De Havilland DH 108 Swallow
de Havilland considered a radical new design shape for its futuristic DH106 jet airliner. It proposed a highly swept, tailless airliner of 75 0001b all-up weight to the Brabazon Committee, which was investigating the needs of British commercial aviation after the Second World War. This design envisaged an aircraft with wing sweepback of 40 degrees and a thickness/chord ratio at the wing root of eleven per cent. The four de Havilland Ghost engines were to be mounted beneath the wing near the trailing edge.
Practical experience was required before this design could be proceeded with, and de Havilland successfully tendered to Specification E18/45 for the construction of two prototype DH108s. The primary purpose in building these aircraft was to provide a half scale representation of the proposed Brabazon 4 tailless civil jet airliner on which could be measured in-flight performance, stability and control and aerodynamic pressures on representative sections of the wing. Following this, less specialised research was to be undertaken to explore the characteristics of the swept wing at all speeds.
To facilitate production the DH108 was designed around the complete fuselage nacelle of the Vampire Fl, and two fuselages, TG283 and TG306, were taken from the Vampire production lines at English Electric, Preston, during the autumn of 1945. To these fuselages were added 43 degree swept-back wings of metal construction and a swept fin and rudder. Within seven months, the first of these research aircraft was completed and rolled out at Hatfield for engine runs and taxiing tests on Sunday, April 28, 1946. The first prototype was a low-speed test vehicle with a maximum speed of 451km/h.
As Hatfield was still only a grass airfield, the prototype was dismantled and loaded on to two "Queen Marys " which left by road on May 5 for the emergency RAF airfield at Woodbridge where it began taxiing trials and hops six days later. The wheel brakes overheated and delayed the first flight by Geoffrey de Havilland until May 15 when the DH108 became the first jet aeroplane with swept wings to fly successfully. It was also the first tailless jet aircraft to fly.
The need for horizontal tail surfaces was avoided by the provision of elevons, a combined aileron and elevator control situated on the wing trailing edge in the usual aileron position. The first prototype was powered by a de Havilland Goblin 2 engine of 3 000 lb st and was to be used for low speed research covering the handling and stability of the aircraft. The RAE Farnborough had warned de Havilland of the possibility of dutch rolling or wing drop at low speeds with complete loss of control, from its wind tunnel tests. Therefore anti-spin parachutes were fitted at the wingtips, together with fixed Handley Page slots along the outboard section of the wing leading edge. However, when the aeroplane flew their fears proved unfounded and the aircraft had no trouble flying alongside a Proctor for air-to-air photography.
The new prototype returned to Hatfield from Woodbridge on May 19 when it showed its easy manouvrability while being demonstrated over the airfield by Geoffrey de Havilland.
This aircraft was joined by the second prototype which made its first flight from Hatfield on August 23, 1946, and was shown publicly for the first time at the SBAC display at Radlett on Friday, September 13. The second prototype differed from the first in having a modified 45 degree swept wing with powered flying controls, automatic slots on the wing leading edge, and the more powerful Goblin 3 engine giving 3300 lb st. This prototype was intended for high-speed research, concentrating on pressure plotting and performance. After only a few weeks of flying the DH108s it was decided that as far as the projected DH106 jetliner was concerned, the tailless layout was uneconomical. The use of ailerons as elevators, with their small lever arm moment from the centre of gravity, limited the effectiveness of the flaps. This meant that for a given wing area and runway length, the take-off and landing weights had to be less than if a normal tailplane and elevator layout were used. The sweepback itself also proved uneconomical at this stage, partly because of the lower maximum lift coefficient and the higher structural weight for a given span. Eliminating the sweepback almost doubled the prospective Atlantic payload with the comparatively low power available.
Despite these results the DH108 still had a great deal of valuable research work to do, particularly with respect to high-speed flight. The second prototype appeared a very promising vehicle in this respect and after its Radlett appearance began a concentrated programme of increased speeds.
It was not intended to exceed Mach 0.87 without power-boosted controls, this speed having been attained without difficulty, but it was soon found that the aircraft was well capable of overtaking the current absolute speed record of 616mph reached by a Meteor not long before. The aircraft was cleaned up and prepared for the record attempt, followed by a programme of several dives to determine handling, first at high equivalent airspeed, secondly at high Mach number, and finally a combination of the two.
On September 26, 1946, a level speed of 637mph was attained at 9 000ft, and on the following day, when weather conditions appeared smooth, Geoffrey de Havilland took off for the final practise flight before going to Tangmere for attempts over the official course. He was planning to undertake two tests to embrace all the conditions for the record attempt. Initially he was to dive the aircraft from around 10 000ft at a high Mach number so as to check its behaviour and controllability in these conditions. He was also planning to fly level at high power near the sea to check the speed and behaviour in record attempt conditions. He took off from Hatfield at about 5.30pm with fuel for about an hour's flying. It was a perfect evening with a slight haze, very little cloud and the air was believed to be relatively free from bumps. He had indicated that the tests would be flown over the Thames estuary.
When his colleagues waiting at Hatfield began to realise the aircraft was overdue, messages were received through air traffic that an aircraft had been seen to break up in mid-air and fall in the Thames near Egypt Bay, just north of the village of Cliffe and north-east of Gravesend. Despite fog and low cloud, the wreckage was found the following day and the majority, including the engine, was recovered.
An investigation into the accident began immediately, and although taking some time to complete, did allow a statement to be released on October 7, ten days after the loss of this short-lived prototype. This stated that it was clear from the evidence available after a thorough inspection that the engine was not responsible for the accident. Sir Geoffrey de Havilland announced in the same statement that the de Havilland Aircraft and Engine Companies intended to continue with the full effort of the research work in the DH108 and that John Cunningham had been appointed to succeed Geoffrey de Havilland as chief test pilot of the de Havilland Aircraft Company.
The speed record attempt had been only incidental to the programme of experimental work which was in hand. It had been found desirable to obtain precise measurements of the speed of the DH108 by utilising the accurate timing apparatus set up on the south coast of England for the Gloster Meteor speed-record attempts and it had been decided to make an attack on the record at the same time.
The accident investigation found that wind tunnel tests combined with the structural evidence showed that the aircraft had suffered a sudden catastrophic disintegration during low level flight. Gusts had proved a problem at low altitude, a horizontal one of only 15mph encountered at M 0.87 would instantly raise the Mach number to 0.89. This is only a very moderate gust and such do abound at times at low altitudes. With this sudden increase in Mach number, tunnel tests showed that the aircraft would be in a very dangerous condition. It was noted that gusts had been experienced on an earlier flight that day. This sudden change could have given the aircraft a nose-down pitching moment combined with low stability, causing it to pitch down in a shock stall with ensuing failure of the wings.
Meetings were soon in hand to discuss the provision of a third prototype E18/45 (DH108) to replace the one so suddenly lost. de Havilland had requested immediately after the accident a go-ahead with this third aircraft and a meeting was held at Thames House on November 15, 1946, to work out the details.
Pilot ejection capability was to be built-in, using the then relatively new Martin-Baker equipment. Provision was to be made for a tailplane on the lower part of the fin, but a decision to fit it left until a later date. The power plant was to be a de Havilland Goblin 4 of 37501b / 1701kg st, although a Ghost engine was considered. An important addition was the fitment of power boost control to the elevators, five months after the first flight, to help the pilot overcome some of the excessive loads expected at high speeds.
As before, the fuselage was taken from the Vampire production line at Preston, which by now was producing FB Mk5s, and it was allocated the serial number VW120. This third prototype differed from its predecessors in having a more streamlined canopy (tried out on Vampire TG443), lowered pilot's scat, more pointed nose (tried out on Vampire TG281), better shaped rudder horn and a strengthened structure. It was rolled out for engine runs on July 21, 1947, and made its first flight three days later (24 July 1947), piloted by John Cunningham. It appeared publicly at the SBAC display at Radlett from September 9 to 12 and continued with the first prototype in its original programme of research into the characteristics of the swept-back wing.
The next milestone attained by the DH108 was on Monday, April 12, 1948, when VW120 raised the international speed record around a 100 kilometer closed circuit to 605.23mph (974.0259km/hr) flown at the time by John Derry. This speed was 40mph (64km/hr) better than the previous holder of the record, the Supermarine Attacker, which in turn had stolen the record from the Gloster Meteor. The de Havilland aircraft in fact employed considerably less power than either of the aircraft which it outclassed.
Only six flights in the 108 were found to be necessary before the actual record attempt and they provided all the confirmation the aerodynamics department needed for its preparation of an efficient flight plan. On the day of the attempt it was decided to do a final flight at midday to check for any snags and to make one final run round. Immediately after take off a paraffin leak developed, causing a cloud of vapour to form in the cockpit. The vapour eventually cleared sufficiently for a landing to be made and repairs were made by late afternoon.
It had been decided before the attempt that owing to the difficulties of flying in bumpy air at over 600mph the attempt should not be made before 4.30pm. At 5.00pm a further check flight round was made, but although the 108 was fully serviceable, the air was very bumpy, so a decision was made to wait an hour before John Cunningham made a final cheek in a Vampire. This time the air was found to be quite smooth apart from some isolated spots and visibility was fair though poor on the second leg. Since better conditions could not be expected that day it was decided to make the attempt.
After take-off a climb to 1500ft was made towards Elstree and a turn over Harrow to head for the starting point at Brookmans Park. The line was crossed as close as possible to the aerial masts at about 300ft above the ground with a speed of about 620mph. On settling down on the first leg towards the Puckeridge aerial the air was found to be smooth making a high speed more feasible. The next leg was more hazy making the identification of landmarks difficult and consequently seconds were lost on course correction, but after the brickworks corner the haze cleared and the ASI showed 635mph as the next turn at Sundon cement works was approached, level at 500ft. From here it was possible to see the smoke markers at Gaddesdon Row indicating the last turn, before turning into the final straight over the north of St Albans, past the end of Hatfield's runway and on towards the BBC masts at Brookmans Park.
On landing back at Hatfield one of John Derry's colleagues thrust a note in his hand giving the speed and time and proving the success of the team effort in gaining the FAI Class C record.
During the work up to the speed record John Derry discovered a phenomenon which in retrospect was found to be the true cause of the disaster to Geoffrey de Havilland. When high IAS and Mach numbers were combined, an unstable pitching oscillation developed which was too rapid for the pilot to control, so that at around M 0.88 at sea level divergent oscillations powerful enough to break the wings could develop in less than one second.
Following this demonstration of high speed of the DH108 John Derry continued the programme of increasing the speed throughout the summer of 1948. The ultimate was reached on September 6 when, for the first time in the United Kingdom, the speed of sound was exceeded. Since the Goblin engine in the '108 was of insufficient power to propel it at high speeds in level flight, the high speeds had to be investigated in dives from high altitude. In building up to Mach 1.0 John Derry had reached M 0.91 to 0.95 in successive flights during the previous days. The flight commenced at 10.00am on September 6 with a take-off from Hatfield for a flight which lasted 45 minutes. On this occasion while flying between Farnborough and Windsor the aircraft was climbed to 45 000ft, the pilot wearing a pressure waistcoat and mask. After being flown level to obtain the maximum level speed of M 0.85, owing to engine temperature limitations a dive of about 30 degrees was commenced. At M 0.95 the engine was opened up to 10750rpm but there was no increase in Mach number due to the large increase in drag. In fact it was necessary to move the stick forward slightly to prevent the Mach number from decreasing. Eventually a point was reached at which a very small elevon movement caused a sudden, but not violent nose-down pitch. The first nose-down pitch was over and the aircraft back in a 30 degree dive before an attempt could be made to hold a steeper angle. This small stick movement was repeated a second time, giving a more violent nose down pitch with a return to 30 degrees. On the third attempt a much more violent pitch down was experienced, which was similar to a half bunt and put the aircraft over the vertical. An immediate pull-out was started, but during this the Mach number increased rapidly until 0.97 was indicated. As the aircraft reached a 60 to 80 degree dive with Mach still increasing, the stick force rose rapidly until it became impossible to hold the aircraft, which took over and increased the dive angle to the vertical. At this point M 1.0 was reached and passed.
The engine was throttled back as soon as recovery became impossible and full strength with both hands failed to make any impression on the aircraft. In spite of the power being shut off the Mach number stayed in excess of 1.0 and little time was wasted when once it was realised that the elevons were immovable. The only apparent hope of recovery was then tried and full trim flap was applied at once. Almost immediately a very gradual recovery began until at M 0.94 the aircraft finally levelled out at 23 000ft, giving an indicated airspeed of just under 500mph. The small split flap was set somewhat forward of the trailing edge of the inboard wing and in this position fortunately helped the recovery, although normally it was only used for landing.
During this dive the aircraft appeared extremely stable, being rock steady and no buffet of any kind was experienced in spite of large amounts of negative ' g'. The original intention of the flight had been to take pressure plotting shots with trim flap up and down at varying Mach number, increasing to a maximum of 0.96, but being so near to the sound barrier plans can always change.
Following this unprogrammed flight in excess of the speed of sound a further series of tests was set up to approach the problem more slowly. Mach 1.0 was again achieved on March 1, 1949, in the same aircraft, VW120, by John Derry in a flight of 40 minutes.
With the now familiar nose-down pitch in the high M dive, when 0.99 was reached and just before trim flap was applied for recovery a further steepening of the dive occurred. Full trim flap was selected with no effect at all on this occasion. The engine was throttled. Negative ' g ' increased and a slight roll to port developed. The stick was held on the stops with both hands, and because the rolling was steepening the dive full opposite aileron was applied which was partially effective in levelling the aircraft laterally. However, by now the dive was beyond 60 degrees and although recovery appeared in sight, the aircraft again rolled slowly but uncontrollably to port until, after 90 degrees of roll and considerable negative 'g', it was in a true vertical dive. There was now absolutely no control, even the lateral control having been lost. From a vertical attitude the negative ' g ' now increased even more and the spiralling had almost stopped. The aircraft completed on its own the latter half of a bunt and then rolled slowly round at the bottom so that it became laterally level in a shallower dive. Having lost an enormous amount of height it suddenly began to respond again to the controls and recovered from spiralling into level flight and very shortly into a climb as the trim flaps and elevons became effective. The aircraft was finally levelled out at 29 000ft and a return made to Hatfield.
Mach 1.0 was exceeded during this uncontrollable dive, but it was noticed that control was always regained at about 26 000ft, which of course pointed the way to the realisation that recovery was probably due not to trim flap or luck, but to the increased drag with increased air density at lower altitudes, which also in effect reduced the Mach number to a safe figure.
By this time, when trials were complete and the aerodynamic characteristics of the swept wing had been fully investigated the first prototype, TG283, had been transferred to the RAE, Farnborough, in October 1948. Here it continued further investigations into stability, control and landing trials and when fitted with the longer stroke undercarriage from a Sea Vampire to avoid hitting the tailpipe on the runway, landings were made as low as 95 knots in February and March 1950.
These low-speed trials abruptly ceased on May 1, 1950, when the aircraft crashed near Hartley Wintney, Hants, during stalling tests, killing the pilot, Squadron Leader G. E. C. Genders, AFC, DFNI. The aircraft had apparently entered a spin from which it failed to recover.
Towards the end of 1949 the third prototype, VW120, had joined the first aircraft at Farnborough when, amongst other tests, it was used to investigate longitudinal stability at high speeds. Shortly after its arrival this last prototype was lost on February 15, 1950, when it crashed near Birkhill, Bucks, killing the pilot, Squadron Leader J. S. R. MullerRowland. This accident was apparently due to lack of oxygen for the pilot as no fault was found in the aircraft and no break up had occurred.
The DH 108 was found to be a pleasant aircraft to fly, with d nose-high approach and rate of sink fairly close to the stall. Although this was thought a little alarming at the time it was in fact only exhibiting the modem tendencies of the swept wing. It was a remarkably clean design which was demonstrated by the fact that it could attain high speeds on very low-power. VW120 was easily capable of maintaining Mach 0.88 at 35000ft. Both TG283 and VW120 were fitted with large wing fences, TG283 having two on each wing in November 1947. The wing of TG283 was also given dihedral for a while, making its first flight in this configuration on September 16, 1948.
Despite all three prototypes ending their careers in accidents, the DH108 made some remarkable advances in aviation in the UK. This was not just restricted to pure high-speed research. The powered controls were similar to the ones destined for the DH106 Comet and were developed on this aircraft. The experience gained with the swept wing was used to benefit the development of the DH110. This later became the Sea Vixen all-weather fighter with the Fleet Air Arm.
First Flights, TG283 - 15.5.46, TG306 - 21.8.46, VW120 - 24.7.47.
Engine: de Havilland Goblin 2, 3000 lb st
Wing span: 39ft.
Length: 25ft 10in.
Wing Area: 328 sq ft.
Wing span: 39ft.
Length: 24ft 6in.
Wing Area: 328 sq ft.
Engine: de Havilland Goblin 4 turbojet, 3,750 lb (1,701 kg) st.
Wing span: 39 ft 0 in (11.89 m).
Length: 26 ft 9.5in (8.17m).
Wing area: 328 sq ft (30.47sq.m).
Max take-off weight: 4064 kg / 8960 lb
Max level speed: approx 560 mph (900 km/h) at 45,000 ft (13,720 m) (Mach 0.85)
Max attained speed: Mach 1 in a dive between 40,000 ft (12,200 m) and 30,000 ft (9,145 m).