In early 1987 the Tornado GR1 was coming to the end of the stores clearance and envelope expansion trials. The final sorties by the military test teams were to be flown on a pre-series airframe, P14, at the Italian flight test centre at Pratica Di Mare near Rome. Yet another week in Rome - it was going to be hell but someone had to do it! I was the Tornado GR1 project pilot on 'A' Squadron at Boscombe Down so I hastily volunteered myself. I was accompanied by one of 'A' Squadron's navigators, an experienced F4 man who had a few hours on the Tornado F3 and was easily coerced into this plot with: "Fancy some GR1 flying - and by the way it's in Rome". The flight test engineer (FTE) for the trial was an experienced helicopter FTE, but this was his first real Tornado trial having "crossed over".
The first sortie was to be flown with two 1500 l underwing fuel tanks and five BL755 cluster bombs. The first test point was a wind-up turn to the g/AoA limit (irrelevant to this tale) and the second test point was a rapid rolling test point in 25 cruise wing (wings fully forward, flaps and slats up) at 340 KIAS and 20,000 ft. The test was to commence at 18 degrees AoA in a constant speed descending turn, the aircraft rolled through the erect using full lateral stick (whilst maintaining constant longitudinal stick position), and after a 180 degree bank angle change the roll was to be rapidly reversed with full lateral stick and held for a further 180 degree bank change. The test was to be flown with the SPILS (spin prevention and incidence limiting system) switched off, a normal practise for Tornado handling trials at the time as this was the worst case. Now, I was a little circumspect about this test point as the clean aircraft SPILS off AoA limit for rapid rolling was 18 degrees and we had a lot of stores (albeit with empty underwing fuel tanks for this rolling test). Knowing that the company had not flown this test but had cleared it on modelling data, I asked how far we were from the departure boundary to be told: "You're well clear". At this stage I was less than 18 months into my first flight test tour and had not yet developed a healthy enough level of scepticism and cynicism about "LCCs", an abbreviation which I shall not define for fear of libel. Therefore, I accepted the brief in good faith.
The initial wind-up turn was uneventful and we set up for the rolling test. A very steep nose down, overbanked attitude was required with COMBAT power (fully rated Mk 103 engines) to try to maintain speed, and there was only a very brief period of stabilisation at 18 AoA in a left turn before the very rapid (less than 0.5 seconds) application of full right stick. The aircraft rolled smoothly but VERY slowly to the right and I remember thinking: "Why do they call this rapid rolling?". After 180 degrees of roll, the stick input was reversed to full left, again in about 0.5 seconds, and the aircraft started to roll to the left but at an even lower rate than before. Before the aircraft had reached wings level I realised that the roll response was not normal and started to centralise the stick (the rudder pedals had been neutral throughout). Too late! The nose sliced and the aircraft pitched up as it departed more rapidly and violently than anything I have seen before or since. Subsequent data analysis showed that it went from 340 KIAS into a fully developed spin in less than 2 seconds with a yaw acceleration of at least 140 degrees/second/second. As it departed I immediately centralised the stick, which did nothing. Based on Hunter spinning experience, I continued to full forward stick but that also did nothing; the aircraft was in a fully developed spin. I should add that I had flown a Hunter spin practise sortie only five days before so I was very current in swept wing spinning. I started to carry out the Aircrew Manual full spin recovery procedure: airbrakes in (they were), throttles idle (they were still at the fully forward COMBAT position, although the reheats had extinguished, so had to be retarded), full back stick (to minimise fin blanking) and full in-spin lateral stick. This latter action was to saturate the CSAS (the analogue fly-by-wire flight control system) by demanding a roll rate greater than can be achieved in spin, thus maintaining full spoiler and differential taileron deflection and providing some anti-spin moments. Now, I thought that we were spinning left so applied full left stick; the aircraft stayed in a fairly smooth and very steady spin.
At this stage I transmitted that we were spinning to our FTE who was on a radio link but had no telemetry data as this was not supposed to be high risk nor a high AoA trial! He instantly called back; "Centralise". We were a little past that stage. Two thoughts then went through my mind. Firstly, as altitude reduces, air density increases, aerodynamic damping improves and spin recovery characteristics get better. Then, followed about a nanosecond later by :"S**t, we're going to have to eject"! We were over the sea and the spinning ejection height was 10,000 ft. I had decided to go at 9,000 ft just to give the aircraft every chance to recover, knowing that the seat was still well within parameters. At 12,000 ft, a little voice from the back seat was heard for the first time during this event: "Twelve thousand". He was obviously thinking about ejecting as well. I have no idea how many turns we did from our 21,000 ft entry altitude down to 10,000 ft but it was quite a few. And then, as if by magic, at 10,000 ft the strip AoA gauge (an indicator to the left of the HUD) went from full scale deflection high to reading on scale. Based on Hunter experience, this told me that the aircraft was recovering from the spin although I could not detect any change in the rotation. Therefore, I immediately centralised the stick. I was very disorientated by this stage and could not read the HUD accurately but became aware that the airspeed indications changed from two to three digits; the airspeed was increasing. Now, the book said to level the wings when the speed builds to 200 KCAS then pull out of the dive. In this configuration, there was so much drag that it had only reached 130 KCAS by 8,000 ft and acceleration was very slow. Therefore, at that point I selected the manoeuvre flaps and slats down, rolled wings level, and started a gentle pull out of the dive. It was soon obvious that we were going to have plenty of separation from the sea, at which point the second rear seat comment was made: "Yeeeee Haaaaaa"! We bottomed at 3,000 ft.
The engines had not surged, there were a few CSAS failures (most of which reset) and every piece of navigation equipment had failed. We flew straight and level for about 5 minutes to allow our gyros to re-erect then followed a road back to Pratica for a gentle downwind join to land. Straight away we went to see the boss of the test squadron to tell him what had happened to his aeroplane. His reply? With a glare and very positive eye contact, "So, we almost get rid of the P14". I think that he would have preferred us to leave what was quite a hangar queen in the Mediterranean!
So why did it spin? Well, although the AoA had appeared to be steady at 18 degrees, it continued to increase after the initial lateral stick input was made, without any further aft stick input from me, to 20 degrees. This, unbeknown to us, was the predicted departure boundary. The sideslip generated during the roll reversal triggered the departure, and we validated the model! It is worthy of note that all SPILS off rapid rolling clearances had been set at 5 to 6 degrees below the departure boundary. Therefore, we would never have attempted the test point had we known that we were only 2 degrees below. This configuration was cleared to a limit of 15 degrees for service use.
And why did it take so long to recover? I had "sensed" that we were spinning left and had thus applied left lateral stick. We discovered when we analysed the flight test instrumentation data that we were actually spinning right and thus I had applied outspin roll control which generated some pro-spin moments. Subsequently, I discovered that just about everyone who had departed a Tornado inadvertently had either mis-identified the direction of departure or had been uncertain of the direction. The Tornado does have a very small turn indicator at the bottom of the attitude indicator but in my 36 hours on type at the point I had never noticed it making it, in my opinion, not likely to be found and interpreted correctly in the disorientation of a spin!
Why had other departures in service generally not resulted in a fully developed spin? Most had been with the same wing sweep angle (25 degrees) but with the manoeuvre flaps and slats down. In this configuration, the Tornado does not have a fully developed spin mode such that centralising the controls after departure will result in a recovery. With the manoeuvre flaps and slats up, especially with no stores on the outboard wing pylons, it will invariably go into a fully developed spin post departure.
I could carry on a lot longer analysing this incident, but I will await any questions. As one final thought, my recent Hunter spin experience when this happened was probably what gave me the little capacity that I had to recover this situation, and even then I mis-identified the direction of spin. If you are involved with manoeuvrable aircraft, take every opportunity that you can to practise spinning.