Understanding Aerodynamics of Stalls

Recently, most commercial transport airplane manufacturers have been revisiting their FCOM procedures for “stall recovery” (actually, procedures avoiding that an approach to stall turns into a stall). This may be related to the spate of recent accidents in which commercial airplanes have been stalled: Colgan Air in Buffalo, Turkish Airlines in Amsterdam, XL Airways in Perpignan. Such a spate of loss-of-control (LOC) accidents is a sudden new development in aviation accident statistics. People are concerned it might signal a trend and are looking for possible causes of this trend, if it is one.

A discussion on such matters started on the Professional Pilot’s Forum PPRuNe on a thread with the ungrammatical title of New (2010) Stall Recovery’s @ high altitudes. I agree with the moderator, who goes by the handle of John_Tullamarine, that the discussion has been stimulating, although I had my doubts as it started, which readers of the thread may observe.

The discussion has been enlightening in a number of respects. One aspect which startled me is the degree of understanding of stall – what it is, when it happens, and its functional relationship with other aerodynamic parameters. The stall is one of the most important, if not the most important, phenomena with which pilots must cope (preferably by avoidance). As is buffet. I conclude that such understanding amongst line pilots could be easily improved. There are graphs used by aerodynamicists; they are all more or less the same shape no matter what the airplane. You can find them in any intro-aero book, say John D. Anderson Jr.’s Introduction to Flight, or Richard Shevell’s Fundamentals of Flight, without any numbers on them, as well as in many FCOMs with numbers on them. Why are they not studied in type training and the knowledge tested?

It could be – has been in the thread – argued that “pilots don’t need to know” such things. As off and on a professional educator for the last few decades, I have participated in enough discussions about what technical practioners need, respectively don’t need, to know. I have seen what happens, at enough places. We as a profession are now – have been for a decade or two – giving computer science degrees to people who can’t really program very well, at least not according to the standards we used to have. Do I think this is – ever – a good idea? No, I don’t. Do I think people should be professionally flying complex airplanes without understanding the aerodynamics presented in the FCOM? No, I don’t. Although I imagine not everyone will agree.

My practical philosophy of education is as follows. My default answer to what people need to know is: everything. That said, there are practical limitations (of ability, of time) which entail prioritising knowledge and intellectual skills in using that knowledge (which, while related, are not the same thing).

Can we reduce such knowledge to algorithms, to operational instructions, as has been suggested in the thread: “if this happens, do Y”? I am sceptical. Choosing the correct action as a pilot requires appropriate situational awareness.

There was, for example, considerable debate about tail-plane stalls and training syllabuses following the Colgan Air upset in Buffalo. Colgan Air used a NASA video about stalls in icing as a training aid, and this video emphasised so-called tailplane stalls due to icing, for which the remedy is apparently inconsistent with the action required for a main-wing stall. The Q400 in the Buffalo crash is not at all susceptible to tailplane stalls, and is equipped with a stick pusher to prevent main-wing stalls. However, the pilot flying pulled the stick back, overpowering the stick pusher repeatedly, which is exactly the wrong thing to do if the main wing is on the point of stalling (it is, after all, the purpose of the stick pusher to do the right thing in this situation) but might be appropriate for tailplane stalls. It was therefore questioned whether the pilot had appropriate awareness of the aerodynamic situation the aircraft was in, and it has been concluded that he did not.

Having appropriate situational awareness requires understanding the phenomena, as well as understanding the limits of understanding. Needing to distinguish between what some people call “stalled” (namely, at or just beyond the maximum value C_L_Max of the coefficient of lift, C_L, when large parts of your wing may still be flying) and “fully stalled”, for example (when none of your wing is flying). The question arose in the thread whether one may use ailerons to lift a dropped wing at stall. The obvious answer is that you can if that part of the wing is still flying, but you most definitely should not if it is not. How do you tell which situation you are in? Trying and seeing is not a wise option.

Consider, for example, FCOM procedures concerning “stall warning” on a popular large airplane after lift-off (A330, 3.04.27 P 5a):

At the same time:

This assumes that you are at high angle of attack (AoA) but not yet stalled, and that the ailerons are still flying. The stall warning may also go off at high altitude, in which case: “relax the back pressure on the sidestick and reduce bank angle, if necessary”. In both cases, it is assumed that the wing is flying, but that bits of it are telling you they might not for much longer, and you need to back away from that point. These procedures obviously won’t help much at all if your nose gets to be way up in the air at 45°-60° of pitch, as happened at Perpignan with a related airplane.

The answer to telling which situation you are in is probably found in a good intuitive understanding of the aerodynamics in the FCOM, and for that one needs a good basic understanding of aerodynamics in general. One illustration of this is the suggestion that was made in the thread on a potential means of discriminating stall buffet from Mach buffet: the feeling of the frequency of the buffet.

This also illustrates the limitations of simulation, a topic on which it seems not all thread contributors are clear. It seems that many people still seem to think that flight simulators, including the expensive moving kind used for airline pilot training and recurrency, are veridical around upset scenarios. How on earth do these people suppose simulators can reproduce veridical buffet? That some aerodynamicist has sampled the frequencies of buffets in the wind tunnel, and given that to some simulator programmer to reproduce, as well as some engineer to make sure that none of it coincides with the resonant frequencies of the simulator? And most of those wind tunnel models that generate the data fly without horizontal tail pieces; what is the effect of the tail? Mostly, one doesn’t actually know, but extrapolates from one’s experience as an aerodynamicist. I feel that a basic understanding of aerodynamics would cure many illusions about the veridicality of flight-simulator behavior outside the normal flight envelope.

Whatever one thinks about what pilots should know or not know, it seems to me a good idea to clean up the vocabulary, suggested through the following examples.

“Stall” is a term of art: for example, sometimes it means the same as “at C_L_Max”, and sometimes it means “the point at which buffet is severe enough to discourage further increase in AoA” (cf. the definitions used in the airworthiness certification document, CS 25). Does being at or over the stall mean you have no lift? No, actually you may have more lift than in most other regimes of flight (just over C_L_Max) even though you might be shaking severely, or you may have much less (AoA way over that for C_L_Max).

Another terminological inexactitude resides in the terms “low-speed stall” and “high-speed stall”. The first refers to the situation in which the AoA is too high for the speed; the latter often refers to a transsonic overspeed situation, in which lift is reduced because of the formation of shock waves over certain parts of the wing, which waves, because they form at or near the leading edge, reduce lift forward and thereby move the center of aerodynamic lift rearwards, leading to a nose-down moment about the center of gravity of the aircraft, which gives nose-down pitch or “Mach tuck”. Use of this terminology leads one to the anomalous-sounding phenomenon of the “low-speed” stall at “high-speed”. Maybe the terminology “high-alpha stall” or “high-AoA stall” would be preferable to “low-speed stall”, and to use the word “transsonic” rather than “high-speed” to indicate effects of shock waves on lift?

Another vocabulary hang-up occurred in the discussion on the thread of V_s1g, or stall speed at 1g. Is it a constant speed or not? If not, with which aerodynamic parameters does it vary?

V_s1g would occur when lift at C_L_Max is equal to weight (W). Lift = q x S x C_L, where q is dynamic pressure and S is an area term usually taken to be the area of the wing planform. So at V_S1g, W = q x S x C_L_Max. Given that q = ½ x density x V^2, we can solve for V: V_s1g = Sqrt( (2 x W)/(density x S x C_L_Max)). S is obviously constant for a given airplane. What about C_L_Max? If you can ignore compressibility effects (i.e., below about 0.3 Mach for most wings) then C_L_Max is effectively constant, as is the AoA at which C_L_Max is achieved.

Now consider density. Air density obviously varies with altitude, indeed with the properties of the atmosphere on the day and at the place. So if one wants V_s1g to represent the true airspeed (TAS), then this obviously varies, but with a bunch of parameters not measurable with equipment on board most commercial aircraft. However, aerodynamicists like to talk Equivalent Air Speed (EAS), in which inter alia density is defined as sea-level standard-atmospheric density, 1.225 kg per m^3 (kg.m^(-3)).

So V_s1g, as EAS, varies only with (the square root of) weight. Weight obviously varies (with load, fuel burn and so on) but it is not an aerodynamic parameter, and is usually considered constant when talking aerodynamics. It follows that V_s1g, expressed as EAS, is constant.

However, V_s1g, as indicated, say, in the A330 FCOM (3.01.20 P7) is expressed in Calibrated airspeed (CAS), which is the pitot-static-measured airspeed corrected (usually digitally) for the effects of how the sensors are positioned in the air stream, and expressed in CAS there is a correction for pressure altitude, starting at about 20,000 ft for lighter weights, and going down to about 5,000 ft for heavier weights.

So, as a “practical” matter, is V_s1g (at fixed weight) constant or not? As an aerodynamicist, liking EAS, one would say yes, as a pilot, preferring CAS because that is what one sees on the airspeed indicator, one would say no. That could be a source of genuine confusion at times.

A more obvious but less insidious vocabulary hang-up is Mach number. Is it a speed? Strictly speaking, no. It has no units (it is a ratio of speeds: airspeed to the speed of sound, which varies with air temperature); whereas speeds have units of length per time unit (m. s^(-1) or ft . s^(-1) ). However, in response to a question “how fast were you going?” one might well respond “at 0.8 Mach”, and indeed Mach is used in preference to airspeed to adjust for many situations at high altitude. For example, limiting dive is expressed as both speed and Mach number, as is turbulent-air penetration, maximum operating (max. cruise), and so on.

Other vocabulary hang-ups occurred in the thread when talking about “approach to stall recovery” and “stall recovery”, and these I feel are insidious. Some correspondents (including the thread originator) insist they have been practicing “stall recovery” in an airplane with a stick pusher, despite the obvious point that if the airplane has a stick pusher and you respect the pusher, it is not stalled. Indeed, many “stall recovery procedures” are more accurately described as stall avoidance procedures, or approach-to-stall recovery. Surely such confusion would be resolved through a little aerodynamical knowledge and some common sense about safety-system design?

One correspondent, when asked repeatedly whether he thinks that test pilots have been going up and stalling Airbus airplanes, in order to rewrite the “stall recovery” FCOM procedures (actually “Stall Warning” in those for the A330 referred to above) and to calibrate simulators, wisely declined to answer. As a veteran pilot, with the handle 411A, said, Has anyone here actually stalled a large swept-wing airliner? I[f] so, what were the results?. Another, Airclues, replied In the early 80’s I was co-pilot on several C[ertificate] of A[irworthiness] air tests on the Boeing 747 when a full stall was completed (I believe that the UKCAA was the only authority that required this) and described his experiences. In other words, actually high-alpha-stalling large commercial aircraft, even for certification, is ancient history. I very much doubt it was done just to rewrite stall avoidance procedures and calibrate simulators.

A useful discussion indeed, but I suspect it will take more than a pilots’ forum thread to sort these issues out.

Leave a Reply