Stall investigation [Archive]

macho macho man

11th Jun 2005, 00:50

As part of my job a few times a year I have to compare stall speeds of aircraft in various configurations against their charted speeds. The direction out of our operating instruction for the aircraft (P-3) with respect to the rate of deceleration is that the ideal deceleration is 1 knot per second, which seems reasonable and manageable.

I have heard conflicting views on this: one technique, as apparently taught at the USN Test Pilot School, recommends a maximum of 1 knot per second - I think to allow smoother investigation of deceleration/stall characteristics, while a second technique (Jay Beasley's theory (http://www.vpnavy.com/beasley_07sep99/beasley_23.pdf)) advises avoiding a slow deceleration (even 1/2 knot per second) since doing so would introduce a greater vertical component to the flight path, and thus a higher angle of attack resulting in an accelerated stall.

I confess I don't understand the second view - while I understand that a slower decel will result in more altituded being lost due to this "vertical component", AOA is AOA. I'd think that slowing at 1 knot per second or 1 knot per 20 seconds would not change the ultimate buffet onset speed. While I'm happy with the way I currently investigate stall buffet I'm less happy with my understanding of basic AOA, if Beasley was correct.

Ideas? Thanks in advance, MMM

Mad (Flt) Scientist

11th Jun 2005, 07:33

1 knot/sec nominal is the accepted basis for stall speeds for performance purposes, for a number of reasons.

One, on aircraft with artificial stall protection systems there is indeed often an 'angle of attack rate limit' input to the system, which will cause the stick pusher and/or stick shaker systems to activate at a lower angle of attack for a more dynamic entry. The purpose of this is to provide the necessary degree of protection against a higher peak angle of attack resulting from a higher entry rate; for example, suppose my aircraft has benign characteristics up to an AoA of 25 degs and a maximum aerodynamic CL at 20 degrees. Obviously I want to take advantage of my aero CLmax if I can, so let's say I set the stick pusher to fire at 21 degs AoA. Now, if a 1knot/sec entry rate equates to 2 deg/sec entry rate and the elevator effectiveness at stall AoAs and the control system behaviour is such that I have a nominal 1 second 'overshoot' in alpha after the pusher fires, I will see a peak AoA of 23 degs (21+2*1) which is still inside my 'good characteristics envelope (AoA<25) so everything is fine.

Now I do the 3kt/sec entry rate dynamic stalls (and yes, the rules keep changing) and find that my overshoot is indeed proprotional - i.e. 6 degs alpha. That would result in a peak AoA of 27 deg and whatever "nasty" was waiting for me at the 25 deg cliff-edge just grabbed me. One (common) solution is to add an entry rate term to the firing angle calculation - in this case, I want to make sure I fire at 19 degs or less, such that the peak value is 25 or less and everything stays tickety-boo. Something like "phase advance term = AoA rate - 3" would work quite nicely - with a 6 deg/sec entry the adjusted firing angle would be 18 (21-3) and the peak 24. Everyone goes home happy.

That covers artificial systems. However, even with a natural stalling aircraft entry rate matters, because the stall characteristics and development is going to be different. Suppose, for example, my aircraft has a natural wing stall at 25 degs AoA, but has vicious roll if stalled there (very sensitive outer wing perhaps). In order to 'tame' the characteristics a stall strip is added to the inboard leading edge, which is placed so as to induce an inboard wing stall at 21 degs AoA. Again I do the 1kt/sec stall, get a 1 sec overshoot in alpha, and the stall on the outboard wing does not get a chance to occur, so everything is again fine. I know do the 3kt/sec stall, and this time the overshoot takes me above the critical alpha for an outboard stall, and I find the aircraft suddenly rolling violently during the stall recovery. In this case I might have to adjust my 'wing dressing' in order to address the characteristics. Note that due to the variation in profile along the span, such spanwise differences can occur regardless of the dressing on the wing.

Finally, entry rate can also affect the stall developement at a given wing station, even if there is not another station with a stall waiting to ambush us. You may have a wing dressing, for example, which provides for warning buffet of suitably increasing magnitude as the AoA increases in a normal stall, followed by good handling, but this may be tied into the way the stalled airflow develops on the wing; while the airflow at a given AoA may be stable if left for some time, it's possible that the rate at which the stalled flow spreads over the wing is slow relative to a high entry rate. If that is the case, a slow entry rate may result in a very different flow pattern than a high entry rate, with corresponding differences in both handling and stall speed.

Finally, I don't understand the logic of the second methodology. Even if a higher gamma is obtained (and I don't see that even that is a given unless you set up the test to make it so), that won't change the AoA at which the stall occurs - it will change the PITCH ATTITUDE instead. Consider this case. If I conduct a 1 kt/sec approach to the stall, but STOP a Vstall+1knot. Is the stall speed now going to vary wildly if my final entry rate is essentially zero? As long as the entry rate is low enough to prevent the dynamic flow development effect, and low enough to prevent system behaviour being an issue, the stall AoA should be the same. Typically, 2kt/sec or less would suffice for that on most aircraft, I see no real need to aim for anything special other than the nominal 1kt/sec.