Professional Documents
Culture Documents
DYNAMICS
OF HELIC~~ER
ROTOR
BLADES
P. J. MAGARI, L. A. SHULTZ and V. R. MURTHY Department of Mechanical & Aerospace Engineering, Syracuse University, Syracuse, NY 13244, U.S.A.
(Received 14 September 1987)
Abstract-A rotating blade finite element with coupled bending, torsion and axial stretching degrees-offreedom is developed. APL is used for symbolic manipulations required for the development of the element. The element is implemented in MSC/NASTRAN to generate the numerical results. The results are compared with the experimental and other published numerical results and are found to be accurate. There are several immediate potential applications. The implementation of the element in the existing finite element software greatly enhances their utility for helicopter applications.
NOTATION
area of cross section, in2 Youngs modulus, lb/in2 mass centroid offset from the elastic axis, positive in front of the elastic axis, in. shear modulus, lb/in2 coordinates of load paths on clevis, in. blade cross section moment of inertia about y-axis, in4 blade across section moment of inertia about z-axis, in4 St Venants torsional stiffness constant, in4 element of stiffness matrix, ith row, jth column mass radius of gyration (kf,=k& +k$) element length, in. element of mass matrix, ith row, jth column mass/length, lb-sec2/in2 radius of rotor, in. tension, lb strain energy axial deflection, in. elastic displacements in the y (chordwise) and z (flapwise) directions, respectively, in. kinetic energy global x coordinate to root of element measured from center of rotation, in. undeformed coordinate system rotating with rotor pretwist angle, rad rotation about y-axis, rad rotation about z-axis, rad elastic twist about x-axis, rad rotor rotational speed, rad/sec frequency of vibration, radfsec
Subscripts and superscripts
pose [l]. A survey of these programs clearly shows that they are incapable of analyzing rotating helicopter blades other than simulating a simple centrifugal force field. The finite element method is currently being applied to isolated blade dynamic problems [2-131 and shows a great potential for isolated blade as well as coupled rotor-body analyses. In a recent reference 1141 the finite element method was applied to determine the free vibration characteristics of pretwisted blades with coupled bending degrees-offreedom. The main limitations of this paper for application to helicopter rotor blades are (i) no blade rotation, and (ii) no torsional or axial degrees-offreedom. The main objective of this paper is to develop a rotating blade finite element which can be implemented easily in the existing finite element software such as MSC/NASTRAN. The element is to have the following features: i. No assumptions regarding the coincidence of neutral, elastic and mass axes are to be made. ii. Significant coupling-type terms associated with the centrifugal forces are to be included in the model. The equations derived by Houbolt and Brooks [I 51 provide an example of the coupling that can exist between the various degrees-of-freedom due to blade rotation. The derivation of the stiffness and the coupled mass matrices is analytical and involves very tedious algebraic operations, integrations and differentiations, and computer programs are developed in APL for these symbolic manipulations. AFL is an intermediate level computer language, which is very convenient for scientific calculations involving array processing. APL was introduced 20 years ago and its growth and development is still going on. Because APL treats arrays uniformly, there is new interest in the use of APL for describing problems for parallel, vector, or array computing. The computer programs developed in APL to carry out the symbolic operations are given in Appendix I.
derivative with respect to x derivative with respect to time ith load path
lNTRODUCIION The finite element method is a universally applicable approach for obtaining numerical solutions to a wide variety of problems in the structural, fluid flow and heat transfer areas. The method is extensively used in the aerospace industry for structural analyses and several computer programs such as MSC/ NASTRAN, ANSYS, MARC etc. exist for this pur763
764
P. J. MAGARIet al
In order to validate the present rotating blade finite element, it is first utilized to perform a coupled vibration analysis on a conventional propeller. Solutions in the form of natural frequencies are generated via the present formulation and compared to the experimental data [16] for the vibrating propeller. Excellent accuracy is obtained with the use of only a few elements, thereby showing high computational efficiency. Then the new element is used to perform a coupled vibration analysis on a specific advanced configuration bearingless multiple-load-path rotor blade[17], similar to those in NASAs ITR/FRR project [18]. The bearingless rotor technology from the ITR/FRR program may well be incorporated into the Armys next generation helicopter program. A fast converging method is developed to take into account the nonlinearities associated with the multiple load paths and the rotation. The rotating blade finite element developed in this paper can be implemented in MSC/NASTRAN in the following ways. 1. Through GENEL, Bulk Data Cards. 2. Through DMI Bulk Data Cards. 3. Through the INPUTT4 DMAP Module. 4. By incorporation as a dummy element. 5. By incorporation into the library as a distinctly recognizable element. The first three of these approaches are used to generate the results presented in this paper.
HAMILTONS PRINCIPLE
Elahic
axis
plane which is normal to the undeformed elastic axis of the blade. The cross section of the blade is assumed to be symmetric about the chord and the n-axis is defined as the major principal axis along this chord. The C-axis is perpendicular to the chord and passes through the shear center as shown in Fig. 2. The angle between the global y-axis and the crosssectional major principal axis n is defined to be the pretwist of the blade and is denoted by /l (see Fig. 2). The locations of a point P before and after deformation in the cross-sectional plane are shown in Figs 3 and 4, respectively. The position vector of this point in deformed coordinates (x, , y, and z]) can be
The equations of motion can be derived using the extended Hamiltons principle [ 191which states: of all the admissible displacements between specified initial and final conditions at times t, and t2 which satisfy the prescribed boundary conditions of the system, the displacements which also satisfy equilibrium and stress boundary conditions are selected by the extremum principle given by the following eqn. (1)
Fig. 2. Blade cross-sectional axes.
where V = kinetic energy U = potential energy 6 W = virtual work done by the body and surface forces 6 E variation.
COORDINATE SYSTEMS
The coordinate system (x, y, z) rotates with the blade such that the x-axis falls along the undeformed elastic axis of the blade as shown in Fig. 1. (n, I) are cross-sectional axes located in the cross-sectional
Elastic
axis
765
I
written as
_____ )1 _____& i
Y,
------4
Pi
The velocity vector of a cross-sectional point is given by ~fferentiating the position vector r given by eqn (2) as shown below.
6
I
dr
Yl
&=f+wXr,
where w is the angular velocity vector of the triad (i, j, k) and is given by w=kQ. (9)
r=ix,+jy,+kz,,
(2)
where i, j, k are unit vectors along the undefo~ed rotating axes x, y and z, respectively. The relations between the undeformed global (y, z)-axes and crosssectional principal (Q [)-axes are given by (see Fig. 3) y=qcosfl--[sin/? z =q sin/3 -4 cosfi. (3)
oR{m~~ti-*~)+(i+Ctx+*u)Z+ti2] -Ciu)(ti+Ci)
+2mecos/?[-(ti
-(tj+Rx+SZu)~u+~~-Rx#(~+nw)] +2me
sin/31 -(ti
-C&)(6-Q4)
The relations between the undeformed and deformed global coordinates are given by (see Fig. 4) x,=x+u-yu-zw y,=y+u--z# z,=z+w+y&. (4)
-(d+nx+nu)(b+nw)+52*,~,] + m(k& cos* /!I+ k& sin2 fl) x [(d+sQ*+ (Qu)* + fj-nz@ f2Qrgw
+SW($
STRAIN ENERGY
+nw)-n(~o~+~lj~)))dx,
(10)
The strain energy of the rotor blade undergoing coupled flapwise bending, chordwise bending, torsion, and axial stretching deformation is given by U=i s0
+EZ~(ucos~ +
where
pydrtd5
=mecos/3
pzdqd[=mesinfl
ss A
s
R x
Q*mx dx.
(6)
py2 d? di = m(k& COG8 f k& sin2 8) IJ A pz* dtl dC = m(kj,, sin* /I + k$ cos2 8)
ss A The derivation of the above equation is based on the principles of en~nee~ng beam theory and follows the treatment of Houbolt and Brooks [15].
166
P. J. MAGARI et al. The displacements in terms of nodal displacements and shape functions are given by u(x) = siui + s3u2 (12)
For the case of conventional helicopter rotor blade applications, the above kinetic energy expression can be simplified to the following eqn: If=; R{m[li2+dZ+~2+R2U2+R2y2+*2X2 s0 - 2nvli + 2R3xu + 2Rxd + 2nuti1 + 2me cos b [R2u + biuj - Q2x+v - R2xv] +2me sin/l
+m(k&cos2j?
4(x)=s34,+s442
(13)
~(x)=s,u,+s,e,,+s,v,+s,e,,
[-tifj + +~2x~v'-R2v~-~2~~'] k&
(14)
+ m(kf,,, sin/l + k& cos2 jI)[d2 + Ci2d2] - 2m cos fl sin fl(k& - k$)Ci2$)} dx.
(15)
MASS AND
DAMPING
MATRICES
SHAPE
FUNCTIONS
In order to evaluate the integrals of eqns (5) and (11) the displacement fields within the element are assumed as shown to be Axial stretching u = a, f a,x Torsion 4 = b, + 6,x
By taking the variation of eqns (5) and (1 l), the following eqns can be obtained: 6U= R{[EIY(t(sinj?-wcos/?)] I0
x
+[El,(v"cos~+w"sin~)]
The following shape functions are derived from the above displacement functions to express the displacements in terms of the nodal displacements. Axial stretching s,= l-x/l s2 = x/l. Torsion s,=l-x/l sq = x/l. Chordwise bending s5 = 1 - 3x2/12 + 2x3113
s 6-- x s, =
+ TwGw} dx
6V=
+m(ti -Rv)&i
I + [m (fi*v - f&i) + me cos /I R2 - me sin bn24]& + [m(ti + Rx + Qu) -me + (mti + me cos fi&)Sti +
+me
sin /I$]%
[me cos /?R'xw
-m
- m cos /I
(k& - ki,)R2]64
sin pi, +
m ki t$ )SC$
2x211 + x3/12
+ (- me cos fif12x + me
+(-me -me cosj?R2x~
sin j3f12x~)6v'
3x2/13 - 2x3/13
sg = -x2/1 +x3/P. Flapwise bending s9 = 1 - 3x3/12 + 2x3113 SIO = - x + 2x211 -x3/s s,, = 3x3/12 - 2x3/13 s,2 = x2/1 - x3/12.
(17)
By substituting eqns (16) and (17) into eqn (1) and integrating by parts with respect to time, one can show that Hamiltons principle becomes SH= -6U+6V=O, where 6U is given by eqn (16) and 6 V
is
161
6V= j
a(I-
+[-m(ii-R2u)+me
sin/%$
The expressions for the resulting stiffness and mass matrices are too lengthy to present in this paper and therefore only the stiffness terms that arise due to blade rotation are presented in Appendix II. These can be added to the nonrotating blade stiffness matrix computed from the existing finite element codes, thereby allowing fully coupled modeling of the rotating systems.
TENSION COEFFICIENTS
+[-Q2mevsin/?+me(iisin/J-ticos/?) - R2m (ki, - k$)(cos - mki$ - Rzmex(-v j3 sin B + 4 cos 28) sin /l + w cos /I)]&$
- timex (cos /I - 4 sin /I)& -R2mex(sin/?++cosfl)6w}dx. If a finite element discretization is employed Hamiltons principle, then eqn (18) becomes
6H= i 6Hi= i (-SUi+6Vi)=0, i=l i=l
(19) for
The stiffness matrix of the element depends on the tension, and in the case of a conventional rotor blade the tension can be calculated in advance from eqn (6). The resulting equation in the element coordinate system is given by T(x) = T2 + where Xi +1, R2mx dx, X, +X 1 (27)
(20)
R
T, =
n2mx dx +4 s Xi
(28)
where
6Hi= -SUi+SVi. By substituting (21)
ii = length of the ith element i, i + 1 = nodes of the ith element xi = distance of the ith node from the axis of rotation. The fully articulated rotor (Fig. 5) is in use on most production helicopters in the world today. This rotor system contains three hinges: the flapping hinge, the lead-lag hinge and the pitch hinge. The flapping hinge was initially introduced to solve the body rolling problem due to the differential lift generation of the advancing and retreating blades in forward flight. The lead-lag hinge was introduced to relieve high stresses in the blades resulting from the periodically varying in-plane forces due to blade flapping motion. The pitch hinge was provided to facilitate the blade collective and cyclic pitch controls. The presence of these hinges complicates the rotor system and recently development of bearingless rotors was initiated to eliminate this complexity [18]. The
eqns (12)-(15) into eqns (16), (19) and (21), and by recognizing that the limits of integration of eqns (16) and (19) are xi and xi+, the following equation can be obtained for the variation of the Hamiltonian: 6Hi = {hi}T([W{4iii) + [Cl{4,1
+ [W(cliI+ {fI)v
where [M] = mass matrix [C] = damping matrix [K] = stiffness matrix {f } = motion independent The components by
Mjk =
(22)
force vector.
a2(mi)ja~ja(sqk)
a2@Hi)/a+%%k)
cjk
Hub rotation
768
P. J. MAGARI et 01.
I I I k 1 Load
R Dath
I I I I
Load path 3
Fig. 6.
bearingless rotorcraft offers reduced weight, less complexity and superior flying qualities, and it is most likely that the next generation rotorcraft will be equiped with those rotors. The recent advances in materials and understanding of rotor dynamics are responsible for the development of the bearingless rotors. All practical designs of bearingless rotors include multiple load paths, and the one that was flight tested by Boeing Vertol had three load paths [ 191.The model of such a triple-load-path rotor blade is shown in Fig. 6. While the tension on the blade is calculated using eqn (27) as before, the tension in the n load paths are calculated as follows. Assuming that the n load paths are coincident with the blade at the clevis, it can be shown [17] that the tension in the ith load path at the clevis is given by
are needed to achieve the convergence. The final converged values are the desired tensions in the load paths and the blade.
NUMERICAL RESULTS AND DISCUSSION
where fi a,=a,a,...aJai
j=l r' dx
a=
J .Ei
In order to validate the present formulation, a coupled vibration analysis is performed on a rotating twisted propeller blade having a nonuniform, unsymmetrical cross section and cantilever boundary conditions at the root. The computed results are compared with experimental data obtained in the form of natural frequencies of the system. The propeller blade analyzed is the WADC S-5 scale model [16]. The propeller radius given is 24 in. (measured from the center of rotation), with the blade root effectively cantilevered 6 in. from the center of rotation. Except for rotational speed, all the data necessary to compute the natural frequencies is given in Table 1 for a propeller pitch setting of 0 at 0.75R. For different pitch settings the only change in the input data is a change in the values of twist corresponding to the desired pitch setting. The first, second and third mode natural frequencies are computed for cases corresponding to the rotational speeds and pitch settings of the experimental data[16,20]. Table 2 shows the very close correlation between the natural frequencies generated Table 1. Data for the WADC S-5 scale model propeller
Grid point location, x (in.) m (lb-&/in*)
I = length of the load path. Now, the tension in the ith load path corresponding to the coincident load path is given by
I T(x) = Q2mx
(lb-n?)
(lb-in )
Et2
(deg)
dx + T:.
sX
(30)
Once the approximate tensions in the load paths and the blade are calculated as indicated above, the static stiffness matrices of the system can be calculated and new tensions in the load paths at the clevis (T,) can be calculated. These new values are used to update the Ti in eqn (30). At most, two or three iterations
6 1.026 x 1O-3 8 0.696 10 0.660 12 0.608 14 0.564 16 0.535 18 0.520 20 0.506 22 0.498 24 (tip) 0.498 x,=6in.; I=2in.;
30.5 0.200 x 106 63 x lo6 49 25.2 0.110 46 20.1 0.083 44 14.8 0.058 43 9.6 0.042 43 4.7 0.03 I 44 0 0.027 47 -4.2 0.026 51 -7.5 0.025 56 - 10.0 0.024
R=24in.
Dynamics of helicopter rotor blades Table 2. Comparison of natural frequencies (Hz) Mode 1 2 3 1 2 3 1 2 3 1 2 3 1 R @pm) 1567 1589 2609 2614 3583 FEM 40.79 109.32 279.95 41.31 109.97 280.62 60.16 139.65 319.38 60.44 139.47 319.59 78.27 Natural frequencies TMM EXFT 40.08 107.53 58.73 137.02 76.52 40.96 109.22 279.79 41.35 109.77 280.47 60.07 139.52 319.40 60.16 139.68 319.62 78.34 Mode 1 2 3 4 5 6 7 8 9 10 Table 4. Comparison of natural R=Orpm frequencies
769 (rad/sec),
Single load path FEM TMM 7.3184 42.4490 45.9095 128.5689 133.6289 251.9575 266.1199 401.4146 416.5278 622.2776 7.3243 42.4708 45.9001 128.5038 133.6382 251.8232 266.1357 400.5921 416.4330 621.0142
Triple load path TMM FEM 11.4351 60.8580 70.4045 160.3119 181.7632 258.2888 490.8899 430.9312 475.4824 666.7434 11.4298 65.8368 70.3664 160.2549 181.7120 258.7264 400.2951 430.5055 475.0182 665.9004 (rad/sec),
Table 5. Comparison by the present formulation (FEM), experimentally obtained values (EXIT), and natural frequencies of the system computed by the transfer matrix (TMM) [20]. Single- and triple-load-path systems are analyzed. Cantilever boundary conditions are assumed at the inboard ends of all the load path member(s) for the single- as well as the triple-load-path system. The load path members are connected to the blade through a rigid clevis as shown in Fig. 6. Noting that the clevis is completely rigid, the multipoint constraints across the clevis are as given below:
Mode 1 2 3 4 5 6 7 8 9 10
Single load path FEM TMM 36.7365 48.0705 105.1274 138.0839 202.6147 280.5532 336.7368 402.8845 508.3706 671.6609 36.7738 48.1092 104.9309 138.2931 202.4001 280.5927 336.3352 402.5505 507.5868 669.1642
Triple load path FEM TMM 45.0691 72.3661 125.2526 164.1234 232.3592 318.5969 426.8616 477.1384 499.9177 742.3208 45.0998 72.3376 124.9840 164.3721 232.1099 318.3011 426.1332 476.4666 499.3885 740.2850
(31)
The subscript i denotes load path quantities and can vary in value (l-n) dictated by the system under analysis. The coordinates of the location of the load paths on the clevis relative to the blade root location on the clevis (point 0) are given by hi and hi.
The physical constants of the single- and tripleload-path systems analyzed are given in Table 3. The natural frequencies obtained from the present formulation are presented in Table 4 for single- and triple-load-path blades for the nonrotating case (a = 0 rpm). The natural frequencies for the rotating case (Q = 360 rpm) are presented in Table 5. With the exception of the nonrotating single-load-path frequencies (which are simply a degenerate case of the single-load-path rotating formulation), all the results presented in Tables 4 and 5 are compared with the transfer matrix solutions for the same rotor system [17]. This provides further validation of the present element and formulation. Both Tables 4 and 5 show the excellent agreement between the results
Table 3. Data used for the numerical calculations Radius of the rotor Distance of the clevis from the root Rotational speed (Q) Flapwise bending stiffness @I,,) Chordwise bending stiffness (EI.) Torsional stiffness (GJ) Axial stiffness (EA) Mass per unit length Built-in twist Collective pitch Distance between mass and elastic axes #,, mkk For triple-load-path
h,, = 1.0 in.; h,, = 2.0 in.;
= = = = = = = = = = = = =
260 in. 52 in. 360 rpm 0.2977 x 10s lb-in* 10 x lo8 lb-in* 0.2 x IO8lb-in2 10 lb 0.0015 lb-sec2/in2 0.0 deg 15.026 deg - 0.6 in. 0.8945 x lo- lb-se2 0.04 lb-sec2
h, = - 1.0 in.;
rotor
770
P. J.
MAGARI et al.
Table 6. Comparative convergence of the element. Natural frequencies (rad/sec) for triple-load-path rotor, R=36Orpm Mode
1
Case 1 Lumned Coupled 44.7709 71.7704 120.7130 163.7905 224.2091 306.2759 412.1135 465.9194 469.8479 677.3142 45.0829 72.4075 125.2827 164.3912 232.4769 318.7578 427.0327 484.0181 501.1094 753.9558
Case 2 Lumped Coupled 44.9860 72.1452 124.2684 164.0283 230.5968 315.7322 423.6169 474.7075 492.0505 732.0105 45.0521 72.2845 125.2544 164.1602 232.3641 318.6079 426.8689 477.8970 499.9707 744.2107
Case 3 Lumped Coupled 45.0320 72.2875 124.6992 164.0494 231.3711 316.9766 425.0334 475.3533 495.4612 736.2747 45.0691 72.3661 125.2526 164.1234 232.3592 318.5969 426.8616 477.1384 499.9177 742.3208
TMM 45.0998 72.3376 124.9840 164.0571 232.3490 318.3011 426.1332 476.4666 499.3885 740.2850
2 3 4 5 6 7 ! 10
Case 1. Three elements for each load path, 7 for the blade. Case 2. Five elements for each load path, 15 for the blade. Case 3. Seven elements for each load path, 20 for the blade.
from the present finite element formulation and the transfer matrix method. Finally, the rapid convergence of the element utilizing both lumped and coupled mass matrices may be seen from Table 6. All the natural frequencies in Table 6 are for a tripleload-path rotating case (Q = 360 rpm) corresponding to the triple-load-path case presented in Table 3. Again, transfer matrix solutions are provided for comparison. All the above calculations are done in MSC/ NASTRAN by feeding the element through the following means. 1. GENEL Bulk Data Card. 2. DMI Bulk Data Card. 3, INPUTT4 Module. The present formulation greatly enhances the capability of MSJNASTRAN for helicopter dynamics analyses and implementation can be further improved by employing the following means. 1. Incorporation as a dummy element.
highly twisted propeller blade operating at various high rotational speeds. The results generated agree very well with the experimentally obtained results. The element is applied to analyze single- as well as multiple-load-path blades. For multiple-load-path blades, a fast converging iteration scheme for computing the distribution of blade tension in each load path is developed. The results from the present finite element formulation closely correlate with calculated results from other formulations. There are several immediate potential applications. The element can be implemented in MSC/ NASTRAN for coupled rotor-body analyses within NASTRAN. Also, the new element could be used for the design and analysis of multiple-load-path bearingless rotors, without making any equivalent single-load-path approximation.
a nonuniform,
REFERENCES
finite element analysis. Advances in computational methods in structural mechanics and design. Papers
presented at the 2nd U.S.-Japan Seminar on Matrix Methods of Structural Analysis and Design, August 1972. 2 V. T. Nagaraj and P. Shantakumar, Rotor blade vibrations by the Galerkin finite element method.
J. Sound Vibr. 43, 575-577 (1975).
CONCLUSIONS
In the present formulation, a new universal finite element is derived for nonuniform, pretwisted rotating blades with flapwise bending, chordwise bending, torsion and axial stretching degrees-of-freedom. Also, for dynamic applications, a coupled mass matrix consistent with the stiffness matrix is derived. New APL symbolic routines developed here allow the development of stiffness and coupled mass matrices in an analytical form. Since no numerical integrations are required to evaluate the matrices for any given rotating system, the element has a high computational efficiency and yields excellent accuracy. The element is thoroughly validated by analyzing
Dynamics of helicopter rotor blades 9. A. V. K. Murty and A. Raman, Non-linear dynamic analysis of rotors by finite-element method. J. Sound
Vibr. 69, 559-568 (1980). 10. F. K. Straub and P. P. Friedman, A Galerkin type finite
771
element for rotary wing aeroelasticity in hover and forward fliuht. Verricu 5. 75-98 (1981). 11. D. H. Hodges and M. .J. Rutkbwski, Free-vibration analysis of rotating beams by a variable-order finiteelement method. AIAA Jnl. 19, 1459-1466 (1981). 12. N. T. Sivaneri and I. Chopra, Dynamic stability of a rotor blade using finite element analysis. AZAA Jnl20,
716-723 (1982). 13. N. T. Sivaneri and I. Chopra, Finite element analysis for bearingless rotor blade aeroelasticity. J. Am. Helicopter Sot. 29, 42-51 (1984). 14. M. Sabuncu, Coupled vibration analysis of blades with angular pretwist of cubic distribution. AZAA Jnl 23, 1424-1430 (1985).
Design considerations for bearingless main rotor hubs. Presented at the 39th Annual Forum Proceedings of the American Helicopter Society (1983). 19. P. G. C. Dixon, Design, development and flight demonstration of the loads and stability characteristics of bearingless main rotor. USA AVRADCOM-TR-80-D3 (1980). 20. V. R. Murthy, Dynamic characteristics of rotor blades.
J. Sound Vibr. 49, 483-500 (1976).
POL YMULT The left and right hand arguments of this function are executable APL statements of multivariate polynomials of the following form:
15. J. C. Houbolt and G. W. Brooks, Differential equations of motion for combined flapwise bending, Chordwise bending, and torsion of twisted nonuniform rotor blades. NACA Report 1346 (1958). 16. W. F. Hunter, The integrating matrix method for determining the natiural vibration characteristics of propeller blades. NASA TN D-6064 (1970). 17. V. R. Murthy and A. M. Joshi, Free vibration characteristics of multiple load path blades by the transfer matrix method. J. Am. Helicopter Sot. 31,43-50 (1986). 18. W. G. Bousman, R. A. Ormiston and P. H. Mirick,
+ ((2 x co1 x co2 + 14) x (X*7)) WY constant variable +((-5xcO6xco7+l)x(x*1))+~~~ ww constant variable The returned value is also an executable APL statement which is the multivariate polynomial multiplication of the right and left arguments.
Cl] C23
n~t~tr*~~**t~~tf*Lt***
Skkkkkkkkkkkkkkkkkkkkkkkkkkk Skkkkkkkkkkkkkkkkkkkkkk
PfiTRICK
kkkkkkkkkkkkkkkkkkkkkkkkkkkkkk
kkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk
C73
C83
n
R
OF
POLYMULT @PL
CIND
RETURNS
ILGEBR&IC
STaTEMENTS
[9]
PLUS
SIGNS
BETWEEN
TERMS
INDEXSPl+(((iQ(=Pl)h+
=Pl)/\FPl)r(F~l)+l
=P2)A+=P2)/\FP2),(FF2)+1
xNnEXSP2+(((1@( orRl+l a
LOOP1
0 PlP2c
SELECTS CI TERM IN F2 TO OPERfiTE ON Pl WITH
LOOPl:CTR2+1
LOOP2
LOOPS
ON
GILL
TERMS
IN
Pl
FOR
ONE
TERM
IN
P2
t-oop2: a
FIND ERICH TERM IN Pl fiblO F2 TO(INDEXSP1CCTR1+1]-I)] TO(INDEXSF2[CTR2+1]-I)] OF TPRHTl &ND TERMT2 TERMT~cP~CINDEXSP~[CTR~] TERMT2cP2[INDEXSP2[CTR2] R LOCATE CONST
POSITIONS
c211 c221
[23] 1
R==Pl+L/() RP=P2+L/()
PULL OUT
712
P.
J. MAGARIet al.
C241 C253
[26-j K
C273 C283 t291 c301 [31-j n C321 C333 c343 [35] L363 n A
CONSTSIGt~t~((~(l~TERwl),'l')X~(lfTEKM~),'1')~'~-'
R FIND
LTSZt((x
FIND
MULT
SIGNS
aND
DIV
SIGNS
IN
TERM1
TErPM2
LTSlt(('x
=TERMl)+(+ =TERMz)+(
NUMBER
CONSTINT
OUTFTt,(+TERMlC2
FIND
DENOYINRTOR
TO(l~LTSl)-13)XzTE~~2C2
OF CONSTaNT
T'J(l+LT~2)-11
BnCKTt+(+TERnlC((-l~LTsl)+l)+l)
FINQ
THE NEW TOTaL CONST TERM
TopTERMl])XiTERM2[((-lbLTS2)+1)
TOFTERHZ]
NCONSTt+,(( TERY2[((lfLTS2))
TO(l+LTSl)-I]),
[37]
AMULTIPLY VaR
TERMS
VTERMlt-l+RPCPlfTERMTl VTERM2+-1+RPCp2+TERMT2 CTR3t3 0 NVTERMC It= THERE ARENT =() VTERMl aND JUST TRY TO FIND VaR IT at31 VCIRI~~LES IN VTERM~
IF (pVTERH1)
EaCH TERM aDD IN EXP
IN
AND
c451 C461 c471 C481 c491 c501 c511 E521 c531 c543 c551 C563 c571 t581 c591 C601 Lb11 C621 C631 1643 C651 C663
IF
FOUND
IF
NOT
CaTENaTE
IF
VIIR
IS
IN
BOTH
TERMS
IF+/VaRWHEREtVTERMl[CTR3]zVTERM~ VaR TERM FROM VTERHl TO NEW VaR TERM WHICH WaSNT IN VTERM2
CaTENaTE
TO(CTR3+3)-J
IF
CTR3(FVTERMl
-tLOOP-jOUT
Ll:EXPt+(+VTERMl[CTR3+2])+~VTERM2[(V~RWHERE~l)+2] NVTERMtNVTERM,VTERMlC(CTR3-2) K DROP OFF VaRS TO(CTR3+1)],EXP,) EXPS -2 THaT -1 aRE 0 USED ,\3]tl IN VTERM2
nt4Q
IF
CTR3(pVTERMl )
VTERM2, TERMS
PlP2tPlP2,NCOnST,NVTERM R R GO IF BaCK LNV GET a NEW TERM NOT FROM USED p2 UP aNn KEEP SaME TERM FROM Pl
aLL
TERMS
0 aND aLL 0
IN
~2
aRE IF
NEW IN pi
CTR~~(FINDEXSP~)-~ TERM HavE FROM BEEN pl aND USED REPEaT UP SWEEP aCROsS P2
IF
CTRli(fINDEXSPl)-1
173
The right argument of this function is an executable APL statement of a multivariate polynomial of the form + ((2 x co1 x co2 t 14) x (X* 7)) WW constant variable
+((-5xCO6xC07tl)x(X*l))+... -+ constant variable The left argument is the variable with which the user would like to integrate with respect to. The returned value is an executable APL statement which is the result of integration with respect to the desired variable.
C63 c71 C81 c91 Cl01 Cl11 Cl23 El33 Cl41 El51 11161 Cl71 Cl81 Cl93
[20] A A Li F INDEXSPC(((~G(=P)A+=P),~+~ CTRlel APICK 0 OUT
IPc"
fi TERM IN P 1130 TO TO DE INTEGKfiTE IT
INDEXSPCCTRl] Vf3RICIDLE
LOCtJTE
AND
ClIV
COMST
BY
HEW
EXPONENT
c211 c221
TOpTERM]
St-4 a
TRIKE CARE OF TERMS THhT DONT
COMTLIIN
THE
VC)KICIBLE
OF
INTEGRATION
C271
IF
INDEXSPCCTRl](pP
POLYPARTIAL
The right argument of this function is an executable APL statement of a multivariate polynomial of the form + ((2 x co1 x co2 + 14) x (X*7)) WCYJ constant variable
+((-5xCO6xCO7+l)x(X*l))+~~~ -cyJ variable constant The left argument is the variable with which the user would like to differentiate with respect to. The returned value is an executable APL statement which is the result of differentiation with respect to the desired variable.
114
t43
&ND fi
ES3
C63
c73 E83 c91 Cl03 Cl13 Cl23 Cl33 Cl41 El51 Cl63 Cl73 El83 El91
[20]
POLYNQMICIL
TO
f&ND
THE
VFIRICLBLE
WITH
WHICH
YOU
DIFFERhNYIIXtE
IN P
WRT
TERMS
INDEXSP4(((1)(=F)~+irP)/tpY)to+1 CtRl41 K PICK 0 OUT PP4 A TERM IN P &ND TO TO DIFFERENTIflTE INDEXSP[CtR1+1]-l] WRT IS NOT It.8 TERM XT
I?IFF
LOCLITE
Gt4D
MULT
CONST TO
DY
OLD
EXPCItMEiT
c213
LOCOPC],(~(~TERM[(LOCOFC+~) TOFTERY]
LaCTIMES-i])K~TERMCLOCV+
L223 LOCV4(V=TERU)/\fTERM C231 TERXCLOCV+234t(+T~RXCLocv+23t-l r.243 SL4 f253 fiTLKE C&RR OF TERMS
[26] L2:TERMt L4:PFcFP,TERX .$Ll IF 0
THCIT
DONT
CORTd#XN
THE
VPlRIClDLE
OF
VIFFERENTIb3TION
[273
C28]
CTRl4CTRftl
INDEX?iP[CTRl]c$P
INDEX()4(PFC((VaPP)/~pPF)+~]~())/(V~PP)/\FFF
kRnSEt,(8(~sW)fIN~EXO)+(~H~(~IN~E:<O)?6)~ -2 -1 0 i 2 3
PFc(M(\FPP)EER~.~SR)/PF
Kll - -o!lKaAs * n * (I t z*ac9*L2+ c9%9*L3) K22 - 4.%S*cS*L3 + lz,aCfnCb*L4 + 9.aC6nC6W TKH - L - z.*cs*L3- 2.%6*L4 + CSaCS*LS+2.*cS*c6*L6 +C6'%6*L? X22 = Clmc22 - OifKGAS~M K33 = Cl*cl*L3 t 2.*cl*c2*14+ CZW?*LS K33 - K33 * Cl1 1142 = L + C9*L2+ CS*L3+ (2.*Cb+ CS*c9)*L4 + 2.*C6%9*15 TKH - CS*L*L+ (3.%6 + 2.%S*Cgf*W + 3.6C6*9*L4 K42 - (1142 + X~X)~~GAS~E*SSETA K43 - Cl*L3 + (C2 + ClWS)*L4 + cZ*C9*L5 gEE - cl*L2+ (C2 + clac9)*L3 + cz*c9*L4 - -(K43+ ~J'VKM)*oHXGA5 * H * K * CBKTA K44 - L + 2.%9*12 + c9nC9*w K44 - OHKGAS* (K?lZS - KMlS) * CTSKTA* 1144
Dynamics of helicopter rotor blades K53 * -Cl*L2 + (ZnClW3 - C2)*L3 K53 - -(K53 + (Cl%4 + 2.%2%3)*14 KS4 = -L2 + (2.%3 - C9)013
+ (2.%3%9
TEH - -I + (2.*c3 - c9)*12 + (2.ac3W9 + c4)*L3 TEH - XJ*(TEn + C4*C9*L4) K54 = OMEGAS * H * E * CBETA * (TEH + K54)
K62 = 2.*C5*L2+(3.*C6+ B.*C5%7))"13 + 6.*(2.*C6"C7+ C5*C8)*L4 K62 - (K62 + S*Cb'W*L5)*(-Cll) TEH = I2 + *c6ac7 2.*C7*L3 (ca _ (2 + + C5*CB)aL5)*L4 - (C6 + 2.*c5*C7)*L5 - C6*Ca*L7 K62 - K62 : TEnaOnEGAB*n K64 - 2,*C7*L3 + 2.*(C7*C9 + C8)*L4 + 2.*C8*C9*L5 ;g - (L + (c9+4.*c7)*L2+ (4.*c7ac9*3.wa)*L3 + 3.*ca*cg*L4)*xJ = -OWGAS*WE*SBETA*(TEn + K64) K66-L+a.nC7*L2+(16.aC7*C7+6.*Ca)*L3+24.*C7~Ca*L4+9.*Ca*Ca*L5 TEH - L3+4.*c7*L4+(4.*c7w7+2.*ca)*L5+4.~7*Ca*L6+ca*ca*L7 K66 - Cl1 * K66 - 0HEGAS"nWEn K71 = -OHEGAS*WClO*(L2 + C9*L3) K77 = -0HEGAS * H * Cl0 * Cl0 * L3 K82 = -C11*(4.*C5*CS*L3 + 12.*CS"C6*L4+ g.*C6*C6*L5) TEH = C5"L3 + C6*L4 - C5*C5*L5 - 2.'%5*C6*L6 - Cb*Cb*L7 'K82= K82 - TEn'WnEGAS*n K84 =-OtfEGAS*n*E*SBETA*(C5*13 + (2.*C6+C5*C9)*L4 + 2.%6*C9*L5) TEH - XJ*(2.*C5*12 + (3.*C6+2.*CJ*C9)*13 + 3.*c6'W9*14) K84 - K84 - OliEGAS*n*E*SBETAWEH K86 -2.'W.C5*L2 + (8.%59c7 + 3.*Cb)*L3 + 6.*(C5*C8 + 2.%6%7)*14 + 9.%6WI*L5 TEH - C5*L4 + (2.+c5*c7 + C6)*L5+(C5'%8+ 2.'%6%7)*L6 + CbaC8*L7 K86 = cll*K86 - OnEGAS*HWEH K88 = 4.*cS*cS*W + 12.*CS*Cb*lk+ 9.*C6*6*U TEH - CS*cS*L5 +2.*cSW6*16 + Cb*c6*17 K88 = Cl1 * K88 - OMEGAS *n *TEn K93 = -Cl1 * (cml*L3 + 2.aC1sC2*L4+ c2%2*L5)
K94 = Cl"L3 + (Cl%9 + c2)*L4 + c2*c9*L5 TEH - XJ*(Cl*L? + (Cl'W9 + C2)*L3 + CZaC9*L4)
K94 - OUEGAS*H*EWBETA*(K94 K95 K95 + TEH)
K99 - Cll*(Cl'W*L3 + 2.sCl*C2*L4+ C2'%2*L5) K102 = L2 + C5*L4 + 2.*Cb*LS TEH = 2.*C5*L3 + 3.'W6*L4 KlO2 - OHEGAS*n*E*SBETA*ClO*(K102 + XJ * TEH) K103 - -OnEGAS*n*E*CBETA*C10*(Cl*L4+C2*L5+XJ*(C1*L3 + C2*L4)) K104 - OHEGAS*(Kll2S-KHlS)*ClO*CTBETA*(L2 + C9*L3) KlO5 = 2.*C3*L4-L3+C4*L5 + XJ*(2.%3*L3 - L2 + C4*L4) K105 = OHEGAS*n*E*CBETA*ClO*KlO5
776
P. J. MAGARIet al.
K106 = 2.*(C7*L4 + C8*L5) TEH = L2 + 4.*C7*L3 + 3.'%8*14 K106 = -OMEGAS*H*E*SBETA*ClO*(K106 + XPCEH) K108 = CS*L4 + 2.*C6*L5 TEH = 2.*CS*L3 + 3.*C6*L4 K108 = -OtlKGAS*H*E*SBETA*ClO*(K108 + XJWEH) K109 = OMEGA8*ti*E*CBETA*C1O*(Cl*L4 + C2*L5 + XJ*(Cl*L3 + C2*L4)) KlOlO = OMEGAS * (KH2S - KHlS) * Cl0 * Cl0 * CTBETA * L3 K113 = -Cll*(Cl*C3*L3'+(Cl*C4 + C2*C3)*L4 + C2*C4*L5) K114 = C3*L3 + (C4 + C3%9)*L4 + C4*C9*L5
TEH = C3*L2 + (C4 + C3*C9)*L3 K114 = OHEGAS*H*E*CBETA*(Kll4 Kll5 + C4*C9*L4 + XWTEH)
K119 = Cll*(Cl*C3*L3
KlllO = C3*L4 + C4*L!i + XJ*(C3*L3 + C4*L4) KlllO = OHEGAS*H*E'WBETA*C10*KlllO Kllll = Cll*(C3%3*L3 + 2.*cPC4*14 + C4'W4*L5) K122 = -Cll*(4.*cS*c7*L3 + 6,*(CbaC7+CSaC8)*L4 + 9.*C6W8*L5) TEH = C7*L3 + C8*L4 - C5%7*L5 - (C6aC7 + C5%8)*L6 - CbaCB*L7 K122 = K122 - OklEGAS*VTEH K124 = C7*L3 + (2.*cO + C7*C9)*L4 + 2.*C8*C9*L5 TKM = XJ*(2.%7*L2 + (3.*c8 + 2.%7*C9)*13 + 3.%8%9*L4) K124 = -OHEGAS*tPE*SBETA*(Kl24 + TEH)
K126 -2.*C7*L2+(8.~7*C7+3.*C8)*L3+18.~7aC8*L4+9.*C8*C8*L5
TEH = C7*L4 + (2.*c7aC7 + C8)*L5 + 3.%7*C8*L6 + C8%8*L7 K126 - Cll*K126 - OHEGAS*@TEH. K128 = 4.%5*c7*13 + 6.*(Cb*c7 + C5'%8)*L4 + 9.%6%8*L5 TKH = CS*C7%5 + (C6=+C7 + CSaC8)*16 + Cb*C8*L7 K128 = Cll*K128 - OHEGAS*WVEH K1210 = C7*L4 + 2.'%8*L5 + XJ*(2.*C7*L3+ 3.*C8*L4) K1210 = -OHEGAS*H*E*SBETA*ClO*Kl210 K1212 = 4.%7*C7*L3 + l2.%7%8*L4 + 9.*C8*C8*L5
E = distance between mass and elastic axes, e SBETA = sin fl CBETA = cos p KMlS=k;, KM2S = k;, CTBETA = cos 21 R
T= C&nx dx
Ll = I/7
M = mass per unit length XJ = distance of the element from the axis of rotation
IL
1<=1/2+XJ.