FW190 vs. BF109: Turn Radius (2011)

[drgondog]

Ardy, I'll make you a serious offer. If you want my copy of "Fundamentals of Flight" (shevell - a guy I took a couple of classes with, in fact), I will send it to you (no scheisse). My wife keeps bugging me to get rid of old text books and it's a good one for basic linear aero stuff. It's in a box in my basement and hasn't been touched since we moved to the new house last October. I bought it new in like '84 or some other Stone Age date.

Shevell is good, Anderson (both Introduction to Flight and Fundamentals of Flight) is good for basic level but sound aero across theory of flight, flight mechanics, S&C, etc.  Shevell has several tomes on the web also.

probably get from Amazon or Albris cheap

[PJ_Godzilla]

'Flexing' is all about stiffness (deflections) for loads applied in both bending and torsion.   Impossible to analyze by 'inspection'. Comparisons would first examine the aero load distribution and the spar/torque box design but even the t/c ratio is deceiving for a deep chord design like the Spit as the actual depth of the main spar, spanwise was probably close to the other designs.  

Thinking about this more, it really wouldn't be all that hard, IF I had section properties of the wing at spanwise stations, or, far easier, closed form descriptions of:

the main wing spar section as f(span)
the lift distribution as f(span) at a given Cl, alt,V
the wing chord as f(span) and the position of the spar within as f(span)
some relevant material data (G and E for whatever aluminum they used on the F-dub)

to get a twist and bending flex distribution for the wing - especially if we made some crude assumptions (model it as main spar, offset lift from torsion axis along a .25chord line using given distribution).

I'm kind of wondering about that wing box structure, though. Messerchmitt pioneered, my recollection, the use of a single-spar wing on the 109. The 190 probably isn't that simple - but such an approx would make for a pretty tractable analysis, so long as I knew the section geometry, taper, material, and chord location of that main spar. OTOH, it's also probably do-able, but harder, to just take a wing structure x-section and figger out the bending and polar inertias numerically. That's hardly elegant - and more time consuming.

I'm also kind of thinking that it'd BE SUPER COOL if HTC modeled, using simple torsion/bending combined loading beams, WING FLEX in our beloved cartoon aircraft. That'd be really cool, especially on something like a B-24. If, as indicated, SPitty's got a significant enough problem with twist that it gets into aileron reversal, that'd be one way to pick it up - a crude but sufficiently descriptive aeroelastic model in-game...

[drgondog]

Drgondog,
How much did areo engineers know back then. I mean, were they able to mathematically predict the general behavior (such as stall handling), or was it more along the lines that they guessed (based on simple mathematical models), then they built it and tested to see if they got the result they wanted?

Good questions. Simple answer - yes to axis-symmetric flight from low speed stall prediction, to drag analysis, to climb and range estimation.  Not so well in asymmetrical aerodynamics, flight mechanics and drag analysis -

Aero is non linear and analysis in pre design is all about enough knowledge to remove insignificant variables from the important ones and they had a very good handle on sub Mcr incompressible flow flight prediction study - but considering the number of failed prototype designs one can safely say they didn't always get it right.

Aeroelasticity and fatigue analysis were non-existant.  Compressibility and transition aero from incompressible to compressible regions were largely a mystery. Analyzing complex airframes and deriving variable pressure distributions to supply loads to airframe structures guys were in infancy and numerical methods, CFD, etc were not possible.. slide rule was the computer.

[drgondog]

Thinking about this more, it really wouldn't be all that hard, IF I had section properties of the wing at spanwise stations, or, far easier, closed form descriptions of:

the main wing spar section as f(span)
the lift distribution as f(span) at a given Cl, alt,V
the wing chord as f(span) and the position of the spar within as f(span)
some relevant material data (G and E for whatever aluminum they used on the F-dub)

to get a twist and bending flex distribution for the wing - especially if we made some crude assumptions (model it as main spar, offset lift from torsion axis along a .25chord line using given distribution).

I'm kind of wondering about that wing box structure, though. Messerchmitt pioneered, my recollection, the use of a single-spar wing on the 109. The 190 probably isn't that simple - but such an approx would make for a pretty tractable analysis, so long as I knew the section geometry, taper, material, and chord location of that main spar. OTOH, it's also probably do-able, but harder, to just take a wing structure x-section and figger out the bending and polar inertias numerically. That's hardly elegant - and more time consuming.

I'm also kind of thinking that it'd BE SUPER COOL if HTC modeled, using simple torsion/bending combined loading beams, WING FLEX in our beloved cartoon aircraft. That'd be really cool, especially on something like a B-24. If, as indicated, SPitty's got a significant enough problem with twist that it gets into aileron reversal, that'd be one way to pick it up - a crude but sufficiently descriptive aeroelastic model in-game...

[Stoney]

Spits' flaps aren't split type?

(Image removed from quote.)

You are correct.  I should have said "split".  Sorry, it was late...

[PJ_Godzilla]

...on something like a B-24. If, as indicated, SPitty's got a significant enough problem with twist that it gets into aileron reversal, that'd be one way to pick it up - a crude but sufficiently descriptive aeroelastic model in-game...

Thinking a little more on this... Even if we had, e.g., a simple multi-node aeroelastic model underlying the graphics, the graphics themselves would probably be upset by this, yes? Some of the elastodynamic models I'd do would flex but only in a "piecy" way w/r the representative graphics - for example, a heavy truck frame rail would look like an elemental representation (because it was). I suspect any rendering of an elastic wing would have to use an awful lot of elements to look convincing, and this probably prices it out of usefulness, computational intensity-wise.

[Stoney]

The 23015 airfoil was known to have sharp stall charcteristics but I seem to recall that the Fw airfoil was a modified 23015-3?

As to torsion issues, hard to say relatively speaking.  The roll reversal on the Spit at high speed was almost certainly due to wing torsion.  With the decided question regarding aileron authority of the Fw 190 in High AoA/G I can't help but wonder regarding outer wing torsion combining with the near simulataneous reaching of CLmax on the inboard 80%.   ???

I think the root chord thickness was 15.3%, so the airfoil most precisely would be listed as a 23015.3.  The only modified 23000 I've ever heard of was the airfoil used on the P-47, which supposedly was a modified 23000 designed by Republic.  I've never done any analysis to determine what the changes introduced, but the profiles look almost identical, and I don't know if something like XFOIL would show a conspicuous difference.

[Stoney]

Thinking a little more on this... Even if we had, e.g., a simple multi-node aeroelastic model underlying the graphics, the graphics themselves would probably be upset by this, yes? Some of the elastodynamic models I'd do would flex but only in a "piecy" way w/r the representative graphics - for example, a heavy truck frame rail would look like an elemental representation (because it was). I suspect any rendering of an elastic wing would have to use an awful lot of elements to look convincing, and this probably prices it out of usefulness, computational intensity-wise.

Well, would the flex even be perceptible visually at a glance?  I think, short of some serious FEA analysis couple with some CFD panelling, you could get anything that would be reliable enough to bother modelling.  The FW-190 roll handling is probably sufficiently dicey as it is in-game to get us a 95% solution--which is probably enough for a good relative representation.  Just my opinion.

[drgondog]

Thinking about this more, it really wouldn't be all that hard, IF I had section properties of the wing at spanwise stations, or, far easier, closed form descriptions of:

Let me ask some questions - reflecting what I would need...

the main wing spar section as f(span)

 I would need -spanwise - the complete drawings - including BOM, so that I could look at the complete inner structure from leading edge to flap hinge line. This detail would give me what I had to know about a.) actual spar and secondary spar (torque box) geometry, b.) beam cap, shear web and rivet type/distribution (to give me a notion of both the assumed bending moments via cap/skin/web geometry as well as the shear transfer from axial to compression loading transfer of the bending loads from upper to lower caps) - from root to the tip.

the lift distribution as f(span) at a given Cl, alt,V

I would particularly like to have the pressure/lift distributions in max G pull out, asymmetric predictions for high speed slow roll and yaw in dives (primarily for eppenage more than wing but also to look at torsion from aileron on the wing torque box).  I would like to have 'best guess' from the aeros on the assumed aero load distribution in high G near max AoA turn.

the wing chord as f(span) and the position of the spar within as f(span)

I want the bleeping dwgs - lol

some relevant material data (G and E for whatever aluminum they used on the F-dub)

with the BOM and any relevant 3rd party data (i.e casting/forging/stamped part spec, etc)

to get a twist and bending flex distribution for the wing - especially if we made some crude assumptions (model it as main spar, offset lift from torsion axis along a .25chord line using given distribution).

Crude won't do you much good but if you had the drawings you could make a very good assumtion on expected loads, and therefore some speculation and knowledge regarding assumed deflections

I'm kind of wondering about that wing box structure, though. Messerchmitt pioneered, my recollection, the use of a single-spar wing on the 109. The 190 probably isn't that simple - but such an approx would make for a pretty tractable analysis, so long as I knew the section geometry, taper, material, and chord location of that main spar. OTOH, it's also probably do-able, but harder, to just take a wing structure x-section and figger out the bending and polar inertias numerically. That's hardly elegant - and more time consuming.

Lol - get the friggen drawings..

I'm also kind of thinking that it'd BE SUPER COOL if HTC modeled, using simple torsion/bending combined loading beams, WING FLEX in our beloved cartoon aircraft. That'd be really cool, especially on something like a B-24. If, as indicated, SPitty's got a significant enough problem with twist that it gets into aileron reversal, that'd be one way to pick it up - a crude but sufficiently descriptive aeroelastic model in-game...

I did this 'stuff' for living before I decided money was better than fun.  I actually pioneered NASTRAN at Bell Helicopter for the Army and some work at NASA along the same path - introducing fatigue, aeroelasticity and high temp into my simple world..

[drgondog]

Thinking a little more on this... Even if we had, e.g., a simple multi-node aeroelastic model underlying the graphics, the graphics themselves would probably be upset by this, yes? Some of the elastodynamic models I'd do would flex but only in a "piecy" way w/r the representative graphics - for example, a heavy truck frame rail would look like an elemental representation (because it was). I suspect any rendering of an elastic wing would have to use an awful lot of elements to look convincing, and this probably prices it out of usefulness, computational intensity-wise.

Well - if you could buckle NASTRAN or Stardyne as a subroutine, model the wing in rods and shear panels, load up to a supercomputer, then solve for Navier Stokes with a very fine mesh to plow through the load aero load model (and back to NASTRAN) in real time through a flight path, you could probably get fairly close.  I know I couldn't and NASTRAN has probably ben superceded by far better methodolgies.

[PJ_Godzilla]

Well, would the flex even be perceptible visually at a glance?  I think, short of some serious FEA analysis couple with some CFD panelling, you could get anything that would be reliable enough to bother modelling.  The FW-190 roll handling is probably sufficiently dicey as it is in-game to get us a 95% solution--which is probably enough for a good relative representation.  Just my opinion.

Well, I'd guess the game doesn't use CFD panelling right now to develop the loads. So, whatever error exists in load development is a separate increment of error, if you like. Separated thus, the question boils down to accuracy of a combined load (somewhat grossly) discretized beam model. Typically, the question at this point becomes how good is good enough? In the case of a chassis-cab Transit I modelled for a rollover reconstruction back in the 90's (this never happened for all recorded purposes) correlation to test improved, naturally, as the element count improved- but the count was by no means especailly CPU costly in the finished model. I'll cast around and see if I can find what kind of convergence to test there is for a fairly complex beam section model under combined loading as you increase the element count. I'm certian 80-20 rule obtains, it's just a question of how much discretization could be supported.

As for visibility of flex - I'd guess not on a 190, except for maybe the TA. I'd bet money you could see it on the 24 and 29, though.

[PJ_Godzilla]

Dragon, you are one crazy motorscooter. I interviewed down at Bell way back in '91. That was before I went into consulting with MDI (the ADAMS people - before they got bought by MSC). I think you guys had an internally developed rigid-body dynamics analysis tool but the interviewers were pretty close-lipped about it. In any case, I think it safe to save that your method would blow the computing and data-transfer capacity of the purveyors and users ALL TO HELL - but it'd be really cool, nonetheless.

I too left engineering for the dark side back in '98. Condolences...

I still contend that, if I could get good test agreement using a flexible-element ADAMS model on a structure as complex as the Transit chassis-cab frame, a wing should be doable. It sounds like you're expecting the secondary spar and the caps to be pretty critical to development of any simplified beam representation. I'm wondering how closely you could rep such a complex structure with a tapered beam. And that'd be the key to making any such thing workable; simplification. As for the lift distribution development, well, how does the game do it now? Clearly, using, for example, your aileron deflection example, the spanwise lift distribution is affected in game because we roll in response. I'm sure the pressure dist on those sections would be affected, thus affecting the torsional moment on the box - and I'm pretty sure the game doesn't care about that.

Any idea on how they're doing their load development?

[drgondog]

I think the root chord thickness was 15.3%, so the airfoil most precisely would be listed as a 23015.3.  The only modified 23000 I've ever heard of was the airfoil used on the P-47, which supposedly was a modified 23000 designed by Republic.  I've never done any analysis to determine what the changes introduced, but the profiles look almost identical, and I don't know if something like XFOIL would show a conspicuous difference.

Good catch - I was actually disconnecting on the trailing modifier to the five digit series after the dash as contrast to a number after the decimal.

[drgondog]

Dragon, you are one crazy motorscooter. I interviewed down at Bell way back in '91. That was before I went into consulting with MDI (the ADAMS people - before they got bought by MSC). I think you guys had an internally developed rigid-body dynamics analysis tool but the interviewers were pretty close-lipped about it. In any case, I think it safe to save that your method would blow the computing and data-transfer capacity of the purveyors and users ALL TO HELL - but it'd be really cool, nonetheless.

LOL. I was at the Beginning of the development back when we started noodling with elastic deformation to predict crash energy absorption.. which of course neither NASTRAN nor Stardyne could handle. Where we were in the 70's wouldn't strain a laptop today, however so I really don't know what the 'hog vector' would be. Iterative solutions on the deflections within the modelled geometry due to applied loads shouldn't be as cumbersome as say, a fine mesh NS procedure - So, Godzilla I am afraid I can't shed too much speculation on where they are these days.

I SUSPECT the V-22 airframe would drive the structures guys to seek higher wisdom for such conditions as wing flex (and associated movement of flap shaft to bearing relationships, as well as the dual torque shafts to and from the gear mixing box to engine.rotor pylon issue. 

I was all over the XV-15 back in the day and part of the Noda Magic pylon design team.

I too left engineering for the dark side back in '98. Condolences...

You are just a youthful lad, grasshopper and have a long and glorious debt ridden life ahead of you as you matriculate to a higher state..

I still contend that, if I could get good test agreement using a flexible-element ADAMS model on a structure as complex as the Transit chassis-cab frame, a wing should be doable. It sounds like you're expecting the secondary spar and the caps to be pretty critical to development of any simplified beam representation. I'm wondering how closely you could rep such a complex structure with a tapered beam. And that'd be the key to making any such thing workable; simplification. As for the lift distribution development, well, how does the game do it now? Clearly, using, for example, your aileron deflection example, the spanwise lift distribution is affected in game because we roll in response. I'm sure the pressure dist on those sections would be affected, thus affecting the torsional moment on the box - and I'm pretty sure the game doesn't care about that.

I don't have access to the model, the assumtions relative to Inertial moments about any axis or assumed load distributions.

I would actually be surprised to discover that the models are that sophisticated - but have an open mind.
The two really tricky parts would be to build in changes to 1.) aerodynamic pressure distributions of a rolling, yawing construct which has a thrust axis and rotational stream tube 'off angle' from free stream velocity vector, and 2.) the tarnsition model from axis symmetric, low vortex/form drag environment (we can look to Anderson, et al for this) to an abrupt asymmetrix, increasing drag, decreasing propeller efficiency (the Thrust thingy), uncertain CLmax due to contributions from elevator/aileron/rudder inputs and operator induced rapid changes to control deflections.

Oh, yes - and predict a real time change in huge Viscous/form drag rise as boundary layer begins to deteriorate and adverse pressure gradients rear their ugly head behind the wing?

These scare me and I'm fearless (also not very smart)

Any idea on how they're doing their load development?

I don't know enough about either the Transit or ADAMS to speculate.

My first reaction to modelling using a tapered beam is that a better analogy is a series of beams, each representing a slightly different geometry and property, coupled with very stiff springs along the spar(s) line where the bending loads are expected to be absorbed.. have to ponder what to do with the outer shear panels, however, as they are the keys to torsion absorption/deflections.

The different beams have to be modelled to provide an analogy for a series of torque boxes with shear and axial load representation (IMO) and are mostly empty space - versus homogeneous constructs

A materially homogeneous beam should not yield the same response to combined vertical and in-plane loads as the box models.

Just my long out of practice humble opinion..good to chat with someone who knows what is involved in airframe type structural analysis.

Back to Transit. My reaction is that modelling it would be one order of magnitude below a WWII fighter and even less aligned with the complexity of a Transition subsonisupersonic capable/wierd flight profile bird like a F-22?

I say that with full understanding that I don't know the complexity of the Transit, nor do I have an understanding of the forces the structure is designed to withstand, but

- an airframe designed to take great loads (Q, and intertial, about six degrees of freedom) with variable combinations both in magnitude and direction while remaining light and easy to produce by the thousands seems to offer more complexity?

[PJ_Godzilla]

I don't know enough about either the Transit or ADAMS to speculate.

My first reaction to modelling using a tapered beam is that a better analogy is a series of beams, each representing a slightly different geometry and property, coupled with very stiff springs along the spar(s) line where the bending loads are expected to be absorbed.. have to ponder what to do with the outer shear panels, however, as they are the keys to torsion absorption/deflections.

The different beams have to be modelled to provide an analogy for a series of torque boxes with shear and axial load representation (IMO) and are mostly empty space - versus homogeneous constructs

A materially homogeneous beam should not yield the same response to combined vertical and in-plane loads as the box models.

Just my long out of practice humble opinion..good to chat with someone who knows what is involved in airframe type structural analysis.

Wow. I can almost see the skin ripple on the wing when you talk torsion. In torsion, the section looks like a thin-walled tube with all the load carried by stressed skin, yes? Bending-wise, you're talking discretization and that makes sense to me - except the boundary conditions on the elements (why springs - connect them rigidly and youd get the same effect anyway, wouldn't you; a local inflection that'll converge to continuous as element count goes to infinity?). The taper I proposed as a means of getting all the properties inot some convenient f(X) form (the dream of a closed-form solution) - but I think I see what you mean about the torsional characteristics... I mean - the beam is not going to rep that thin-walled tube as well as your beam plus torque box. I'm thinking that's because the inertial props can be completely divorced in your rep.

As for Transit structure, it's probably not as bad as a wing but it still involved a fabricated thin-walled structure. Otherwise, your humble input strikes me as really kewl.

Is there any chance you're retired?If so, you might have the time for a development like that...

The rest, re: loads - I'd also be surprised if its all that sophisticated too. Many of the conditions you describe as difficult still frequntly occur in game, though. I look at that crazy tail slide. Isn't that a fine example where you've probably got detached and/or reversed flow and bad pressure dist on one side of the fuselage? Is that all discounted in force development? I seem to be able to get good speed reductions out of yaw oscillation, but suspect that's just based on a linear model of Cd w/r sideslip... In fact, I''d guess it's all just linear range stuff but have no idea.

Finally - life advice goood. The grasshopper, though, is more like an ant. I just sort of like work.