TSR2

by Damien Burke

  Flight 17 ended with a touchdown on a 300yd (275m) carpet of foam laid on runway 26 by Warton’s fire section. With additional strain gauges on the undercarriage, Flights 17 to 20 all ended with landings on foam carpets to see what effect a low-friction surface had on the touchdown oscillations that were being experienced. BAE Systems via Warton Heritage Group

  As camera footage showed that most of the yawing motion set up in the undercarriage appeared to begin at the torque links, the first attempt to solve the problem was to fit stiffer torque links. Unfortunately, after several further flights the evidence showed that these had no effect. The spin-up rate of the wheels still seemed to be the primary cause, but this was affected by the descent rate at touchdown, and the period between port and starboard main gears touching the runway also had an effect on the severity of the oscillations. Flight 12 had produced a particularly violent sequence of oscillations on landing, owing to a particularly high descent rate.

  Here is XR219 in flight, probably on Flight 21, with the ‘Aylesbury tie’ modification. This was a tie-rod connecting the bogie to the fixed portion of the oleo to alter the bogie trail angle so that the front wheels touched down first (similar to the accidental configuration on Flight 7). On the few occasions it was flown, this modification was successful in reducing the violent oscillations experienced on touchdown, but it is likely that some structural changes to reduce fuselage resonance would also have been needed to effect a complete cure. BAE Systems via Warton Heritage Group

  Examination of film of the main undercarriage during these landings revealed a bewildering variety of movements at touchdown. The bogie was yawing from side to side and pitching up and down as individual wheels contacted the runway. The main leg was spreading outwards and also ‘twanging’ fore and aft, and it was this fore-and-aft movement that proved to be the key to solving the problem. It was counter-intuitive, but the various lateral motions of parts of the undercarriage were much smaller contributors to the fuselage oscillations than the fore-and-aft movement of the entire assembly. Adding a strut between the oleo leg and the rear of the bogie not only stiffened the whole assembly but also slightly altered the bogie trail angle. It was hoped that this would improve spin-up characteristics as the wheels contacted, and allow far less fore-and-aft movement in the oleo leg. A fixed strut was installed on XR219 for Flight 21. This meant that the undercarriage could not be retracted, but it would serve to demonstrate if the strut worked or not. If it did, a retractable strut could be designed and fitted.

  The first flight with the modified undercarriage showed that oscillations were hugely reduced. There was still some lateral movement, but it was no longer a cause of disorientation. Several more flights followed, each with either minimal or no oscillations. It appeared that the new strut was not a 100 per cent cure, but was very close to it. After the aborted Flight 25, XR219 was laid up for a number of modifications, including the installation of the retractable struts, and these were also fitted to XR220. Unfortunately neither aircraft was flown with the retractable strut before the project was cancelled. The lay-up of XR219 would also have included the fitting of manual controls to allow the intake cones to be moved, strengthened intake lips, an AFCS, a fin mass-balance and the two-stage reheat fuel restrictor system. The last, in particular, would have permitted the aircraft to operate at higher take-off weights and thus carry more fuel, necessary for proving flight up to Mach 2 (though the wing tank system would still have been inhibited, limiting Mach 2 endurance to 15min per sortie).

  XR220 and XR221

  The accident suffered by XR220 on arrival at Boscombe has often been blamed for losing BAC the opportunity to get more than one aircraft into the air before the nearly inevitable cancellation. While taxy runs did not start until March 1965, this was primarily due to the extremely late delivery of flight-capable engines from BSEL. Even if XR220 had not spent several months being repaired and reassembled, it is not likely that engines would have arrived any sooner, and taxy tests could not, therefore, have begun any earlier than March 1965.

  By the end of March XR220 was frustratingly close to flying, and BAC was looking forward to possibly taking it to France for the Paris Air Show in June. The airframe had gone through just about every test programme apart from undercarriage cross-functioning. These tests were scheduled for the 4 April, with the first flight on the 6th. However, a minor foul was found on an undercarriage sequencing valve, and the final tests were expected to have to be delayed until the 7th, with a flight on the 9th. The BAC team at Boscombe worked hard and regained some of the lost time, with the result that XR220 was very nearly ready on the morning of the 6th. A minor hydraulic leak needed to be dealt with, however, and, as it became clear that the aircraft would not be ready until the afternoon, the aircrew knocked off for lunch.

  Meanwhile, at Weybridge, XR221 was already ‘flying’, with a crew aboard and taking it through a complete simulated sortie from Wisley to Land’s End and back as a final check of the AFCS installation. Having made a completely successful ‘flight’, the crew began their debrief and were looking forward to flying the aircraft for real once it had been transported (via road) the short distance to Wisley later in the month. That afternoon, in the Budget Speech, the announcement was made that the TSR2 programme had been cancelled. At Boscombe Down the flight crew arrived back at the airfield in a hurry, anxious to fly XR220, only to find that orders had already been given to prevent this. Thus XR220 would never fly, and XR221’s one and only ‘flight’ was made without it leaving the ground.

  Repairs to the damaged areas had been just about completed by the time this picture was taken of XR220 in the hangar at Boscombe Down in late 1964, but final reassembly was delayed by the need to use some components on XR219. BAE Systems via Warton Heritage Group

  Delivery of engines for XR220 was delayed time and time again, and it was not until March 1965 that the aircraft was ready to begin engine runs. BAE Systems via Warton Heritage Group

  Initially, XR220’s flight-test programme was to permit the aircraft to fly at up to Mach 1 at low altitude, extending the envelope to Mach 1.7 at higher altitudes and then up to maximum altitude during the first 30hr of flying, with the investigation of flutter characteristics as the primary task. A large number of accelerometers were fitted within the airframe to measure vibrations on the various flying surfaces, along with a number of ‘bonkers’ (explosive inertia exciters). The housings for the latter are evident on XR220 to this day, as chamfered plates scabbed on to various locations on the aircraft, such as the wingtips, fin and tailplanes. Accelerometers were also fitted to measure loads on the undercarriage. Mounted within the weapons bay instrumentation pack were a pair of more substantial inertia exciters, hydraulic rams that could set up vertical and lateral oscillations within the fuselage. The aircraft’s initial flights would primarily be aimed at ensuring that it was in a fit state for the ferry flight to Warton, which was planned for about Flight 5. With gradual modification to bring it to a higher standard, including the fitting of a higher standard of engine (Olympus 320 instead of 320X) and the activation of wing fuel tanks to permit longer sorties at higher Mach numbers where fuel consumption would be higher, XR220 would gradually widen the envelope. Some problems with AFCS structural coupling meant that a ‘frig box’ was to be fitted to enable limited auto-stabilization, unlike XR219. Also unlike the first aircraft, XR220 had a sprung mass-balance in the fin to combat flutter, permitting flight at higher Mach numbers than XR219 would be capable of. A fix for the airbrake closure problems would be embodied in XR220’s second phase of flying, along with a production-standard rear fairing.

  Details of the ‘bonker’ layout on XR220. A number of these inertia exciters were mounted on the tail surfaces and wingtips of specific development-batch airframes (not XR219), and have been misidentified as ECM aerials in some publications. BAE Systems via Brooklands Museum

  A typical ‘bonker’ installation on XR220 (this pair is near the leading edge on the underside of the st
arboard tailplane). Damien Burke

  As well as the primary task of flutter testing, XR220 was also instrumented to measure engine compressor-blade strain, and LP shaft stresses, to verify the effectiveness of the various fixes introduced on the Olympus 320 by BSEL.

  TSR2 handling characteristics

  After six flights, Roland Beamont wrote a summary for the Air Ministry of his opinions on the qualities of the aircraft:

  Briefly the handling assessment of take-off, low-speed flying within the initial landing gear envelope, and approach and landing, has been completed with 100 per cent success. The flying qualities in this configuration are as good or better than predicted in every case, and it is without doubt the easiest high-performance aircraft to land that I have flown. All six landings to date have been, we are told from the visual point of view, perfect; and they have been accomplished within a recorded scatter of +/– 4kt [4.6mph; 7.4km/h] from the scheduled touchdown speed, and at a measured vertical velocity of 4ft/sec to 1.5ft/sec [1.2m/sec to 0.45m/sec]. Perhaps of more significance than these numbers is the fact that the pilot feels in every case that he can do just this. With this excellent control harmony we feel fairly certain now that there will be no cost growth on the basis of a need to develop basic low-speed stability and control.

  A TSR2 model in the Aircraft Research Association windtunnel at Bedford. The streaks of oil gave an indication of the airflow during the run. This shot shows the results of a Mach 1.2 run with three small vortex generators on the side of the nose in line with the bottom of the windscreen (these were fairly successful in cleaning up the flow on the sides of the nose), and a centreline fence added to the top of the fuselage to try to regain some stability (to no significant effect). Aircraft Research Association

  Flights 15 to 24 by XR219 were all flown from Warton, and, before the fitting of the additional undercarriage strut, served mostly to expand the flight envelope further and give the crews more experience of various drills and configurations. The aircraft performed well, with serviceability as good as could be expected for the first of any type, and its flying qualities continued to impress the three crews assigned to the flight-test programme. No serious problems other than those already mentioned were experienced, though the view through the canopy transparencies was a constant cause of complaint. Redesigned units were under test when the programme was cancelled.

  Sadly, the single example of the TSR2 to fly was never fitted with the AFCS developed for the aircraft, and thus was limited to manual flight control. Regardless of the limitations of flying without automatic assistance, the test pilots who flew the aircraft considered it viceless in the envelope in which test flying took place, and well within the capabilities of the average Lightning pilot. Indeed, test pilots Jimmy Dell and Don Knight both described it as ‘just like flying a big Lightning’.

  The TSR2 was naturally stable up to about Mach 1.5, when directional stability began to be lost, resulting in negative stability by Mach 1.7. Windtunnel tests at the Aircraft Research Association’s facility at Bedford in mid-1962 had determined that this loss of stability resulted from vortices shed from the upper ‘shoulders’ of the forward fuselage, which interfered with the fin, but no practical aerodynamic solution had been found to cure this. Instead, the hope was the fin auto-stiffener system, with a lateral accelerometer sensing the amount of sideslip and the auto-stabilization computer calling for a proportional amount of counteracting fin movement, would be able to compensate for the loss of stability. The auto-stiffener was to be active at all times, not just above Mach 1.5, to improve handling. Because an auto-stiffener failure (non-moving fin, or fin runaway) could result in loss of the aircraft, the entire system was triple-redundant, with a voting system taking the majority decision and applying it.

  Some handling aspects of the airframe had given rise to concern, based on windtunnel and theoretical work, and in June 1960 English Electric wrote to the CAL in the USA to ask if it could use the laboratory’s variable-stability NT-33 research aircraft. This was a highly modified T-33, fitted with a Lockheed F-94 Starfire nose to make room for additional electronic equipment. The front cockpit had its flying controls connected to the control surfaces via an AFCS using servo-mechanisms, the output of which could be varied by the safety pilot in the rear cockpit. Additional servos attached to each control surface could also be controlled by the safety pilot. Each control surface could be tied to a combination of outputs. For example, the rudder could be made to respond not only to a basic rudder input from the pilot but also to a combination of rolling and yawing velocity and acceleration, angle of sideslip and its rate of change, or products of these and various other flight parameters. In this way the aircraft could be set up to simulate the flight dynamics and feel of another type, while the safety pilot had direct and conventional control from the rear cockpit should the simulated flight characteristics prove uncontrollable.

  Simulated TSR2 flights using the CAL NT-33 began on 27 October 1962, and twenty-four flights were made, totalling 40hr of flying, the last flight being made on 17 November. The CAL staff were extremely helpful, holding together a tight flying programme regardless of poor weather conditions, and even managing some extra flights despite strictly limited funding being provided. Don Knight, the evaluation pilot, flew sixty separate TSR2 configurations, forty-three simulating low-speed and seventeen simulating supersonic configurations. The general result was an impression that handling was going to be better than predicted, the ground simulator proving overly pessimistic in many cases. In particular, Don assessed non-autostabilized flight in the approach configuration as ‘acceptable and satisfactory’, which was a surprise, but it was no surprise at all that the aircraft was nigh on unflyable in Mach 1.8-plus cases. From these flights the required damping measures for the autostabilizer were verified against the predicted values. In fact, many of the handling problems predicted and simulated were not present on the aircraft when flown.

  The Cornell Aeronautical Laboratory’s NT-33 variable-stability aircraft provided valuable experience of simulated TSR2 flying characteristics. Don Knight made twenty-four flights in the aircraft, coincidentally the same number of flights made by the TSR2 before cancellation. Calspan Corporation

  Handling of the TSR2 itself was described by all of its test pilots as easy and pleasant. Pitch control was responsive and well damped, there was no tendency to pilot-induced pitch oscillations, and the aircraft was exceptionally easy to control on the approach and for touchdown, with good speed stability (the last was particularly surprising to the simulation team, who had predicted somewhat more squirrelly behaviour). The pilots reported resemblances between the ground simulator and the real thing, but said that, in general, the flight handling was a great deal easier and more pleasant than on the simulator. So XR219 proved to have less drag than predicted, and far superior handling; a stunningly good start. Alas, that was all it ever was.

  The members of Warton’s windtunnel team were particularly disappointed that they were never able to find out just why all of the windtunnel data analysis had led them to believe that the aircraft would have difficulty in raising its nose during lift-off, yet XR219 exhibited far superior pitch response and there was never any need to use the extending nose-gear leg. The opportunity to take the aircraft to Mach 2 was also denied them, so the aircraft’s real stability characteristics beyond Mach 1.2 remained an eternal mystery.

  CHAPTER SIX

  The Aircraft

  Fuselage structure

  The fuselage was built in four sections: forebody, forward and aft centre sections, and the rear section, all of fairly conventional construction with skin stringer panels over bulkheads and frames.

  Forebody

  The forebody was initially built in two halves, split longitudinally, these being brought together once the basic structure on either side was complete. Sandwiched between the external panels and crew compartment walls was polyurethane foam covered with Melinex, to act as insulation against the high skin temper
atures generated at supersonic speeds. Within the forebody were the FLR bay, crew compartment, SLR bays and the main equipment bay (split into port/starboard sides with a web in between). Most of the nose section (crew compartment and main equipment bay) was pressurized, with a secondary pressurized area encompassing the radar bay and radome. (An auxiliary pressure bulkhead also protected the crew compartment in the event of loss of pressurization in the equipment bay.) The radome was a glassfibre/ resin-laminate cone fitted on a machined mounting ring, with an inner spherical inner radome that protected the FLR assembly. Unlike those of more-modern aircraft, the radome could not be easily opened and folded to the side for access. Removal required undoing multiple fasteners, and suitable storage equipment for the radome itself once it was removed.

  The crew compartment had two sloping bulkheads on which the ejection seats were mounted, with a floor sitting on bracing members linked to the close-pitched frames rising up on either side. The windscreen and quarter panels were glass, with gold film heating and demisting. The pilot’s and navigator’s canopies were mostly metal framework with metal stressed skin and relatively small transparencies. Unlike the windscreen and quarter panels, these were not glass but a triple laminate of Perspex, once again with gold film heating/ demisting.

  Systems, control runs, wiring and ducting ran for the most part through the lower fuselage, where access could be had through the undercarriage bays and a plethora of access panels. A spur from the central fuel gallery ran forward through the port side of the forebody to a blanked-off position just under the inter-canopy structure. This was designed to be a mounting point for a retractable in-flight-refuelling-probe pack (similar to that later fitted to the Panavia Tornado). The main equipment bay housed various systems in rack-mounted packages, and a separate upper equipment bay above housed some air-conditioning, lox and fuel-system components.