Originally posted by Brooke
I've completed my study of F4U-1 stall speed in Aces High vs. the stall speeds listed in the F4U-1 pilot's manual for clean, power off; clean, power on; dirty, power off; and dirty, power on. I find excellent agreement between the manual and Aces High.
I did some math to figure out a technique that would be repeatable, that would allow steady-state measurements (which are much easier to take), and that would allow reasonably easy avoidance of changing g loads on the aircraft.
My results and analysis are posted here:
http://www.electraforge.com/brooke/flightsims/aces_high/stallSpeedMath/stallSpeedMath.html
If anyone finds errors with my analysis or data, please let me know. I've checked it a couple of times, but that doesn't mean there aren't errors, and it would be better if the analysis and data stands the test of being looked at by others.
I'll be interested to see if anyone repeats my tests or applies it to aircraft other than the F4U-1.
Brook, I agree with dtango, excellent write-up. In particular, your description of the differences in the literature is on the nose. I have also been working on refining the test technique and am now able to come within a few mph of Pyro's numbers. Changes I've needed to make to do this includes using 100% fuel and tailoring my stick response curves to allow for very precise pitch adjustments at slow speed; however, this has a detremental affect on normal flying.
Overall, I think what we're butting heads up against the "engineering" answer and the "practical" answer. An "engineering" answer derived from precise instrumentation and very specific test conditions is required for an understanding of aircraft design and may be the basis for fulfiling the contract but, from a practical standpoint, it is often of little use to the pilot in the cockpit. I'll talk about what that means at the end.
Many times you'll find "engineering" data in flight handbooks that can never be met by the average pilot flying an average airplane on an average day. I think it's probably a mix of engineering vs practical information (and occasionally some factually incorrect data) in literature that results in at least some of the confusion. The differences between calibrated vs indicated lose their subtleties when the pilot can only see indicated and is in no position to apply position error corrections. Also, errors sometimes make it into publications and are never corrected. For instance, NATOPS for 40 years has listed the roll-rate of the A-4 Skyhawk as being 720 degrees per second. That has been taught in flight training the entire time and is even the roll-rate the Blue Angles announcer used to describe the A-4 to audiences during airshows. The problem is it's wrong and not just a little wrong. The correct number is 270 degrees per second which is easily demonstrated with a stopwatch. A simple typo turned 270 into 720 and has remained there to this day.
Based on the numbers provided by Pyro and yourself I believe you've demonstrated that the modeling of the actual stall point is probably pretty accurate. I think the discussion and data provided by everyone contributing to this thread also demonstrates the difficulties and significant variations resulting from different techniques and, in particular, the use of airspeed as the principal measurement of stall when in fact AOA is the culprit, airspeed is only a secondary indicator. Variations in loading, variations in environmental conditions and instrument errors shows why the military has switched to AOA and as the principal indicator of stall. The nice thing about AOA is that it's a constant and, for the same configuration, the wing always stalls at the same AOA regardless of these other factors. For instance, in the F-14, 18 units AOA, not airspeed, always indicates the point at which rudder becomes your primary roll control rather than lateral stick. Doesn't matter what your weight, speed or altitude is, it'll always be 18 units.
Regarding your write-up, there are a few things you should look at. In your description of "Usual Method" you say that accelerating or decelerating along the plane's direction of travel doesn't affect the stall speed as long as you remain level. Actually, it does and it can be quite significant but not because of instrument lag, it's because of nonsteady state flow effects. Yes, there are delays in instrument response but if you're deccelerating at a rate that causes significant lag then you aren't going to get good data anyway. Changes in flow over the wing take some finite time and a rapid deceleration, even if remaining level, affects the stall point because the flow has not stabilized. Depending on the wing, the difference between approaching stall at -5mph/sec vs -.5mph/sec can be 5 to 10mph in observed stall speed with the faster deceleration equaling a lower stall speed. That said, Pyro's technique is probably the technique with the absolute smallest deceleration and therefore (at least from an engineering standpoint) the most accurate measurement.
Another area you should take a look at is your discussion of stalls in climbs and "Figure Climb". It does not include the vertical (i.e., lift) component of thrust. This can have a very significant affect on the result just as does G and, the greater the thrust the higher the climb angle and the more stall speed will be thrown off. To take an extreme example, assume the aircraft has a 1 to 1 thrust to weight ratio. The airplane would be able to park 90 degrees nose-high at O airspeed and never stall. In another, less extreme example, make a constant speed climb with a 45 degree pitch angle (assuming thrust is coincident to the aircraft centerline) and 1/2 your lift is coming from your engine and only 1/2 from the wing. The greater the thrust the greater the climb angle and the less useful the results are.
These are the two reasons why I've duplicated Pyro's test condition of 100% fuel. The greater weight decreases the thrust to weight ratio and makes it easier to conduct the power-on test with minimum vertical thrust component and it makes it easier to control the rate at which you approach stall. Assuming no (or small) changes in CG the results at higher weights can be extrapolated to the lower weights. Also, slight excursions in G can be negated by taking the average of multiple runs while the same cannot be said of either pitch angle or high deceleration rates to stall.
All that aside, where does that lead us regarding the AH stall speeds? We are pretty sure the specific stall speeds are accurate, at least from an engineering perspective, and they can be replicated under very specific, controlled circumstances. So here's the rub...why can't we duplicate them or even come close in the case of the power-on numbers without this engineering approach? Remember, when it comes to flight manuals (versus engineering studies) they are attempting to provide numbers which the average pilot in an average plane on an average day can use using the instruments he has in the cockpit (i.e., indicated AS). If, as we're seeing here the stall speeds are accurate but not achievable except under very exacting conditions the flight manuals would (or should) say so. One way to take these practical issues into account would be in the determination of recommended approach speeds and acceleration/climb schedules for launch but I have no information on these so don't know if they take into account a higher minimum usable flying speed; however, since they don't mention difficulty in reaching these numbers (or at least nobody has pointed out where they do this) maybe we're asking the wrong question. Is there something about the way that approach to stall is modeled that is incorrect? Is this sensitive to specific flight controllers and their setup (I suspect it is to a degree)? I don't really know the answer at this point but I do know the "effective stall speed" is much higher than the actual speed.
Mace