A fundamental decision had to be made – which engine to use? de Havilland were committed to powering the Type IV, now designated DH 106, as a pure jet.
de Havilland had their own engine genius in Frank Halford. His H.1 turbojet, later to be known as the Goblin, was available and, better still ‘in house’ being built by one of the de Havilland group of Companies – the de Havilland Engine Division.
de Havilland had been conducting design studies on a twin jet version of it’s famous Mosquito fighter/bomber. This ‘design study’ eventually evolved into the D H 100 Vampire. The Vampire flew in 1943 and, folk-lore has it, that de Havilland’s C.C. Walker was to say to BOAC’s Campbell-Orde that the airline could have Vampire performance, and a cruising speed 530 mph., in a civilian transport aeroplane if it was powered by turbojets. This ‘vision’ sold the concept to the Chairman.
D H 95 Flamingo
By 1941 a twin engine (Goblin) medium range transport called the D H 95 Flamingo was under consideration. It was unique aeroplane for de Havilland, not only in it’s use of jet power in a passenger aircraft, but also because it was the first time an all metal stressed skin type of construction had been built by the company. However it was never destined to received jet propulsion.
Yet another design study was considered which was, in effect, a larger four engine version of the Vampire proposal!. Chief Designer of de Havilland R E Bishop described this study as a, “stupid aeroplane”.
The main design team under Bishop working on the Type IV included chief of aerodynamics Richard Clarkson, and his assistant was David Newman. Many different design concepts were to pass across their desks before the 106 design was finalized in 1946.
After initial Goblin Vampire/Ghost Vampire variations were considered the design was to change radically. The fall of Germany disclosed some amazing developments in airframe and propulsion technology. Always innovative, German engineers had been experimenting for some time with jet and rocket propulsion systems and also with some interesting concepts in the field of aerodynamics.
A concept of great interest to de Havilland designers was the use of swept wing configurations in high-speed flight.
Bishop made a fact-finding visit to Germany in late 1945. Perhaps influenced by what he saw a new proposal emerged in May 1946. Configured as a 24 passenger airframe and featuring sharply swept back wings – it was tailless too!
Power was from four 5000lb thrust Halford designed engines located in pairs either side of, and next to, the fuselage and buried in the wing root. An all metal pressurized fuselage some 8ft in diameter was designed to take eight rows of seats arranged three abreast. It was anticipated that this version would operate on Empire routes where a maximum stage range of 2200 miles would not be an embarrassment. A reduction in payload (to 18 passengers) would give North Atlantic capability – but only with a stop-over en-route.
The Ministry accepted that such a radical configuration required detailed evaluation and, with this in mind, commissioned the building of the D H 108 development prototype in August 1945 – the 108 was unofficially named Swallow.
Swallow was based on the wooden fuselage of a Vampire. It had attached to it metal wings swept back at an angle of 40 degrees. At the rear a fin and rudder topped the engine nacelle. Two prototypes were ordered one (Goblin 2 powered) to test low speed flight characteristics and another (Goblin 3, later Goblin 4 3750 lb. thrust) was destined for high speed trials.
Swallow was fast. During practice for an attempt on the world air speed record it achieved 637 mph. at 9000 ft. The record was held by a Meteor at 616 mph. Tragically on September 26th 1946, when flying low level, the aircraft broke up killing Geoffrey de Havilland Jnr.
A third prototype was produced, strengthened and made more aerodynamic. VW 120 finally set a new record of 605.23 mph. over a 100km closed circuit on April 12th 1948 in the hands of a de Havilland test pilot John Derry. Derry also piloted the 108 when, in a dive, it became the first British plane to exceed the speed of sound. Sadly all three 108’s were destined to crash with fatal consequences. Test flying was a hazardous business.
While all this was going on the technical risks inherent in tailless design, and the apparent cost and weight penalties involved, were beginning to be appreciated. Experience with the 108 confirmed fears that this configuration was not yet viable. Back at the drawing board a swept-back conventional tailplane was added to the design and this revised version was the configuration outlined when the first Public Relations brochure for the 106 was published in May 1946.
Finally, in July 1946, the design team announced a change to the angle of wing sweep. By opting for a 20° sweep, instead of the original 40°, over a ton of weight in wing structure was saved. The change significantly improved handling characteristics – particularly take-off and slow speed capabilities. The penalty? Cruising speed was down from a projected 535 mph. to 505 mph.
As development continued the length of the fuselage was increased to accommodate 32 passengers seated four abreast in a fuselage of diameter 9ft 9 inches. Gross weight was now projected to be around 100,000
In September 1946 the Ministry of Supply (formerly Ministry of Aviation) placed an order for two prototypes to specification 22/46. A price of £250,000 was set by de Havilland before the final design was fixed – clearly there was to be no profit with the first batch.
As months slipped by, increased production costs and all-up-weight problems began to weigh heavily on the design team. It was proposed that a large, single wheel design of undercarriage be adopted early on so that there would be the minimum of delay in getting the prototype airborne – it was never intended for production. Development was well underway of a new type of four wheel bogie and, it was hoped, it would be available by the time production aircraft were under construction.
Drag calculations indicated that the design was now 18 percent above specification. So critical did performance calculations become that it was decided to compensate for a possible ‘over-shoot’ in all-up-weight targets by making provision for rocket assistance on take-off!
The design was modified to incorporate a DH Sprite 5,000lb thrust rocket engine – located between the two Ghost jet pipes on each side of the fuselage. Also evaluated was ammonia injection to boost the Ghost’s power. Even in-flight refueling was considered!
The obvious choice of engine for the 106 was the Ghost turbo-jet built by de Havilland Engines Division. The Ghost was a development of Goblin – which had been available since 1941 and was already extensively developed. Goblin experience was incorporated in the larger Ghost which had a first run on the test bed on 2nd September 1945.
In-house Ghost Engine used in the early Comets
Ghost was too wide at 53 ins. – and created difficulties for the design team. Although it was known that a slimmer, more powerful, alternative engine (the axial-flow AJ65) was going to be available from Rolls-Royce in the early 1950s, it was decided to civilianize the Ghost and avoid delay. ‘Military’ Ghost was re-designed with some sixty percent of the detail design drawings re-drawn.
Development of the Ghost was undertaken using two modified Lancastrian aircraft (VM703 and VM749) for low altitude evaluation. The two outer Rolls-Royce Merlin engines were replaced by Ghosts. For high altitude work a Vampire – TG278 – was modified. In June 1948 the Ghost received type approval as a civilian power unit. It was rated at up to 5000 lbs. st. thrust.
Further progress came in January 1947 when the Ministry of Supply placed an Intention to Proceed with de Havilland for the construction of eight aeroplanes – all intended for BOAC. Later six more aircraft were ordered for use by British South American Airways thus making a total of 14 orders. However BOAC merged with BSAA. and the total order was revised to nine aircraft. The price per aircraft was still set at £250,000 each … and against this figure costs were rapidly rising – as always.
DH106 becomes the Comet
In December 1947 the DH 106 was christened the ‘Comet’. Development continued in parallel: systems and equipment tests were conducted using various other aeroplanes (Vampires, Lancastrians etc.). The shape of the nose, for example, was finalized by modifying a Horsa glider. This strange looking craft was towed by a Halifax bomber during evaluations.
The Comet was designed to operate at up to 40,000ft for relatively long periods in an environment that was little understood at the time. Designers had to create a structure that was robust enough to withstand high internal and external pressures and sustain very low temperatures – parameters that in many respects could only be guessed at.
On military aircraft limited pressurization had already been used. Now the intention was to pressurize the whole crew/passenger environment. At 8.25 psi. the pressure was twice that in use on other airliners of the time.
A de Havilland innovation – subsequently to become standard practice on all passenger turbojets – was to use compressed air taken directly off the engine’s compressors for cabin pressurization but for a while this practice caused problems with engine over-heating. The Engine Division were constantly under ‘pressure’ themselves to improve the Ghost’s specific fuel economy and ‘bleeding off’ of hot compressed air did not help matters. Eventually compressed air destined for the cabin was directed via coolers, pressure regulators and humidifiers. More ‘bled’ hot compressed air was used for de-icing purposes.
The Comet was the first aircraft in which the pressure loads were higher than the flight loads. In simple terms previous aircraft had pressure cabins that were much stronger than necessary in order to withstand the flight loads and they had not, therefore, been exposed to the phenomena of ‘pressure fatigue’ in the same way. This was to be of major significance later on in the Comets history when failures of the pressure cabin were to occur.
Recognizing that the company were moving into unknown territory with the unique Comet, de Havilland devised special testing facilities specifically for it. At Hatfield a test chamber was specially constructed. It was designed to operate at up to an equivalent of 70,000 ft. and down to a temperature of -70 degrees centigrade – and allow a considerable margin of safety with respect to both. Whole sections of the pressure cabin structure along with the windows were tested. Test windows were subjected to pressure loads of 8.25 psi. 2000 times and then, for good measure, tested at 100 psi. A forward section of the fuselage was subjected to 16,000 cycles in the water-tank.
The Royal Aircraft Establishment at Farnborough had been devising and conducting extensive tests to investigate the fatigue-life of aircraft structures. Much work had been done on devising methods of measuring various types of loading, for example, gust loading. The RAE had devised an instrument that could automatically count the number of ‘accelerations’ and grade them for severity.
RAE research had led them to believe that continuously repeated testing of the main components and such things as spar joints – if applied to at least six examples would give very accurate data. The procedure was to apply a constant load and upon that superimpose a heavy and continuous reciprocating load thus simulating positive and negative ‘g’. One minute in the test rig was equivalent to one hour of flight. By extrapolating results it was thought the fatigue life of any structure could be accurately determined.
The RAE would test full-scale specimens to destruction on a strength testing rig e.g. a full wing. Loads were applied by hydraulic rams. In the case of the Comet the wing and a section of the fuselage were first tested through a certain life-span and then the same specimen was loaded to destruction-point in the strength test rig! There was also a ‘cold-box’ to simulate temperatures at high altitude and another form of testing was to make the structure vibrate at its natural frequency.
The RAE were consulted early on in Comet development. The first prototype G-5-1 ended up at Farnborough in early 1953 for further testing. By then it had been ‘flown’ the equivalent of 10,000 hours during de Havilland test flying and structural testing. It was noted though, even after this punishment, that there were a few minor deterioration’s. The main structure ‘continued to creak and groan reliably under this continuous reversal treatment in which loads from 0.7 to 1.3 ‘g’ were being applied.
Other changes Bishop made for the Comet was the extensive use of Redux metal glue for attaching the skin to stringers and doublers. Although used previously on aeroplanes bonding had never been used as extensively as on the Comet. The reason for its use? Thinner aluminium plate could be used and problems associated with the use of rivets eliminated – essentially weight was saved and sealing problems did not arise. In any case Redux bonding added greatly to the structural strength of the aircraft.
For a paper on Redux Bonding that appeared in Flight in 1951 click here