AAE 556 Aeroelasticity The P-k flutter solution method (also known as the “British” method) Purdue Aeroelasticity 1 The eigenvalue problem from the Lecture 33 h2 2 0 h b 1 2 1 ig x 2 0 r 2 1 Lh M h 2 h 2 2 b h 0 0 1 1 x 2 r x h b 2 r 1 h L a Lh b 0 2 0 M x 1 Lh r 2 M h Purdue Aeroelasticity 1 h L 2 a Lh b M 2 Genealogy of the V-g or “k” method Equations of motion for harmonic response (next slide) – Forcing frequency and airspeeds are is known parameters – Reduced frequency k is determined from and V – Equations are correct at all values of and V. Take away the harmonic applied forcing function – Equations are only true at the flutter point – We have an eigenvalue problem – Frequency and airspeed are unknowns, but we still need k to define the numbers to compute the elements of the eigenvalue problem – We invent ed Theodorsen’s method or V-g artificial damping to create an iterative approach to finding the flutter point Purdue Aeroelasticity 3 Go back to the original typical section equations of motion, restricted to steady-state harmonic response 2 AEOM DEOM h BEOM F b EEOM M SC Purdue Aeroelasticity 4 The coefficients for the EOM’s AEOM 2 2 L 1 h 2 h BEOM DEOM 1 1 x L Lh a 2 1 1 x M h Lh a 2 EEOM r2 1 2 Mh 1 M L a 2 Purdue Aeroelasticity 1 Lh 1 a a 2 2 2 5 The eigenvalue problem AEOM DEOM 2 h BEOM 0 b EEOM 0 AEOM DEOM h BEOM 0 b EEOM 0 AEOM DEOM h BEOM 0 b EEOM 0 2 2 Purdue Aeroelasticity 6 Another version of the eigenvalue problem with different coefficents AEOM DEOM h h BEOM A B 0 b b EEOM D E 0 2 2 2 A 1 h h 1 ig Lh 1 B x L Lh a 2 Purdue Aeroelasticity 7 Definitions of terms for alternative set-up of eigenvalue equations for “k-method” h A B 0 b D E 0 1 D x M h Lh a 2 2 1 2 E r 1 1 ig M a h M 2 1 1 L a Lh a 2 2 2 Purdue Aeroelasticity 8 Return to the EOM’s before we assumed harmonic motion Here is what we would like to have M ij j Kij j Aij1 j Aij 2 j Aij3 j 0 Here is the first step in solving the stability problem pt e j j p j p 2 M ij j Kij j Aij1 j 2 3 p Aij j p 2 Aij j 0 Purdue Aeroelasticity 9 The p-k method will use the harmonic aero results to cast the stability problem in the following form 1 p M ij p Bij K ij V 2 Qij ,real 0 2 2 t e pt …but first, some preliminaries Purdue Aeroelasticity 10 Revisit the original, harmonic EOM’s where the aero forces were still on the right hand side of the EOM’s and we hadn’t yet nondimensionalized h 1 it P L b Lh L a Lh e 2 b 3 2 1 h b it P b Lh L a Lh e 2 3 2 h P A11 P Purdue Aeroelasticity h A12 b 11 This lift expression looks strange; where is the dynamic pressure? h 1 it P b Lh L a Lh e 2 b 3 2 V 2 h it 1 3 2 P 2 b Lh L a Lh e 2 V b V 2 b 2 2 P b2 2 2 V k h 1 it Lh L a Lh e 2 b b V Purdue Aeroelasticity 12 Writing aero force in different notation - more term definitions P A11 h b 1 h b 2 A12 V Q11 Q12 2 Q11 Q11 2 A Q12 2 11 V A12 2 b3 2 1 Q12 Lh L a Lh 2 V 2 The Qij’s are complex numbers Purdue Aeroelasticity 13 Aero force in terms of the Qij’s Q11 1 Q12 2b k Lh L Lh a 2 2 Q11 2 bk 2 Lh 2 b k 2 i 2kC k 1 Q12 2 bk L Lh a 2 2 k2 1 2 Q12 2 b ik 1 2C k 2C k k i 2kC k a 2 2 Purdue Aeroelasticity 14 Focus first on the term Q11 Q11 2 b k i 2kC k 2 Q11 2 b k 2 i 2k F iG Q11 2 b k 2 2kG i 2kF Q11 Q11,real iQ11,imaginary Purdue Aeroelasticity 15 The second term k2 1 2 Q12 2 b ik 1 2C k 2C k k i 2kC k a 2 2 k2 1 2 Q12 2 b ik 1 2 F iG 2 F iG k i 2k F iG a 2 2 3 1 2 Q12 2 b 2F k a 2kG a i k 2G 2kF a 2 2 Purdue Aeroelasticity 16 Let’s adopt notation from the controls community to help with our conversion 2 Q11 2 b k 2kG j 2kF Q11 Q11,real jQ11,imaginary If we were to assume motion e pt and p j p then 1 j Q11 Q11,real p jQ11,imaginary j Purdue Aeroelasticity 17 Continue working on the first term in the aero force expression p Q11 Q11,real Q11,imaginary Q11,imaginary Q11 Q11,real p P A11 h A12 b The expression for A11 reads 2 1 1 V A11 V 2Q11,real p Q11,imaginary 2 2 Purdue Aeroelasticity 18 The term with the p in it looks like a damping term so let’s work on it 2 1 1 V A11 V 2Q11,real p Q11,imaginary 2 2 1 Vb V 2 A11 V Q11,real p Q11,imaginary 2 2 b 1 Vb Q11,imaginary 2 A11 V Q11, real p 2 2 k Purdue Aeroelasticity 19 Finally, the exact expressions for each term are as follows 1 Vb Q11,imaginary 2 A11 V Q11,real p 2 2 k Q11 2 b k 2 2kG j 2kF 1 Vb 2 2 A11 V 2 b k 2kG p 2F 2 2 Both terms are real numbers, there is no j here. Purdue Aeroelasticity 20 Aerodynamic moment expression M aero M 1 1 h a Lh b 2 2 4 2 b 2 M L 1 1 a Lh 1 a 2 2 2 3 1 i 8 k M aero A21 h A22 b 1 1 A21 b 4 2 a Lh 2 2 2 1 1 1 4 2 A22 b M L a Lh a 2 2 2 Purdue Aeroelasticity 21 The Qij’s 2 A Q22 2 21 V Q21 Q21, real 2 Real A 2 21 V Q22, real 2 Real A 2 22 V A22 Q21,imaginary 2 Imag A 2 21 V Q22,imaginary 2 Imag A 2 22 V Purdue Aeroelasticity 22 The p-k process Step 1 Choose a value of k and compute all four complex aerodynamic coefficients – These are the complex Aij’s with the Theodorsen Circulation function in them – These will be a set of complex numbers, not algebraic expressions Choose an air density (altitude) and airspeed (V) Purdue Aeroelasticity 23 Perform this computation 2 Qij A 2 ij V Purdue Aeroelasticity 24 Compute the aerodynamic damping matrix, defined as 1 Vb Bij Qij ,imaginary 2 k Qij ,imaginary 1 Bij Vb 2 k Purdue Aeroelasticity 25 Take the results and insert them into an eigenvalue problem that reads as follows 1 p M ij p Bij K ij V 2 Qij ,real 0 2 2 2 Qij ,real real ( Aij ) 2 V pt t e Q 1 Bij Vb 2 ij ,imaginary k Purdue Aeroelasticity 26 Summary 1 p M ij p Bij K ij V 2 Qreal ,ij 0 2 2 Choose k=b/V arbitrarily Choose altitude (, and airspeed (V) Mach number is now known Compute AIC’s from Theodorsen formulas or others Compute aero matrices-B and Q matrices are real Purdue Aeroelasticity 27 Solving for the eigenvalues Convert the “p-k” equation to first-order state vector form 1 2 p M ij p Bij K ij V Qij ,real 0 2 2 displacement vector x j velocity vector v j x j м ь x п п j п п z = э State vector = { j } н п п v j п о п ю Purdue Aeroelasticity 28 State vector elements are related {x j }= {v j } {x j }= {v j } The equation of motion becomes йM ij щ{v j }кл ъ ы йBij щ{v j }+ кл ъ ы йK ij щ{x j }= {0} кл ъ ы Solve for - 1 {v j }= - йклM ij щ ъ ы йй {v j }= кклл йKij щ{x j }+ кл ъ ы {v j } - 1 йM ij щ кл ы ъ йBij щ{v j } к ъ л ы м ь xjп п - 1 - 1 щ щ й щ п п йM ij щ йKij щ йM ij щ йBij щ э ъ к ъън кл ъ к ъ ы л ыылкл ъ ы кл ъ ыы п п v ып j о п ю Purdue Aeroelasticity 29 State vector eigenvalue equation – the “plant” matrix м x j ьп п п {z j }= нпv пэп = п jю п о й [0] I ] щм [ x j ьп к ъп п п н кй- M - 1K щ M - 1Bъпv эп = клкл ъ п jю п ъ ы ыо { йAij щ{z j } кл ъ ы } pt z t = z e ( ) { j} Assume a solution j Result {z j }= p {z j }= йAij щ{z j } кл ы ъ Solve for eigenvalues (p) of the [Aij] matrix (the plant) Plot results as a function of airspeed Purdue Aeroelasticity 30 1st order problem Mass matrix is diagonal if we use modal approach – so too is structural stiffness matrix Compute p roots – Roots are either real (positive or negative) – Complex conjugate pairs 0 Aij 1 M K 1 M B I 1 2 Kij Kij V Qreal ,ij 2 {z j }= p {z j }= Purdue Aeroelasticity йAij щ{z j } кл ы ъ 31 Eigenvalue roots pi pi ,real jpi ,imaginary pi i g i j k b V kV b g is the estimated system damping There are “m” computed values of at the airspeed V You chose a value of k=b/V, was it correct? – “line up” the frequencies to make sure k, and V are consistent Purdue Aeroelasticity 32 p-k computation procedure Input k and V Compute eigenvalues No, change k ki kinput ? yes ki pi i g i j i b V preal ig i i pimaginary i Purdue Aeroelasticity Repeat process for each 33 What should we expect? 0 Aij 1 M K M 1B I Right half-plane Root locus plot Purdue Aeroelasticity 34 Back-up slides for Problem 9.2 Purdue Aeroelasticity 35 A comparison between V-g and p-k h2 0 2 h b 1 2 1 ig x 2 0 r 2 1 Lh M h 1 x 2 x h b r 2 1 h L a Lh b 0 2 0 M x h h2 0 h b b 2 2 2 r 0 r V 2 1 Lh 2 2 V M h 2 2 1 h L a Lh b 0 2 0 M Purdue Aeroelasticity 36 A comparison between V-g and p-k 1 2 x x h h2 0 h b b 2 2 2 r 0 r V 2 1 Lh 2 2 V M h 2 1 2 x 2 1 h L 2 a Lh b 0 0 M x h h2 0 h b b 2 2 2 r 0 r V 2k b Lh 2 2 b m M h 2 2 2 Purdue Aeroelasticity 1 h L 2 a Lh b 0 0 M 37 A comparison between V-g and p-k 1 2 x x h h2 0 h b b 2 2 2 r 0 r V 2k b Lh 2 2 b m M h 2 1 x 2 2 1 h L 2 a Lh b 0 0 M 2 x h h2 0 h b b 2 2 2 r 0 r V 2 k Lh 2 m M h 2 2 1 h L a Lh b 0 2 0 M Purdue Aeroelasticity 38 A comparison between V-g and p-k 1 p2 x h x 1 2 k Lh b p Vb Imag r 2 2 m M h 2 2 2 0 V 2 k Lh h Re 2 2 2 m 0 r M h 1 h L 2 a Lh b M 1 h L a Lh b 0 2 0 M Purdue Aeroelasticity 39 Flutter in action Accident occurred APR-27-95 at STEVENSON, AL Aircraft: WITTMAN O&O, registration: N41SW Injuries: 2 Fatal. REPORTS FROM GROUND WITNESSES, NONE OF WHOM ACTUALLY SAW THE AIRPLANE, VARIED FROM HEARING A HIGH REVVING ENGINE TO AN EXPLOSION. EXAMINATION OF THE WRECKAGE REVEALED THAT THE AIRPLANE EXPERIENCED AN IN-FLIGHT BREAKUP. DAMAGE AND STRUCTURAL DEFORMATION WAS INDICATIVE OF AILERONWING FLUTTER. WING FABRIC DOPE WAS DISTRESSED OR MISSING ON THE AFT INBOARD PORTION OF THE LEFT WING UPPER SURFACE AND ALONG THE ENTIRE LENGTH OF THE TOP OF THE MAIN SPAR. LARGE AREAS OF DOPE WERE ALSO MISSING FROM THE LEFT WING UNDERSURFACE. THE ENTIRE FABRIC COVERING ON THE UPPER AND LOWER SURFACES OF THE RIGHT WING HAD DELAMINATED FROM THE WING PLYWOOD SKIN. THE DOPED FINISH WAS SEVERELY DISTRESSED AND MOTTLED. THE FABRIC COVERING HAD NOT BEEN INSTALLED IN ACCORDANCE WITH THE POLY-FIBER COVERING AND PAINT MANUAL; THE PLYWOOD WAS NOT TREATED WITH THE POLYBRUSH COMPOUND. Probable Cause AILERON-WING FLUTTER INDUCED BY SEPARATION AT THE TRAILING EDGE OF AN UNBONDED PORTION OF WING FABRIC AT AN AILERON WING STATION. THE DEBONDING OF THE WING FABRIC WAS A RESULT OF IMPROPER INSTALLATION. Purdue Aeroelasticity 40 Things you should know Royal Aircraft Establishment The RAE started as HM Balloon Factory. From 1911-18 it was called the Royal Aircraft Factory, but was changed its name to Royal Aircraft Establishment to avoid confusion with the newly established Royal Air Force. Farnborough was known as a center of excellence for aircraft research. Major flutter research was conducted there. Famous R&M’s such as the “flutter bible” came from this facility. The RAE played a major role in both World Wars. So confident was Hitler that he could occupy England with relative ease that he spared the RAE from bombing in the hope of benefiting from its research. Recently the RAE (now known as the Royal Aerospace Establishment) was absorbed into the DRA (Defence Research Agency), itself renamed as DERA (Defence Evaluation and Research Agency). The world famous initials are no more. Purdue Aeroelasticity 41