he recalls riding a police jeep out to see the hissing ruin of G-ALYZ, which had had its fuel tanks ripped open by rocks in the run-off area. Nearly 70 years later, he is still indignant at how the accident was investigated and reported. Like many subsequent analyses, he says the crash involved much more than so-called ‘pilot error’.

The Comet’s pilot, Harry Foote, was demoted from jet operations to flying the Avro York, a boxy freighter based on the wartime Lancaster bomber. ‘His reputation is sullied by that accident report,’ Duffey says, ‘and we knew it at the time’.

‘Foote was a very good pilot, as he demonstrated later when a propeller detached on a York and damaged the right side of the fuselage,’ Duffey says. ‘He did very well indeed to land it safely because the (four engine) York was not a happy aircraft in asymmetric flight. It was a handful with one engine out, but Foote landed with only two engines.’

Duffey cites three technical factors in the crash that Sir Vernon never discussed. ‘First was the lack of attitude reference on the artificial horizon. Unlike most modern aircraft today, it had no pitch markings. So, you couldn’t judge the pitch attitude when you moved the controls back to rotate the nose up. Also, because of the configuration of the windows in the nose, you lost all reference to the natural horizon.’

‘The second point was that the controls on the Comet 1 had only a 20-pound (9-kg) spring bias. So, no matter what speed you were going you had no feedback. This was dangerous, and later with the Comet 4 there was Q-feel artificial feedback, giving progressive stiffening as airspeed increased.’

‘Finally, we know now in hindsight that the wing section produced extra drag in a high nose-up situation. Under these circumstances, to raise the nose inadvertently too high is not a pilot error but a consequence of design.’

Soon after, de Havilland changed the profile of the Comet wing. Author Graham Simons gives details in his 2005 book Comet! The World’s First Jet Airliner. ‘Leading-edge slats had been fitted to the prototype as an insurance against wing-drop, but they had made little difference to stalling speed and their mechanical complexity was dispensed with early in-flight trials. With hindsight, perhaps with more angle they might have prevented the ground stall.’

‘The prototype was modified using wooden frames and wood screws—it worked well with only a small penalty in cruising speed, so the fitment became standard.’

Duffey wonders how de Havilland’s Comet test pilot, World War II night fighter ace John ‘Cat’s Eyes’ Cunningham missed discovering the wing problem. ‘He should have discovered this characteristic of the wing. That was his job as a test pilot. And remember that as a result of this accident all subsequent aircraft certification tests have involved over-rotated take-offs to make sure the aircraft can get off the ground in that situation.’

Even before the Rome overrun, Duffey was becoming aware of the awkward reality behind the Comet’s glamorous image. ‘We felt on top of the world but gradually became very conscious of the fact that we were flying a development aircraft. It was a, er, very interesting experience.’

‘There were any number of problems,’ he says. ‘Every time you put the LORAN (navigation receiver) on you needed to run the inverter, which had about a ten-minute run before it cut out.’

‘On a fast descent the windscreens would mist up and we used to get towels from the cabin crew and constantly have to wipe them clear. The autopilot would disengage without aural warning in mild turbulence and faulty seals caused extravagant leakage of the hydraulic system.’

In his aviation autobiography Comets and Concordes, Duffey described how during descent in icing conditions, using low engine power, the centrifugal compressors and intakes would accumulate ice. ‘There was insufficient hot air (for de-icing) unless the descent rate was slowed by increasing power, causing other problems.’

The Comet 1 also had a fuel filter icing problem. The workaround for this was blowing the ice off the filters using full fuel pressure. Results of the procedure could include an engine over fuelling that sent flames as far aft as the tailplane and cause a rich extinction and an engine rundown.

There were also operational issues. Duffey writes about the surprise of seeing another Comet emerge from cloud on a reciprocal course at flight level 300, over the Burmese coast, at a closing speed of 800 knots. With the few Comet 1s as the only jet airliners in service, high altitude separation was not even considered to be a problem. ‘There were no upper level airways, nor any aids to enable accurate guidance along them had they existed.’ Another Comet, on the African route, became lost near Entebbe and flew round for an hour, becoming dangerously low on fuel. There was no weather radar to help avoid heavy cloud and storms.

In a stop-press to the Rome accident story, the first Aviation Safety Digest reports on a similar accident at Karachi in Pakistan, shortly before publication. This was the second crash that demonstrated the flaw with the Comet’s wing, Duffey says. The Karachi crash differed from the Rome overrun in one terrible way—all 11 on board were killed. ‘They nearly got away with it,’ Duffey says. ‘The left gear hit a sandbank and that was the end of them.’

The captain, John Pentland, had been briefed on the Rome accident by test pilot Cunningham, but had not previously attempted a night take-off at maximum weight in a Comet, according to author Simmons.

Duffey remembers a swift change in procedures by BOAC. ‘At Rangoon we received a signal from London saying, “add 1000 feet to the take-off distance and leave the nose wheels in contact until V2”. That was a safe solution.’

All of these became historical footnotes to the disasters of 1954, when two Comets exploded in cruise flight, one in January, the second in April. The Comet had been grounded after the first crash but cleared to fly after modifications, which had been made despite the cause of the crash being unknown. Another grounding followed the second crash, near Naples, and the remains of the January crash near Elba were dredged from the Mediterranean seabed. An airframe was taken out of service and subjected to pressure testing in a water tank, which after 3057 pressurisation cycles revealed unanticipated metal fatigue, leading to structural failure.

The often quoted ‘square windows’, story of the Comet 1’s structural failure is a media simplification, Duffey notes. The failure on the first aircraft (the second was never recovered) was in an ADF cut-out on top of the fuselage, which was rimmed by a cracked and repaired cast aluminium frame, used, and approved possibly because of tight production scheduling.

Duffey knew the Comet 1 was developing a problem with cracking, as he had visited the Comet maintenance hangar at Heathrow and saw the big tailplanes being changed as part of the routine Check 4. This was because of premature cracking. A Comet crash at Calcutta was caused by structural failure in heavy cloud, and the first piece of wreckage in a long trail was found to be a tailplane.

‘This should have led to a grounding even if the two Italian crashes had not happened,’ Duffey says.

Duffey was on leave with his young family at the time of the second crash. He found out about it from his father-in-law, who wordlessly handed him a copy of that morning’s The Times. The Comet dream was over, and by sheer luck he had survived.

Duffey reverted to first officer on the Canadair Argonaut, a Rolls-Royce Merlin engine version of the Douglas DC-4, but returned to jet flying on the Comet 4, before converting to the Boeing 707, the aircraft that relegated the Comet to a footnote in aviation history. From there he was chosen to be one of the first pilots on the Concorde. He was proud and excited but nagged at first by a feeling history might repeat. ‘When I went onto Concorde my wife told me I was a sucker for punishment,’ Duffey says. ‘But things had changed, and it was an extremely well-made aeroplane.’

He now leads an active retirement in Vancouver, Canada, a country which first charmed him when, as a teenager who added a year to his age, he learned to fly in 1943 with the Royal Air Force. He also has fond memories of Australia, where on a temporary basing he flew the Boeing 707 and lived in Manly. He contributes to online aviation forums and has been discovered as a source of expertise on the aviation standards and cultures of the past by academics and experts, who greatly value his input. He also remains engaged with 21st-century aviation. Certification, automation and human-machine interface are subjects that continue to intrigue him, although he concedes that he watches from the far back stalls these days.

‘The most automated thing I deal with these days is my Volvo XC60, which has computerised aspects that I will probably never discover, much less understand—it’s occasionally ridiculous,’ he says.

But ultimately, the challenge of automation management in aviation comes down to the basics, just as driving his Volvo is ultimately a matter of wheel, stick and pedals. ‘Nothing in aviation is foolproof, but whatever it is, if you set power and attitude, it will fly, if it’s properly designed.’