JAS 39 Gripen crash in Stockholm 1993 Aug 08 report summary

Preliminary report summary

• The only equipment malfunction before the crash was the electronic map which had nothing to do with the crash
• The flight control system, the engine and all other systems worked as specified until the aircraft impacted
• No external cause is suspected
• The pilot was properly trained and equipped
• The limits for minimum altitude and maximum angle of attack were exceeded insignificantly and did not have anything to do with the crash
• The manufacturer and customer knew that large and rapid stick movements could cause divergent Pilot Induced Oscillations, but considered the likelyhood of it actually happening insignificant, so all pilots weren't informed
• The red warning light was too late in telling the pilot that the control system was saturated, for him to do anything about it
• The low altitude of 270 m made it impossible for the pilot to try to regain control

This caused the control surfaces to move with their maximum deflection speed, and as the flight control system then had little or no control surface movement on its own to work with, the stability margin was reduced. At the same time the alert system informed the pilot of this, and he no longer recognized the aircraft's behaviour.

The aircraft started to roll to the left and pitch up. In response to this the control stick was moved almost to the right endpoint and some forward. The result was a roll to the right to 35 deg bank and a lowering of the nose to 7 deg below the horizon, whereupon the stick was pulled forcefully back and to the left. At the same time, the artificial stability system tried to raise the nose, which in combination with the pilot's command caused a powerful pitch up. By this time, the control stick was fully forward, but the aircraft was already unflyable.

From exit of the turn until ejection the sequence took 6.2 s. The time from when the pilot didn't recognize the flying characteristics until stall took 2.7 s, but the control system warning wasn't shown until 1.2 s before the stall.

The cause of the crash was the misjudgements that PIO was so unlikely and that the warning light would tell about any problems early enough for no mention of this in the pilot's manual was necessary.

My own comments

Some doubt has also been cast on the statement that the flight control computer worked as specified, as its log only goes a few seconds back and thus only showed data since the pilot left the aircraft. The crash report doesn't go into this in detail, but it is clear they had its error log to work with, in addition to data from the "black box".

This report also clears up some things I didn't understand from earlier this year, when it was said that maximum angle of attack was 26 deg, and at 35 deg the rear airbrakes come out and the canard gives a full pitch down command automatically. Obviously fly-by-wire does not have to mean you impose hard limits on the aircraft's performance, in the Gripen case the pilot is trusted not to exceed them.

This part of the text last updated 13 Nov 93

Final report summary

The AAIB has had contact with France in order to learn about how flight control systems are validated internationally. Since long time, Saab and the Swedish customer has had, and continues to have, contact with American parties in order to gain access to experience gained with the flight testing of YF-16, YF-17, F-16, F-18, the Space Shuttle, B-2, C-17 and YF-22.

• At the crash, the flight control computer was severely damaged, but all pieces were recovered and all memory circuits identified, which allowed the information in them to be read, so full data from the whole crash sequence was available to the investigators.
• The way the display was planned was not in accordance the goals stated for the flight test programme. Nor were any reasons stated for the exceptions made from the Pilot's Manual.
• Limits set for the display were exceeded by the pilot, which didn't directly contribute to the accident.
• The control laws are complex, which means that there have been problems to fully analyse their function.
• The effects of control surface movement speed limitations had not been fully investigated for the full flight envelope.
• The validation process hadn't successfully identified the pilot induced oscillation properties of the aircraft.
• When flying the simulator, less work with the control stick was needed than when flying for real.
• Simulator studies have showed that it is possible to stall Gripen at less than 20 deg angle of attack during similar conditions.
• During the low speed left turn, the automatic roll trim was disconnected, exactly as automatically is intended to at angles of attack over 20 deg. Because of this, the control stick had to be kept 2 deg to the right in order to maintain constant bank angle.
• The stick movement to the right at the end of the turn was the same as during training, but as it was begun 2 deg to the right already, it reached the end position, leading to a larger roll rate than during training.

As reasons for the crash is stated:

• Pilot commands
• Control stick properties
• Control laws limitations
• Control surface limitations

Contributary reasons:

• It was lighter, giving a higher thrust/weight ratio, as well as making it more unstable in pitch.
• The control stick required lower input forces.
• The control stick could be moved to give a larger roll input, resulting in higher available roll rates.
• The mass distribution was different, requiring the control stick to be kept 2 deg to the right to maintain constant bank angle when the automatic roll trim system was disconnected.