# Why disable stall warning based only on low airspeed, rather than multiple criteria?

One of the factors that potentially contributed to the Air France 447 crash is that the stall warning was disabled when the airspeed fell below 60 knots.

I understand why this is -- you don't want to have the stall warning go off while the plane is parked at the gate or slowly taxiing.

However, this seems like clearly bad design for two reasons (that should have been obvious to the designers):

1. This introduces an extra point of failure. If the pitot-static system fails, then the stall warning may be disabled due to erroneously low reported airspeed.
2. As demonstrated by AF447, a stalled plane may have very low forward airspeed in an extreme stall.

Moreover, it seems like the solution is easy: Only disable the stall warning if a number of criteria are met. For example, the following logic could be used.

The the stall warning (based on AoA vanes) is disabled if and only if (i) the forward airspeed is below 60 knots AND (ii) the landing gear is down/on the ground AND (iii) the radar altimeter reports a height of 0.

This logic would still prevent erroneous stall warnings on the ground, but it is more unlikely to suppress "real" stall warnings, as it would require multiple systems to fail.

Alternatively, if you are worried about one broken sensor resulting in erroneous stall warnings, you could require only two of the three conditions to hold for the stall warning to be disabled. That's still better than only one condition.

My question: Why do aircraft not use multiple criteria, such as suggested above, for disabling the stall warning?

The point of my question is that you want the stall warning disabled when the plane is on the ground. Whether or not the plane is on the ground should be rather obvious and you should not need to try guessing whether the plane is on the ground based on airspeed alone. There are plenty of other sources of information that can indicate whether the plane is on the ground.

• Where does it say it's a contributing factor? It says that you need airflow over the AoA vanes, and if you're under 60 kts, the AoA reading is likely to be invalid, so there's no way to tell if you're in a stall (pg 44). I think the simple fact is that if you're in the air at 60 kts, recovery is highly unlikely. – user71659 Mar 20 '18 at 1:07
• Also, see fig 11 on page 44 and the note. It uses all three measured airspeeds to determine the stall threshold, which is dependent on airspeed. So the idea is to use triple redundancy to always have good data, instead of making up potentially dangerous invalid data. – user71659 Mar 20 '18 at 1:14
• @user71659 I’m suggesting more redundancy, not less. AF447shows how triple pitot tubes can all simultaneously fail - they all point forwards and have the same vulnerability to ice. – Thomas Mar 20 '18 at 4:03
• You're getting two different things confused. The stall system already takes the worst case mach number (pg 44 note). Increasing the reliability of airspeed instrumentation is a different question. The fact of the matter is, if airspeed is really under 60 kts, then you are dead, there's no reason to warn about stalls in that case. – user71659 Mar 20 '18 at 4:13

## 1 Answer

A review of the stall warning system was recommended in the final report:

EASA require a review of the conditions for the functioning of the stall warning in flight when speed measurements are very low.

Here is the outcome of reviewing the "conditions" (plural):

EASA has reviewed the conditions for the functioning of the stall warning in flight when speed measurements are very low. The stall warning is designed to be efficient in a realistic and recoverable flight domain. It is also designed to avoid spurious triggering, hence the 60 knots stall warning inhibition threshold implemented on a large number of large aeroplanes. At very low speed (below 60 kts), corresponding to exceptionally high angle of attack, the aircraft flight characteristics are unknown. Flying below 60 knots is out of the flight envelope. Certification requirements are not established to ensure safe flight outside the aeroplane flight envelope.

Status: Closed — Category: Disagreement

While yes there are multiple factors in any one accident, the stall warning design was not one of the listed contributing causes. The biggest factor is that the plane was stalled by the crew. Tragic as it is.

Adding criteria to a system adds more points of failure to that system. As an example, if one RA (radio altimeter) is inoperative, the crew will have an erroneous stall warning on ground that they can't silence. The solution will be to pull a circuit breaker, but this might lead to a flight conducted with no stall warning.

• The proposed logic would not cause erroneous stall warnings in flight as your last sentence suggests. – Thomas Mar 20 '18 at 1:45
• The quoted text does not seem to consider using other inputs to determine whether or not the plane is on the ground, only airspeed. – Thomas Mar 20 '18 at 1:47
• @Thomas - The proposed logic [to me] reads: the stall warning is disabled when A AND B AND C are met. So if A or B or C fails, the stall is not disabled. Nevertheless the conclusion reached via EASA remains and the issue with multiple criteria remains valid. RE second comment: Note that it reads: "review of the conditions". Happy to discuss this in chat later. – ymb1 Mar 20 '18 at 1:52
• @Thomas - Redundancy is having more of the same thing. The last paragraph: for the stall to be disabled on ground, the suggested 3 different inputs need to be working correctly and sending the correct signal for the AND logic. If one system is inop (quite normal for planes to be dispatched with inop systems), then the logic collapses and hence the example given. – ymb1 Mar 20 '18 at 4:22
• @Thomas - The duration quotation is from the final report. And the outcome of the system's review is as objective as can be. IMO (as is EASA), I don't see a benefit from the proposed idea, as I already wrote in my answer, the stall warning is not listed as a cause. This is my final answer, but let's wait for others. – ymb1 Mar 20 '18 at 4:48