A Wing Design Method for Aerospace Students and Homebuilders; M A Ferman, PHD, PE

Discussion in 'Aircraft Design / Aerodynamics / New Technology' started by craig saxon, Nov 6, 2012.

Help Support HomeBuiltAirplanes Forum by donating:

  1. Aug 20, 2019 #21

    wsimpso1

    wsimpso1

    wsimpso1

    Super Moderator Staff Member

    Joined:
    Oct 19, 2003
    Messages:
    5,881
    Likes Received:
    3,103
    Location:
    Saline Michigan
    The issue with almost all course books is they assume the list of prerequisite classes have been taken, passed, and the knowledge retained. This is engineering, not basket weaving after all. For classes like this, that would be calculus, linear algebra, differential equations, mechanics of materials, structural design, elastic stability, vibration. I really do not think that EI and GJ should need explaining in a senior/grad level AE/ME course. While I believe that the terms used in the math should be defined in the text, any attempts to make a class on this topic into one where they also teach the math and basic mechanics would become much much bigger and hugely more difficult to extract insight and understanding.

    Flutter gets complicated. Start with the fact that it is a resonance behavior, which is a tough topic for many folks to even get their arms around. You need to calculate the resonant frequencies of the various deflection modes of the structure (which is usually both a distributed mass and distributed stiffness problem), then you have to somehow calculate the forcing functions which are usually due to - I am guessing - vortex shedding. I have no idea how this is even done... Then you wrap into elastic stability. If it turns out you have resonant frequencies coming close to pertinent forcing functions, you have to do one or more of: Reduce mass, increase stiffness, change forcing functions (how?), adjust conceptual design (reduce aspect ratios, add structural members, etc), reduce Vne, others... And if elastic instability rears its head, you again get into significant structural adjustments and perhaps into adjusting the conceptual design again.

    To present flutter and related structural subjects compactly and be fairly accurate is a challenge. To convey insights is even tougher, but must be done to achieve applicability. Makes me want to buy a book or two... Intellectual greed has few limits. As if I do not have enough things slowing progress on this airplane...

    Billski
     
    christos likes this.
  2. Aug 20, 2019 #22

    BJC

    BJC

    BJC

    Well-Known Member

    Joined:
    Oct 7, 2013
    Messages:
    9,534
    Likes Received:
    6,308
    Location:
    97FL, Florida, USA
    Are you suggesting that proper flutter analysis is complicated, Bill?

    There is a reason that, for much of the history of aircraft design, flutter speeds have been determined by flight tests.


    BJC
     
  3. Aug 20, 2019 #23

    harrisonaero

    harrisonaero

    harrisonaero

    Well-Known Member

    Joined:
    Oct 31, 2009
    Messages:
    548
    Likes Received:
    278
    Location:
    Coeur d'Alene, ID
    BTW who's Dave?
     
  4. Aug 20, 2019 #24

    plncraze

    plncraze

    plncraze

    Well-Known Member HBA Supporter

    Joined:
    May 12, 2006
    Messages:
    1,592
    Likes Received:
    332
    "Dave's not here, man" LOL
     
  5. Aug 20, 2019 #25

    flyboy2160

    flyboy2160

    flyboy2160

    Well-Known Member

    Joined:
    May 25, 2014
    Messages:
    320
    Likes Received:
    160
    Location:
    california, USA
    A flutter whiz at a big aerospace company where I used to work.
     
  6. Aug 20, 2019 #26

    flyboy2160

    flyboy2160

    flyboy2160

    Well-Known Member

    Joined:
    May 25, 2014
    Messages:
    320
    Likes Received:
    160
    Location:
    california, USA
    Pushing the flutter buttons in Femap NASTRAN is easy.

    Getting a semi-accurate dynamic FEA model of the plane to feed Femap flutter is orders of magnitude more difficult and time consuming. If you're using a composite structure, it's very complicated. One of the reasons I designed my plane from metal was seeing the enormous effort it took the analysts at work to properly dynamically model a complicated composite structure. (I did a really simple one in school...)

    Understanding flutter is even more orders of magnitude more difficult. I haven't found anything that simply explains the math and the physics principles involved. It is the by far hardest subject I've ever tried to study. I understand a few things, but I'd still say I'm an D-/F student. I'd rather try to learn nuclear physics. :p

    Even several of the the available rule-of-thumb guides I've seen, such as FAA report 45 (which I used for my rudder before I used NASTRAN flutter), require the mode shapes and their frequencies. You can swag some of these closed form; most I haven't been able to. If the ones you can hand-estimate match the FEA modes, you just have to assume the rest of the FEA modes are good.

    Indeed, this level of knowledge and effort is why almost no amateur builders do it. I did it because I foresaw a possible problem with my aft fuselage. As much as Hollman has been maligned, the Lancair 'problem' is still an interesting read : http://www.lancair.net/flutter.html
     
    Last edited: Aug 21, 2019
  7. Aug 20, 2019 #27

    christos

    christos

    christos

    Well-Known Member

    Joined:
    Nov 9, 2018
    Messages:
    50
    Likes Received:
    12
    Location:
    Greece
    You are right. Flutter is a dynamic response problem and there are a lot of "tools" to calculate it.
    In any case, you have to calculate correct natural frequencies, deformation shape and fluid matrices. Like any other dynamic response problem, it is a problem of mass, stiffness and forces (fluid) matrices (generally )
    You will take your answer if you solve it.
    You can calculate it by hand calculations but the result will be not accurate enough to trust it because of simplifications. And of course, through hand calculations, you can not solve tail buffeting (it should need a lot of months).
    But in a small and slow aircraft's case, it is not necessary, because of a strong enough construction to deal with regulations.
    You can calculate all of the above with a good enough pc (you will need a really good ram). Also some codes like open foam (c++ background is necessary) or Ansys CFX/fluent, Matlab etc will be useful.
     
  8. Aug 20, 2019 #28

    flyboy2160

    flyboy2160

    flyboy2160

    Well-Known Member

    Joined:
    May 25, 2014
    Messages:
    320
    Likes Received:
    160
    Location:
    california, USA
    This is for the rest of you. I'm through trying to discuss things with christos; in the past he wouldn't or couldn't explain what he claims is his 'flutter' analysis.

    I'm not an expert, but my studies so far say:

    -You DO NOT NEED CFD for flutter analysis. Classic flutter analysis uses panel methods. NASTRAN flutter has them embedded. All you do is pick the points on the model that correspond to flat aero surfaces. NASTRAN does the dynamic analysis to get the shapes and frequencies.
    -A FEA program will provide the deformed shapes and frequencies, but some of them can be calculated/checked by hand.
    -You DO NOT analyze deformed shapes. Panel or strip methods get the aero on the static shape.You mathematically analyze the change in the mode shape frequencies and
    damping as airspeed changes, starting from the static shape.
    -Designing to strength regulations will NOT PREVENT FLUTTER. You need to worry about frequencies; therefore specific stiffness - stiffness/weight- is important. Just look at all the videos of flutter on structures that met the strength requirement.
    -To repeat: I suspect Ferman's book uses a closed form solution for wing flutter called Galerkin's method.
    -Anything can flutter, even a low, slow LSA plane or an RC model. There is a video here somewhere in the crash thread that shows an Italian pod / skinny boom LSA (piuma) fluttering right near it's stall speed. I can't find it now. I'll post it later if I can.

    Here you go:

    The builder-crasher did install a heavier, more powerful engine than spec. There's a debate about whether there were build errors.
     
    Last edited: Aug 21, 2019
    PiperCruisin likes this.
  9. Aug 20, 2019 #29

    flyboy2160

    flyboy2160

    flyboy2160

    Well-Known Member

    Joined:
    May 25, 2014
    Messages:
    320
    Likes Received:
    160
    Location:
    california, USA
    A classic flutter analysis looks like these linked photos. I tweaked the aft fuselage design so that the aft fuselage plunge and twist frequencies did not cross each other anywhere near Vne.

    https://www.flickr.com/photos/140193367@N07/48587888331/in/dateposted-public/

    https://www.flickr.com/photos/140193367@N07/48587891156/in/dateposted-public/

    For a typical ogive, monocoque skinned aft boom, the torsional natural frequency, as you would expect, is higher than the aft boom bending frequency. As airspeed increases, the bending frequency increases, while the torsional frequency decreases. At some airspeed, these coalesce and you have flutter. I completely understand what Hollman did with the Lancair: he added 0s on the on the top (and bottom) to increase the plunge frequency. I couldn't get the 'after' results from his wife.

    I thought my boom should be ok before the NASTRAN flutter run because I tweaked the design to get the plunge frequency above the torsional frequency - thinking they shouldn't get closer together. The wing bending and torsion frequencies behave qualitatively like previous data and analyses - the bending frequency increases while the torsion frequency decreases. So the Femap run makes some sense.
     
    Last edited: Aug 21, 2019
    Doggzilla likes this.
  10. Aug 25, 2019 #30

    christos

    christos

    christos

    Well-Known Member

    Joined:
    Nov 9, 2018
    Messages:
    50
    Likes Received:
    12
    Location:
    Greece
    Flyboy2160,
    I don' know your engineering background but you don't know exactly what flutter is.
    I am not your teacher, i told it to you and i will tell it to you for last time, you still don't understand.
    Flutter is an aeroelastic problem. It is a dynamic instability during flight.
    First of all, you have to solve the "fluid" part of your problem.
    There are a lot of different methods to calculate it (one of these is the British one, NASTRAN use it).
    NASTRAN run in the background a CFD solver and for this reason, you don't know it. NASTRAN use a low order CFD solver.
    It is a good choice to solve easy flutter problems, but:
    You have to know exactly your problem. It is easy to make big mistakes with easy operation software.
    We (engineers) tell that if you put rubbish in a pc it will give back to you, rubbish. You have to know your software limitations, and choose a suitable method and set correct boundary conditions.

    Because flutter is a dynamic instability caused by airflow around aircraft, i chose Navier Stokes CFD solver instead of a low order CFD solver because it is simply better and maybe the best one. Also, it is the only way if you want to be sure about your tail. Because of the main wing's turbulence. Of course, you will spend a lot of computing hours.

    Then you have to solve the dynamic response part of your problem. You can calculate it by hand but you will have a lot of work (because you will be calculated eigenvalues until you reach 90% mass participation ratio or 10 first eigenvalues).
    After, you will be ready to start dynamic response analysis.

    Maybe there is a lot of easy to use softwares for static analysis, dynamic analysis, response spectrum analysis, dynamic response analysis, etc, out there, but keep in your mind that these are tools. It is not good to gamble on your software. It is something like that a non-pilot man flies, thanks to autopilot.
    A real pilot is a much better option.


    Christos
    Master Integrated civil engineer
     
    Last edited: Aug 25, 2019
  11. Aug 28, 2019 #31

    flyboy2160

    flyboy2160

    flyboy2160

    Well-Known Member

    Joined:
    May 25, 2014
    Messages:
    320
    Likes Received:
    160
    Location:
    california, USA
    Correction: I should have inserted '2D' in front of 'panel' above, as in vortex lattice.
     
  12. Oct 1, 2019 #32

    flyboy2160

    flyboy2160

    flyboy2160

    Well-Known Member

    Joined:
    May 25, 2014
    Messages:
    320
    Likes Received:
    160
    Location:
    california, USA
    rrrrrrrrr I need to make another correction/amplification: After the 2D panels are hooked into the real structure, the 2D panels are deformed (mathematically by the computer) by the structural mode shape and air air loads are determined from this deformed shape. These air loads are then applied to the undeformed shape to determine the resulting frequency and damping changes with airspeed for that particular mode.
     
  13. Oct 4, 2019 #33

    Andy_RR

    Andy_RR

    Andy_RR

    Well-Known Member

    Joined:
    Sep 29, 2009
    Messages:
    180
    Likes Received:
    105
    Location:
    Melbourne, Australia
  14. Oct 4, 2019 #34

    flyboy2160

    flyboy2160

    flyboy2160

    Well-Known Member

    Joined:
    May 25, 2014
    Messages:
    320
    Likes Received:
    160
    Location:
    california, USA
  15. Oct 4, 2019 #35

    pictsidhe

    pictsidhe

    pictsidhe

    Well-Known Member

    Joined:
    Jul 15, 2014
    Messages:
    6,854
    Likes Received:
    1,843
    Location:
    North Carolina
    Did you sell Ferman's book?
     
  16. Oct 4, 2019 #36

    flyboy2160

    flyboy2160

    flyboy2160

    Well-Known Member

    Joined:
    May 25, 2014
    Messages:
    320
    Likes Received:
    160
    Location:
    california, USA
    My comment about his seemingly excessive skin thickness was buried in the middle of a long comment I made. It is 'thick' because he is smearing the spar cap inertia into the skins. He is just trying to get an initial weight, strength, and flutter estimate. For that, it doesn't matter where the wing box I is actually located. Later in the book, he refines this.
     
  17. Oct 4, 2019 #37

    flyboy2160

    flyboy2160

    flyboy2160

    Well-Known Member

    Joined:
    May 25, 2014
    Messages:
    320
    Likes Received:
    160
    Location:
    california, USA
    Dr. Ferman also authored a couple of interesting reports about Conceptual Flutter Techniques in which he tried to simplify the math a lot in order to get a 'practical engineer's' answer about possible flutter problems early in the design cycle. He tried to move the math away from a theoretical purist's scheme to a get an answer that was ~95% correct - good enough for the conceptual stage of design.

    These 2 reports are F322 under NavAir contract N0w66-0298-c and F971 under NavAir Contract N00019-67-C-0213. They include tunnel testing validation of his method.

    Alas, these are not the method in his Homebuilders book.
     
  18. Oct 20, 2019 #38

    christos

    christos

    christos

    Well-Known Member

    Joined:
    Nov 9, 2018
    Messages:
    50
    Likes Received:
    12
    Location:
    Greece
    I haven't his book, i only read some pages. It looks interesting, but it also looks too simplified.
    I hope that you know the difference between "95% of cases" and "95% accuracy".
     
  19. Oct 21, 2019 #39

    brad.walker

    brad.walker

    brad.walker

    New Member

    Joined:
    May 26, 2017
    Messages:
    3
    Likes Received:
    0
    Location:
    Charlotte, NC

Share This Page

arrow_white