- de Havilland DH-106 Comet 1
- Accident Overview
- Accident Board Findings
- Accident Board Recommendations
- Relevant Regulations
- Prevailing Cultural / Organizational Factors
- Key Safety Issue(s)
- Safety Assumptions
- Resulting Safety Initiatives
- Airworthiness Directives (ADs) Issued
- Common Themes
- Related Accidents / Incidents
- Lessons Learned
- de Havilland DH-106 Comet 1
Copyright Matthew Clarkson - Used with permission
History of Flights
On May 2, 1953, one year to the day after the maiden flight of the British-made de Havilland Comet, aircraft G-ALYV departed Calcutta Airport for Delhi as BOAC Flight 783. A few miles out of the airport, the flight encountered a severe thunderstorm. While the pilot and air traffic control were both aware of it, the storm did not appear severe enough to restrict flight through it. Furthermore, the captain was well-qualified, had considerable experience on this route, and had experience in similar weather conditions. Just six minutes after take off, while climbing to 7,500 feet, radio communication was lost. About this same time, witnesses at various ground locations saw "an aircraft coming down in a blaze of fire through severe thunderstorm and rain" and then crash into the ground. All 37 passengers and six crew members were killed.
The inquiry into the accident, directed by the Central Government of India, concluded that the crash near Calcutta was due to "structural failure of the airframe during flight through a thundersquall." They determined that one of two possibilities caused an overstressing of the plane enough to crash it: either severe gusts from the storm, or over-controlling by the pilot because of the storm. They recommended that the wreckage be more thoroughly analyzed to determine the primary failure, and that "consideration should be given" to modifying the flying characteristics of the Comet to give it more "feel" when loads are applied to the control surfaces.
Comet G-ALYP at Calcutta/Dum Dum
Copyright Matthew Clarkson - Used with permission
On January 10, 1954, Comet G-ALYP departed Ciampino Airport, Rome for London as BOAC Flight 781. About 20 minutes into the flight, as it was approaching 27,000 feet, transmission from the crew ceased mid-sentence, indicating a failure of the aircraft with "catastrophic suddenness." Witnesses on the island of Elba, Italy, saw the aircraft fall into the sea in flames. All 29 passengers and six crew members were killed.
While a crash investigation is normally conducted by the government or aviation authority in the country of the crash, it was determined that the British authorities would head the Elba investigation. The Comet fleet was grounded while investigation began and while de Havilland made modifications "to cover every possibility that imagination has suggested as a likely cause of the disaster." These modifications were made to address any possible cause of failure including flutter of control surfaces, primary structural failure due to gusts, flying controls, explosive decompression, engine fire, failure of a turbine blade, and fatigue of the wing. Fatigue of the fuselage was not considered as a cause at this time, nor was a modification made to compensate for it.
As these modifications were made, and while wreckage was still being recovered, the British Minister of Transport and Civil Aviation noted "the nature and extent of the modifications planned... and whilst the Calcutta disaster is completely accounted for... we cannot eliminate that the accident might have been due to some other cause which was possibly common to both disasters." Believing the unknown cause of possibly two accidents had been fixed during the massive modification project, Comet flight was resumed March 23, 1954.
BOAC Comet G-ALYX at London, November, 1952
National Air and Space Museum, Smithsonian Institution
Just over two weeks later, on April 8, 1954, Comet G-ALYY departed Ciampino Airport, Rome for Cairo, as South African Airlines Flight 201, chartered through BOAC. About 40 minutes into the flight, while climbing through 35,000 feet, the aircraft experienced a catastrophic in-flight break-up and crashed into the sea near Naples. All 14 passengers and seven crew members were killed.
Immediately following this crash, BOAC suspended all Comet flights. The Airworthiness Certificate was removed from all Comet aircraft and the fleet was subsequently grounded indefinitely. It would take four years for the Comet to conduct commercial airline flights again - this time as the Comet 4.
Very minimal wreckage of the Naples aircraft was able to be recovered due to the great depth to which it had sunk - about 3,300 feet. From what was able to be recovered, it was concluded that there were no inconsistencies with "the view that the accident to Yoke Yoke [Naples] was attributable to the same cause as the accident to Yoke Peter [Elba]."
Now, with the cause of three accidents in the span of one year all possibly hinging on the findings of the Elba wreckage, efforts to recover the remaining pieces were renewed. Underwater television cameras were used for the first time. By the end of August 1954, 70% of the Elba crash had been recovered.
Photo courtesy of John Heggblom, taken by J.C. 'Connie' Heggblom.
With still no definitive cause, investigators decided to do full-scale tests on existing fuselages: unpressurized flight tests on G-ANAV and pressure tests on G-ALYU. To conduct the pressure tests in a safer manner, a water tank was constructed to encase the fuselage. The fuselage was submerged and filled with water, and then additional water was pumped into the cabin until the pressure inside the fuselage reached 1P, the equivalent of flight. This was then cycled to simulate many flights over the life of an aircraft. By using water instead of air, water being a much less compressible fluid, the test would be much safer and the fuselage would be able to be repaired and re-tested as necessary. Had air been used, the results would have resembled the catastrophic in-flight break-ups at Elba and Naples.
Comet G-ALYU in the water tank for pressure tests. The fuselage failed at a corner of the squarish forward escape hatch window.
(View Large Photos)
G-ALYP, Elba, showing the two ADF windows. This piece
was determined to be the origin of the in-flight break-up.
(View Large Photo)
Source: Canadian Forces Joint Imagery Centre, Reference number PL-62095. Department of National Defence.
Reproduced with the permission of the Minister of
Public Works and Government Services Canada, 2008.
G-ALYU had undergone 1,230 pressurized flights before testing and 1,830 tank "flights" before the fuselage failed at the corner of a squarish forward escape hatch window. This failure was the pivotal evidence needed to turn the direction of the investigation towards fatigue. A scale model was next created to test the theory of fatigue failure of the fuselage at a window corner. The results were then mapped to the crash site near Elba, and a new search area created. At this new location, the aircraft's Automatic Direction Finder (ADF) windows, also squarish, were recovered within hours. The ADF windows are on the very top of the fuselage, just forward of the wings. This piece of Elba wreckage, containing the two ADF windows and adjacent material bore the "unmistakable fingerprint of fatigue," and was determined to be the first fracture of the Elba crash.
G-ALYU withstood about 3,060 pressurized "flights," whether in the air or in the water tank. The Elba aircraft had experienced 1,290 pressurized flights. The Naples aircraft made 900 pressurized flights. All these seemed to indicate a much lower fatigue life than the 16,000 successful cycles de Havilland tested.
Even in the design stage, de Havilland knew that the Comet would be a great technological advancement. They were competing to be the first company to offer pressurized jet service to the public. Since there was little experience in the design and production of pressurized commercial airliners at the time of the Comet development, deHavilland placed special emphasis on structural testing. One area of special emphasis involved pressure testing of the fuselage at higher than normal pressures.
Both the International Civil Aviation Organization (ICAO) and the British Civil Aircraft Requirements (BCARs), the applicable regulations for any British-made civil aircraft of the day, required a design pressure of 2P and a proof test of the fuselage up to 1.33P, where "P" is the working pressure difference, or the pressure expected in normal flight. For the Comet, P was approximately 8.25 pounds per square inch (lbs/in2 or psi). Neither ICAO nor the British authorities were fully aware of all the implications and effects of pressurized flight yet, so many regulations remained the same for pressurized and unpressurized aircraft, including the fatigue requirements.
De Havilland significantly exceeded the requirements in their effort to ensure the safety of their aircraft. They decided to design the fuselage to withstand up to 2.5P, and to proof test it to 2P, instead of just 1.33P. A prototype fuselage was pressurized between 1P and 2P approximately 30 times, and then pressurized to "rather over P" another 2,000 times. These two tests were to prove the fuselage as an adequate pressure vessel as well as to prove its structural integrity. Much later, in the summer of 1953 after Comets were already flying, regulations started to be published that required further fatigue testing for pressurized fuselages. Consequently, de Havilland went back and tested the same prototype fuselage with another 16,000 pressurization cycles between zero and 1P to verify its fatigue life. The fuselage finally failed at 16,000 cycles due to fatigue cracks at the corner of a squarish cabin window. The Comet's expected life was only 10,000 cycles, so cracks at 16,000 were not a concern.
De Havilland Comet Prototype G-ALVG. Note the squarish passenger windows.
Photo courtesy of British Airways Museum Collection - Used with permission.
Stress Concentrations at Window Corners
De Havilland ran many tests in pre-production to prove the safety of the Comet: from pressure tests, to flight tests, to stress tests. The extensive proof testing of the fuselage was believed to be hard evidence that the Comet was safe. This experiential knowledge gained from actual testing bolstered de Havilland's confidence in their analyses. Calculations had been made for an average stress "in the neighborhood of the corners" which found the stress to be less than half the ultimate strength of the material. De Havilland did not consider further stress calculations to be any more accurate than the one already done, and preferred to rely on testing as the main evidence for the adequacy of the Comet. Following the failure of G-ALYU in the water tank however, more testing revealed stress at the window to be significantly higher than that originally determined. The testing found high stress concentrations at the window corners.
A stress concentration is a very localized area of much higher stress than the surrounding area. The stress concentrations were high specifically because of the squarish shape of the windows and window frames which is very different from the round/oval shapes of modern airplane windows. With modern windows, the stress flows freely around the curved edges with minimal build up. But with the Comets' squarish windows, stress cannot smoothly flow around the abrupt corners. This creates stress concentrations.
Still image from Comet 1 Traffic Analogy Animation
An animation describing the stress concentrations associated with the squarish windows is available at the following link: View Comet 1 Traffic Analogy Animation.
Although any aircraft will have varying levels of stress concentrations, the Comet's unique squarish window corners resulted in especially high stress levels. De Havilland tested their prototype to 2P, twice the expected operating pressure. The pressure overload combined with the very high stress levels at the window corners, created stress levels at the concentrations great enough to change the material characteristics at these locations. Each time de Havilland increased the pressure load, the material characteristics progressively changed. Upon achieving the highest load of 2P, these locations had fundamentally different material characteristics than a production Comet. The process by which the material characteristics changed is called cold-working.
Material Cold-Work Properties
Cold-working is not, in itself, a safety issue. The testing to 2P proved the Comet could withstand excessive pressure loads. The significant misstep was the decision to perform the fatigue test on the same prototype fuselage that had undergone the pressure test and had been cold-worked. The prototype fuselage withstood 16,000 cycles before failure, due in large part to the fundamentally different material characteristics of the cold-worked material at the window corners. This characteristic change actually improved the fatigue properties at these locations, which would mask the true fatigue vulnerability of the production Comet. An animation describing how the material characteristics can be changed through cold-working is available at the following link: Comet 1 S-N Diagram Animation.
The Comets that crashed at Calcutta, Elba, and Naples, and G-ALYU in the water tank, had not undergone proving tests to 2P, nor had any other production Comet. These airframes did not have the "benefit" of the application of high loads to improve their fatigue characteristics. As a result, the window corners' natural cycles of stress would quickly wear out, or fatigue, the material. The fatigue had such a great effect on the never-overloaded production fuselages that instead of 16,000 cycles of fatigue life, the Comets were only reaching about 1,000 cycles. At the end of their fatigue lives, the worn-out material would rupture catastrophically, resulting in inflight breakup.
Reconstruction of the Elba fuselage wreckage recovered.
Source: Canadian Forces Joint Imagery Centre, Reference number CAL-43-281-16, detail of photo. Department of National Defence. Reproduced with the permission of the Minister of Public Works and Government Services Canada, 2008.