Jump to content
Sign in to follow this  
JtD

Flight attitude of P-47D

Recommended Posts

When I was looking at the flaps and drag of the P-47D in game some oddities came up. I think it's worth addressing one, which is the attitude of the P-47, both flaps up and flaps down. The P-47 wing has a Republic S-3 airfoil which is thicker at the root than it is at the tip. Should be in the region of 15% tapered down to 9%, however, I haven't actually seen data for the root and tip as such. I'm also taking NACA up on the fact that the S3 is fairly similar to the 230xx series, which is also what some aerotools suggest, at least as long as you're not near maximum lift coefficient. There the 230xx is better. The P-47 wings are inclined by 1° against the thrust axis, and the thrust axis is pretty much parallel to the prominent joint line running along the fuselage.

 

image.thumb.png.d531b1f7de8066aa1a21a417a693ec48.pngimage.thumb.png.32834af47094b04c5d7d5c2a8e94588f.pngimage.png.4e49ab30a81aad05e923f5de5e8a1ea9.png

 

The NACA 230xx series is one of the best documented WW2 airfoils, which is why I like to use it as reference. Since there's very little data available on the S3, there's not much choice anyway. Already visible in the Cl v Alpha chart attached above, the airfoil(s) provided a lift coefficient a little in excess of 0.2 at 1° incidence (alpha). You can also take other sources, but you pretty much always get in the range of 0.2...0.25. I'm picking 0.21. So that's what you'd get with the P-47D levelled out (i.e. thrust axis at 0°, meaning the panel line at 0°).

 

The P-47 300sqft reference area consists of three relevant sections (well, for my purpose), this is the fuselage, taking up about 15% of the reference section, the flaps section, with about 50% of the reference section, and the outer section with ailerons at wingtips, with about 35% of the reference area. So the wing area actually proving lift is just 85% of the reference area, for which typically all coefficient for an aircraft is given. Based on the 0.21 chosen above, and the 85% effective area, means something around 0.18. The gradient, meaning the change of lift coefficient with change of angle of incidence, can also be taken from several data sets and is around 0.1 per 1°, and one would also only get 85% of this, so about 0.8/° as a low estimate.

 

The P-47D comes with powerful flaps, as we all know. Technically they are Fowler flaps, but they don't extent backwards as much typically is associated with Fowler flaps. Lacking data on the P-47 as such, we again have to look at wind tunnel data. The mechanics of these flaps are pretty close to what NACA has tested with several slotted flaps, closer than to the Fowler flaps they tested. These slotted flaps also move back and down a bit, while applying the angle, and their performance in terms of lift vs. drag is much better than what is seen on say split flaps used on a Fw190. Anyway, below flaps characteristics of a slotted flap, a Fowler flap and another slotted flap on a NACA 23012 airfoil. The last one is my favourite, as it comes closest to the P-47 design, but unfortunately there's only limited data. (Btw, note the 0.1/° cl increase in all three data sets for the plain airfoil.)

image.thumb.png.4907ee6f036bb6fbdcee1c244b451c1b.pngimage.thumb.png.3102feaa81e3e16e1e8707356ab73fee.pngimage.thumb.png.c4a95da1100baf8345343d49e9f9b843.png

 

Anyway, what we see is a large lift coefficient increase, that means maximum lift coefficient, but also lift coefficient at the same angle of incidence/attack. I want to focus on the latter right now, and also the other way - if I keep lift coefficient constant, how much does angle of attack change. I'd like to look at a lift coefficient of 0.8 and compare the no flaps position and the 20° flaps position.

Slotted flaps 1 say change from 6.5° to -1°, Fowler flaps say from 6.5° to -5° and slotted flaps 2 say from 6° to -2.5°. The change for the Fowler flaps is huge, because the wing with the flap in that position is of about 1.25 the original size, something that doesn't happen with the P-47 flap. There's only a moderate increase in area, as is with slotted flap 2, of about a third the flaps size.

 

So, if we fly the P-47 level at a certain speed, once with flaps in and once with flaps at 20°, we need to lower the nose so that we have a lower angle of attack when we have the flaps out, to compensate the extra lift. If the P-47 was flying at normal load and at a speed of 247 km/h, it would fly at a cl of 0.8 and an angle of attack of 6° with no flaps out. If it used a wing made of the airfoil with the slotted flap 2, we'd need to lower our angle of attack by 8.5° (6° to -2.5° from the chart). However, our flaps only cover part of the wing, so it doesn't work like that. As we lower the angle of attack, the outer section of the wing will provide less lift than it previously did, and the inner section will need to provide even more. As previously said, the ratio is about 50:35 or 10:7. Eventually we come up that we only need to decrease angle of attack by 5 degrees. We're now at 1° angle of attack, the flaps section with a lift coefficient of 1.3 and the outer section with 0.4 ((1.2*10+0.3*7)/17=0.83).

 

If we check the charts we can also see that this change of pitch is less if we deploy the flaps further, it's most evident with the Fowler flaps data set where 40° of flaps get you pretty much the same lift at the same angle of attack as 20° of flaps do, but even with other flaps the lines move closer.

 

I checked the P-47 ingame by flying level, engine idle, prop to minimum revs. I measured attitude with the ruler on the screen method. It worked out pretty well in the clean condition, but flaps produce a lot of drag so the aircraft slowed down fast and measurements were difficult. Reproducing results was easy in the clean config, but hard with flaps fully down.

 

What I got is below. The dotted lines show my expectations. 50% are roughly 20° flaps, 100% about 40° on a P-47. What got me curious to run that test were two things. One was the high angle of attack require in the flaps up position throughout the speed range. I assumes that lift at 0° angle of attack would be too low, and apparently it is. Not by as much as I expected, but still. (I estimate it with 0.14 in game vs. 0.18 in real life). The other was the insane change in pitch that even partially lowering the flaps brought about. I figured that the flaps might create too much lift at the same angle of attack, but as a general observation, this does not seem to be true (only at higher speeds/lower cl's). However, the gradient appears to be quite low with just 0.7/°. This does of course have the same effect, you need to change pitch a lot when you want to change lift a little bit, giving you insane nose down attitudes when you pop out flaps at higher speeds. Additionally, as you go slower with flaps up, you need to increase pitch quite a bit. In a climbing situation, at around a cl of around 0.6, the difference already is 1.5°. Doesn't sound much, but it is very notiable.

I also find it strange that extending flaps from 50% to 100% gives pretty much the same trim change as going from 0% to 50%, even though they shouldn't. It might be down to testing inaccuracies - like I said, flaps fully down was difficult.

image.png.97c4d6eccff83e9abdaa8bfb8bcb16a7.png

Bigger chart - the attitude over speed. The nose down attitude flaps down appears to be overdone in game, the nose up with flaps up as well. This results in a too large pitch change with deployment of flaps.

 

image.png.5e175465908efad8f2a8e8b5056aca0c.png

I know changing polars pretty much means rewriting the FM, as it affects all sides of performance. Personally, if my findings were correct, I'd say it's still warranted. But then ...

 

If a dev ever happens to see this post - I have two questions:

How close did I get to the real in game values? and Do you have any better sources?

 

Edit: Added last chart.

Edited by JtD
  • Like 3
  • Thanks 4

Share this post


Link to post
Share on other sites

Just as a note, couldn't Tacview help you with measuring AoA? Since IIRC Tacview receives enough information from the game to give us an accurate AoA.

Share this post


Link to post
Share on other sites
1 hour ago, Quinte said:

Just as a note, couldn't Tacview help you with measuring AoA? Since IIRC Tacview receives enough information from the game to give us an accurate AoA.

IIRC Tacview has issues with giving you accurate AoA in Il-2, at least in some situations. I can't remember how exactly it works but I remember it coming up in a previous thread. 

Share this post


Link to post
Share on other sites
5 minutes ago, RedKestrel said:

IIRC Tacview has issues with giving you accurate AoA in Il-2, at least in some situations. I can't remember how exactly it works but I remember it coming up in a previous thread. 

 

TacView gets no information about wind, and there seem to be other issues with it giving accurate results in tight turns, but I don't think these would be a factor regarding the issues JtD is discussing.

Edited by AndyJWest
  • Thanks 1

Share this post


Link to post
Share on other sites
2 minutes ago, AndyJWest said:

 

TacView gets no information about wind, and there seem to be other issues with it giving accurate results in tight turns, but I don't think these would be a factor regarding the issues JtD is discussing.

Ah, OK, that's it. I just remember people getting some truly bizarre AOA outputs from tacview and being warned that there were situations it would not work.

Share this post


Link to post
Share on other sites
2 minutes ago, RedKestrel said:

Ah, OK, that's it. I just remember people getting some truly bizarre AOA outputs from tacview and being warned that there were situations it would not work.

 

Yup: see this thread, and this post in particular, along with my later analysis there of where other errors seemed to be arising from:

https://forum.il2sturmovik.com/topic/45722-very-high-angles-of-attack/page/2/?tab=comments#comment-758986

 

TacView usually seems to give consistent and reasonably plausible AoA results in level flight, though it isn't obvious what exactly is being used as the datum - probably the IL-2 GB model design datum, which likely corresponds with the aircraft datum, rather than the wing chord datum, since most aircraft are designed with a positive angle of incidence of a degree or so.

  • Thanks 1

Share this post


Link to post
Share on other sites
2 hours ago, Quinte said:

Just as a note, couldn't Tacview help you with measuring AoA? Since IIRC Tacview receives enough information from the game to give us an accurate AoA.

 

Tacview is a valuable helper, but I don't know how it defines zero. Might be the gun sight line or anything else. It's certainly a solvable question, though. Plus, this was a dynamic thing with strong decelleration and Tacview has some issues with that. But so do I doing things manually. However, the reason I did not even try it for this test is because I don't have it installed. It's the first time since ages I've got some time on my hand for this. Should it stay that way, I'll certainly use it in the future.

 

I also added the attitude over speed chart I wanted to post in the first place.

Edited by JtD

Share this post


Link to post
Share on other sites

I just had a look at the last chart you posted JtD and having given it some thought here is a theory of why you get a slightly different lift slope for the case with flaps deployed in-game: I would expect the game to use some sort of bound vortex panel model and this has the advantage that it captures the wings 3d effects. Now I’m focusing on the lift slope in your figures, i.e. the dCl/dalfa, and if you take the wing 3d effects into account, there is a large discontinuity in the lift at the flap joint when they are deployed. This results in a large vortex shed there which induces changed aoa’s in both the outer and middle panel. Another way of looking at this is that your effective aspect ratio goes down when you deploy flaps. The easiest way to visualize this is to consider the aoa when the outer panels are just unloaded meaning all lift is generated by the middle part which then in effect becomes a wing with a very short span. Of course it’s a bit more complicated than that since the outer panels will still work a bit like horizontal winglets but the idea still captures the first order effects I think. From aerodynamic theory we know that the lift slope dCL/dalfa is a function of aspect ratio so I would expect the lift slope to be lower with flaps meaning you need larger changes in attitude when speed is changed , i.e. just like your measurement´s in BoX shows. Thoughts?

  • Like 1
  • Upvote 1

Share this post


Link to post
Share on other sites

Well, I think that's covered with me being coservative. Below a cl over angle of attack chart of an F4U model with slotted flaps. One line is normal config, one line is 50° flap angle with aileron droop of 9.5°. The F4U comes with a NACA230xx series airfoil, but of course a completely different wing. And it's a model (1/2.75 scale).

Gradient lif/alpha: clean 1/13th, flaps down 1/12th. So 0.77 and 0.83.

Pitch change necessary for same cl of say 0.6: 10°.

Anyway, the point is that the dcl/dalfa in a 3D test is larger here with flaps out than with flaps in, as I'd expect it to be in game, but didn't find it.

image.thumb.png.d15ddcf6bd45bb61c64503493468bf79.png

 

Edited by JtD
  • Like 1

Share this post


Link to post
Share on other sites

OK, that chart was interesting: Even if the figure is for an F4U with quite a different wing planform the principle I was after should be present there as well but it looks like the partial span flap effect on the lift gradient is less than I expected which then would not explain why there seems to be such a large change in pitch attitude when changing speeds in-game with the P-47 with flaps down......

Share this post


Link to post
Share on other sites

Getting back to Tacview (reinstalled it yesterday), here's the comparison between what Tacview says and my manual measurements. I can't really be sure how Tacview calculates AoA, I assume it's flight path between the last two data samples against pitch right now. As my speed is decreasing, more rapidly so with flaps down, errors from that sort of calculation increase. OTOH, ruler on the screen isn't really top precision, so Tacview doesn't have to be the reason the difference between the flaps50 curve is larger than between the flaps0 curves. At any rate, getting this out of Tacview is a lot quicker than the manual thing, in particular as can get a lot of info at once, where I'd have to take several tests if I did it manually. So in the future I'll probably us Tacview again, maybe after finding out why the flaps 50 is a bit different.

 

At any rate, the basic statements that

a) AoA for flaps up is too high throughout the speed range

b) slopes of the curves are lower than my expectations

c) differences in AoA between the curves are higher than my ecpectations

is also confirmed by Tacview data, even though point a) to a higher and c) to a lesser degree.

 

image.png.1fdeaa03074d7f6141c8bd2348ccfe71.png

  • Like 1

Share this post


Link to post
Share on other sites

Excellent findings JtD.

 

Tried to PM you, but is says you can't accept PMs ...

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×
×
  • Create New...