Stall Speed Bug?

[Mace2004]

Originally posted by hitech
Mace2004: Don't confuse aces high buffet with the real thing. I assume we can agree to define stall to be the point at which  more AOA produces  less total lift? In AH the buffet starts at that AOA .

Sure, that's certainly a valid approach but it sacrifices a bit of realism.  Max performing an airplane means to get the absolute most of the turn performance and it's not unusual to ride in mild to moderate buffet (depending on the plane) to get best turn performance while using the seat of the pants feel for how close you are to stall.  This is especially important when you're twisting all around the cockpit keeping track of your opponent and you're not looking at your airspeed or AOA indicator.  If it's as you're describing then I'd think turn performance would suffer anytime there was any buffet at all.  The difference would probably be noticable to some of the pilots that fly with the stall limiter off but maybe not to the average Joe.  Of course you have a stall buzzer that gets louder the closer you get to stall, maybe this is enough warning.  Another thought though, since there are some airplanes known for a nasty stall with little/no warning wouldn't tailoring the onset of buffet to each plane be more representative?

This is what I got today in the F6F after a lot more work, using buffet onset as stall and tweaking my technique.  The numbers I present are the avg of 10 stalls in each configuration.  One of the biggest "tweaks" was for the power-on stalls where I went to idle to get behind the curve and put just enough rudder to roll the airplane a bit to the right.  Adding power rolled the airplane back to wings level flight.  This let me get ahead of the roll.  Just for comparison I put in Grumman's numbers (in parenthasis) and the delta.  Power-on was 44inchesMP:

Power-off, landing: 71.9 (69) Delta: +2.9
Power-on, landing: 75.0 (63) Delta: +12

Power-off, clean: 89.5 (77) Delta: +12.5
Power-on, clean: 91.6 (70) Delta: +20.6

Again, while all stall speeds are above Grummans' the power-on stalls are much higher as Widewing contends.

I'll mention that power-on, clean is a very tough stall to perform, the airplane likes to just slowly nose over with full aft stick and the plane picks up speed.  You have to be around 95-97mph just to get behind the curve and the plane will many times power out of it.  It also has the broadest spread of stall points, from a low of 85 to a high of 97.  The difficulty in actually getting the plane to stall reliably in this configuration also resulted in the greatest difference in pitch angle at the stall which accounts for the wide variation in results.  This makes the results of the power-on, clean tests less reliable because of the inconsistency and lack of solid repeatability.  Also, just a side note but I noticed that on numerous occasions I got the stall buzzer and buffet but it disappeared below 70mph.


Mace

[Pyro]

The buffet actually kicks in when any portion of the wing begins to stall- not necessarily at peak lift.  However, it's a razor thin edge to maximize turning while in the buffet.  If you go deep into the buffet you will degrade your turn.

I gave the autopilot full control authority and did the following power-on and power-off stall tests at 1/5 and 1/10 speed to take screenshots.  Test subject was an F4U-1A at full internal load- 12,904 lbs.  Atmospheric conditions were hard-coded to SL which is why TAS and IAS always match in the screenshots.

Power on test.  Speed auto pilot was used with the speed setting constantly reduced until stall horn was blaring.  Speed was then reduced in 1 mph increments and allowed to settle before the next reduction.

Lift vectors will turn white when max AOA is exceeded.  Plane is in steady flight at 102 mph.  As soon as I input 101 as the target speed, the stall began.  Plane snaps pretty cleanly in this condition.

[Pyro]

The following series is a power off stall.  Autopilot set to level, throttle brought back to maintain flight at about 115 mph before being fully closed.  Notice how the plane wallows around in the stall unlike the clean break in the power on stall.  Also notice that the stall speed is higher than the power on stall.

[Badboy]

Hi Widewing

I just watched your films, and I think I can see one possible explanation for the figures you are quoting. However, I can't be certain, because the film doesn't show the buffet or stall horn, so it isn't easy for me to pick out the precise point when you are recording the stall speed.

However, I've been paying attention to the load factor, and it looks to me as though you might be measuring accelerated stalls, because the needle doesn't appear to stabilize at 1g, it fluctuates a little above and below.

Even if it did stabilize at 1g, there are two other points that have a bearing on this. Firstly, the definition on the scale isn't fine enough to be accurate, and secondly, even if it were, we know that the cockpit graphics aren't always drawn accurately enough for precise readings.

Does that even matter? Unfortunately it does, and it is important, because even tiny differences in the load factor have a significant effect on the stall speed. For example:

Suppose you were measuring the stall speed on an aircraft that had a stall speed of 90 mph. If you stalled at 1.2g the stall speed would read 99 mph. If you stalled at 1.1g the stall speed would read at 94mph. If you stalled at 0.9g the stall speed would read at 85mph.

So if even if you could guarantee holding the load factor within 0.2g you are probably only going to get a stall speed within 10mph of the true value, and even getting within 0.2g is probably optimistic using the reading in the cockpit.

There is another way to measure stall speeds, that I believe yields far more accurate results (doesn't depend on reading the gauges) that I use when testing.    

Hope that helps...

Badboy

[Tilt]

Originally posted by hitech
In AH the buffet starts at that AOA .

HiTech

geez have I had it wrong or what............... the amount of time I spend in buffet when in a turn fight.......to now realise  that that is not optimum turn rate or minumum turn radius...............


Great Graphics Pyro............ thanks for sharing HTC

[hitech]

Tilt: I miss spoke, Buffet starts when any 1 wing section is stalled.

HiTech

[Mace2004]

Holy cow Pyro, those are like test pilot porn.  Ahhhummm....I need to light some candles, put Barry White on the stereo, open a nice bottle of Merlot and spend a little time alone to study them.

Seriously though that's alot of data. Good news that the buffet is correlated to a local stall vice the entire wing.  Just out of curiosity can you tell me what the data are on the wing cut lines?  Looks like local values for lift and alpha but don't know what the last is.  Some measurement of drag?

Badboy, I don't think Widewing is taking the speeds from the aircraft instruments, he's probably doing like I am and using the clipboard values.  Regarding the g's I averaged my numbers over 10 stalls and this would pretty much eliminate any variation of such small g deltas.  As far as the specific accuracy is concerned, I think you're assuming a much larger error (10mph) than should be realistic.  Pilot technique and perception (i.e., what constitutes a stall?), slight manufacturing variations, calibration of instruments, even the quality of the paint job could all affect the numbers but the only variation in AH would really be pilot technique and perception.  Also, don't forget that the stall speeds are for an average pilot flying an average plane on an average day.  It wouldn't be very useful to publish stall speeds that can only be achieved with a highly instrumented aircraft flown on a specific and precise profile.  If plus/minus .2g is how the plane flies then the average is what's going to be published.

Mace

[Tilt]

Originally posted by hitech
Tilt: I miss spoke, Buffet starts when any 1 wing section is stalled.

HiTech

Some (slender) hope for me yet then............

By section do you mean any of the grouped coloured  lines on Pyros graphics or do you mean one wing in "net" stall

[dtango]

Pyro - thanks for posting the results.  That is indeed just a few more lift vectors more than AH1 (at least what I remember seeing in the past in Grapevine)!

Badboy - would you mind sharing your stall testing techniques?  I had always assumed in the past you were doing level power-off stall flight tests but that's obviously not how you are doing it.  Might be fun to try to figure out the physics you might be applying but alas my brain cells are pretty much spent at work and then my family consumes whatever ones I have left when I get home!  I've actually tried using the "wind" feature offline as a way of doing it but only tinkered with that approach a few times but never spent the time to try and refine that process!

Mace- the difference is pretty much in the technique.  With the auto-pilot with full control authority it must be maintaining much finer controls than we are able to do by hand at normal time.

Tango, XO
412th FS Braunco Mustangs

[Badboy]

Originally posted by Mace2004
Badboy, I don't think Widewing is taking the speeds from the aircraft instruments, he's probably doing like I am and using the clipboard values. Regarding the g's I averaged my numbers over 10 stalls and this would pretty much eliminate any variation of such small g deltas.  As far as the specific accuracy is concerned, I think you're assuming a much larger error (10mph) than should be realistic.

Agreed, it is always better to use the clipboard values, I assumed that was what Widewing was doing. And even though small differences in the load factor does make a significant difference to individual stall speeds, you are of course, absolutely correct that averaging several tests should get you much closer to the true value.

Interestingly, the results Pyro posted are in good agreement with the EM diagram posted earlier, which shows a slight reduction in stall speed with wep for the F4U1. But I've known that the stall in Aces High began at the onset of buffet and that flying deep in buffet degraded turn performance, I published as much in a .pdf file some time ago and I've been conducting my own tests accordingly, which may be why I get similar results to Pyro.  

So, if it isn't the variation in load factor, do you think the reason your tests and Widewing's show the opposite result to Pryo's and mine is just that you have been using the wrong visual cue for the stall?  It will be interesting to to see what results you get from your next set of flight tests.  

Badboy

[Stoney74]

What's the difference in Pyro's speeds and the historical POH speeds for the F4U-1A?

[Mace2004]

Originally posted by Badboy
So, if it isn't the variation in load factor, do you think the reason your tests and Widewing's show the opposite result to Pryo's and mine is just that you have been using the wrong visual cue for the stall?  It will be interesting to to see what results you get from your next set of flight tests.  

No Badboy I don't think that's it. My last set of tests where I averaged 10 runs in each configuration and using initial buffet as the stall indication I still have reversed numbers.  The only differences between what Pyro is doing and what I've done is that obviously he's using the F4U and I did the F6F, he's heavier (100% fuel vice 25%), and he's letting the AP fly the plane which will react quicker than we can.  If you've tried to duplicate this and have gotten similar results to Pyro's I can only assume there is some difference in technique or more likely a difference in the way our flight controls react.  

Here are my current theories:

AP vs pilot/control system reaction times.  Think of it this way, in RL, if a wing begins to drop off a bit the AP reacts very quickly with a very small adjustment.  I don't know how AH does it, it could be that AP works simply by saying the wing is level so it's level, there may be no AP action actually taking place.  When we're doing the flying we have a delayed reaction time and the wing will drop further so our response is going to be late and therefore, by necessity, relatively large.  Whether we're using rudder or aileron each work by increasing the local AOA.  Since the AOA is affected most at the wingtip we may actually be initiating the stall followed by rapid stall progression.  Of course the same can be said about pitch, particularly in light of your observation of the delta G.  If stalls are consistently started by control inputs that's what we'd report out to the fleet and the speed that occurs would become the effective stall speed published even if it would stall slower with absolutely perfect control.  Who knows what a real airplane would do with the precise AP control that AH has but would it really matter to the pilot?  No, it wouldn't.  He'd just know that he can't fly the plane as slow as the computer can and the computer controlled stall would be irrelevant to him.

Weight/CG.  Since Pyro did his test at 100% fuel, not only is he heavier, his CG is probably different.  Changes in weight without changes in CG will just raise/lower both stall speeds in proportion but changes in CG effects the Cl at which stall occurs but I just don't think this change could be significant enough to reverse the stall speed relationship.

Delta V/Delta T.  I considered the affect of rate at which stall is approached but discounted it.  Basically the more rapid your deceleration to stall the lower the stall speed. If this were a problem in these tests we'd actually be artificially lowering the power-on stall speeds and making them closer to what Pyro got but then he's approaching stall very slowly so it doesn't make sense that this could be the culprit.

Engine thrust.  I considered this and discounted it also.  Basically, at the nose-high attitude required for the power-on stall you have a thrust component contributing to lift.  This would also artificially lower the stall speed which isn't the problem so this isn't it either.

These are only my current assessment.  If I had to pick one, I'd go with AP vs pilot but I'll look at it again, maybe I can find something I'm doing or haven't considered.  BTW, love your charts.

Mace

[Widewing]

Pyro, after looking at your data, I find that I can duplicate your clean power-on stall speed both in auto-angle and under manual control. If I steepen my climb angle a few degrees, I can get a slight lower speed of 98 mph.

Where I find a discrepancy is in clean power-off stalls.

Pyro's images show a big rate of descent. It appears that the aircraft was in level flight at the beginning of the test.

If I do the same test, but set an angle (about 8-10 degrees nose high) that produces a stall as the rate of climb reaches approximately zero, the air speed is 9 mph slower than what you show. Why does this occur?

I watch the films at very slow speeds and wait until I see a marked increase in airspeed to record the lowest speed immediately before that happens.

I also tested the F6F-5. Using auto-angle, power-on, I was able to get 92 mph at stall. This is consistent with every test I've done, in manual, with and without combat trim. I obviously don't the ability to decrease speed in 1 mph increments.

Power-off, in auto-angle, using a slight nose up attitude that results in a stall at a rate of descent of 800 fpm, I recorded 87 mph. In manual flight, I recorded a speed of 85 mph. If I start out level (in auto) and pull off power, I see a stall at about 90 mph, but the plane is dropping like a brick.

Pyro, set the nose about 5 degrees high (I use auto angle) and see what you get. Stall speed should be slower than when starting level. Then repeat the test with the nose about 10 degrees high. This should produce an even lower speed.

So, my tests tonight demonstrates a split of 2 mph best case, 7 mph worst case.

How about the difference between Navy stall data for the F6F-5 and that in the game?

Power-on stall speeds for F6F-5, clean configuration.

AH2: 92 mph (25% fuel, approximately 11,300 lbs)

Power-off stall speed for F6F-5, clean configuration.

AH2: 85-87 (or even 90) mph (25% fuel)

Navy Stall data from NAVAIR 1335D, Standard Characteristics of the F6F-5:

Power-on clean: 72.2 knots (83 mph)
Power-off clean: 79.2 knots (91 mph)

Navy stall data for full load, 12,420 lb from Navy test report NA-83/44177:

Power-on clean: 96 mph
Power-off clean: 98 mph
Power-on Landing: 79 mph
Power-off Landing: 84.5 mph

Grumman stall data "normal takeoff weight", actual weight not specifically defined:

Power-on clean: 71 knots (81.7 mph)
Power-off clean: 77 knots (86.6 mph)

There is apparently no general consensus of agreeing data for stall speeds of the F6F-5. The differences are substantial. Training films for the F6F-3 and F6F-5 state stall speeds well below that of the manufacturer. One can see why it's hard for you to select performance criteria....

My regards,

Widewing

[Widewing]

Originally posted by Mace2004


Delta V/Delta T.  I considered the affect of rate at which stall is approached but discounted it.  Basically the more rapid your deceleration to stall the lower the stall speed. If this were a problem in these tests we'd actually be artificially lowering the power-on stall speeds and making them closer to what Pyro got but then he's approaching stall very slowly so it doesn't make sense that this could be the culprit.

Don't discount it yet... I think you hit on it here. I can lower the stall speed significantly by increasing the rate of deceleration and do it consistently.

If I raise the nose 5 degrees above level, stall speed drops by 3 to 4 mph. I described this in my post above, and asked why this is the case. Has Pyro modeled stall characteristics so accurately that Delta V / Delta T behavior is modeled too? It seems so.

My regards,

Widewing

[Pyro]

Ok, here's some power-off clean tests using auto-angle instead of level.


This series is at 7.3 degrees.










This series is at 20 degrees.