Sie sind auf Seite 1von 8

H. D.

N@Ison
Professor of Engineering Science, Arizona State University, Tempe, Ariz. MEM. ASf.lE

The Dynamics of Rotor-Bearing Systems Using Finite Elements


A procedure is presented for dynamic modeling of rotor-bearing systems which consist of rigid disks, distributed parameter finite rotor elements, and discrete bearings. The formulation is presented in both a fixed and rotating frame of reference. A finite element model including the effects of rotatory inertia, gyroscopic moments, and axial load is developed using the consistent matrix approach. A reduction of coordinates procedure is utilized to model elements with variable cross-section properties. The bearings may be nonlinear, however, only the linear stiffness and viscous damping case is considered. The natural whirl speeds and unbalance response of a typical overhung system is presented for two sets of bearing parameters: (i) undamped isotropic, (ii) undamped orthotropic. A comparison of results is made with an independent lumped mass analysis.

J, M. MeVaugh
Assistant Group Supervisor-System Dynamics, Airesearch Manufacturing Company of Arizona, Phoenix, Ariz. MEM. ASME

Introduction
Flexible rotor-bearing systems have been analyzed by many different mathematical methods. An excellent comprehensive review of these methods with reference to specific contributors is contained in a recent work by Eshleman [l], 1 hence, need not be repeated here. Recently, the utilization of finite element models, such as introduced by Archer [2], in the area of rotor dynamics has yielded highly successful results. It is the intent of this paper to provide additional information on the use of finite elements for analyzing rotor-bearing systems. Of primary interest relative to this paper, is a contribution by Ruhl [3], [4] who utilized a finite element model of a turborotor system to study stability and unbalance response. Ruhl's finite element includes only elastic bending energy and translational kinetic energy, while the effects of rotatory inertia, gyroscopic moments, axial load, axial torque, and shear deformation are neglected. These effects can be quite significant for some configurations as indicated by several investigators referenced in [1]. This particular paper generalizes Ruhl's element by including the effects of rotatory inertia, gyroscopic moments, and axial load. In addition the element and system equations are presented in both a fixed and rotating reference frame. An example, which utilizes a coordinate re-

duction procedure, is provided to indicate the accuracy of the element.

System Configuration and Coordinates


The typical flexible rotor-bearing system to be analyzed consists of a rotor composed of discrete disks, rotor segments with distributed mass and elasticity, and discrete bearings. Such a system is illustrated in Fig. 1 along with the two reference frames that are utilized to describe the system motion. The XYZ: E F (xyz:Q{) triad is a fixed (rotating) reference with the X and x axes being colinear and coincident with the undeformed rotor center line. 01 is defined relative to 'S by a single rotation cot about X with < o denoting the whirl speed. A typical cross section of the rotor in a deformed state is defined

1 Numbers in brackets designate References at end of paper. Contributed by the Vibration and Sound Committee of the Design Engineering Division and presented at the Winter Annual Meeting, Houston, Texas, November 30-December 5, 1975, of THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS. Manuscript received at ASME Headquarters August 5,1975. Paper No. 75-WA/DE-19.

Fig. 1

Typical system configuration

Journal of Engineering for Industry Copyright 1976 by ASME

MAY 1976

593

The displacements (V, W, B, T) of a typical cross section relative to ff are transformed to corresponding displacements (u, w, /3, 7) relative to (ft by the orthogonal transformation

{q} =
with

MM

(2)

{q} =

w
B

-fc} =

fv\ w

,[*] 0 0
COS Uif

[y]
"coscrf - sinarf sin cot coswf 0 0 3 0 0 0 - sinatf
COSktf

(3)

sinotf

Fig. 2

Cross section rotation angles

and for later use the first two time derivatives of equation (3) are {q} = a)[S]{p} + [R]$\ (a)

relative to *5 by the translations V (s, t) and W (s, t) in the Y and Z directions respectively to locate the elastic centerline and small angle rotations B (s, t) and r (s, t) about Y and Z respectively to orient the plane of the cross-section. The cross-section also spins normal to its face relative to 9?. The abc: Q triad is attached to the cross-section with the "a" axis normal to the cross-section. S is defined by the three successive rotations, illustrated in Fig. 2, 1 r about Z defines a"b"c" 2 B about b" defines a'b'c' 3 (j> about a' defines abc, and the angular rate of relative to ?f is

= [R]{$\with

o>2{p}}+ 2co[S]{p} (6)

[Sl=i[R}

-sinarf - cosarf 0 0 cos a r t - sinwi 0 0 0 0 -sinarf - cosatf 0 0 cos cot - sinM?

(c)

(4)

"- s i n B 1 0 cos B sin 0 0 c o s 4> .cos B c o s <f> 0 - s i n <f>. LB

(1)

For small deformations the (B, T) rotations are approximately colinear with the (Y, Z) axes respectively. The spin angle fa for a constant speed system and negligible torsional deformation, is fit where 2 denotes the rotor spin speed.

Component Equations. The rotor-bearing system is considered to comprise a set of interconnecting components consisting of rigid disks, rotor segments with distributed mass and elasticity, and linear bearings. In this section the rigid disk equation of motion is developed using a lagrangian formulation. The finite rotor element equation of motion is developed in an analogous manner by specifying spatial shape functions and then treating the rotor element as an integration of an infinite set of differential disks. The bearing equations are not developed and only the linear form of the equations as treated by Lund [5] are utilized in this paper. Rigid Disfe. The kinetic energy of a typical rigid disk with mass

.Nomenclature. ' = differentiation with respect to position X = whirl ratio = S2A = differentiation with respect to time ["$?] = matrix of translation displacement SJ = fixed reference frame (XYZ) . functions; fa (s), i = 1, 2, 3, 4 (R = rotating reference frame (xyz) [ < E > ] = matrix of rotation displacement 6 = cross-section reference frame (abc) functions; fa' (s), i = 1, 2, 3, 4 2 3 = kinetic energy u> = whirl speed 0 = potential energy 2 3D, dp = element diametral and polar iner- o>a, ub, <>>c = angular rate components of relative to S F tia per unit length [M], [g], [x] = assembled mass, gyroscop- Vd, id = location of disk mass center in b,c directions ic, and stiffness matrices 2 n(s), f(s) = distributed location of element a = eigenvalue cross section mass center in b, c direc( S , r ) , (/3,Y) = small angle rotations about tions (Y,Z), (y,z)axes riL, to = rj(O), f(0) 4> = spin angle fi = spin speed = 4> VR, to = vV) W) M = element mass per unit length 111. IPI = displacement vector relative to '5, (R2 2 Where appropriate the superscripts d, e, b, s (<Zcl, {<2s! = unbalance response associated refer to disk, element, bearing, and system rewith cos flt, sin 0 i 2 spectively and the subscripts T, R, B, A refer to translational, rotational, bonding, and axial load |Q], {Pj = external force vector relative to respectively. <S, (R2 \Qc], \Qs] unbalance force associated with cos fit, sin Qt2 [R] = orthogonal rotation matrix (V,WO, (v,w) = translations in (Y, Z), (y, 2) directions Md, ID, IP = disk mass, diametral inertia, polar inertia 6 = pojar inertia of element as a rigid Ip body [M], [G], [K] = mass, gyroscopic, stiffness matrices 2 [iff], [G], [k] = transformed mass, gyroscopic, and stiffness matrices 2 t = time s = axial position along an element I = element length [Cb], [Kb] = bearing damping and stiffness matrices cyv, Cvw, cwv, c\vw = elements of [Cb] kvv, kvw, kwv> kww = elements of [Kb] c,k = isotropic bearing damping and stiffness coefficients [I] = identity matrix

594

MAY 1976

Transactions of the ASME

mation of the equation of motion from fixed frame coordinates, equation (6), to whirl frame coordinates, equation (8a), an inertial restoring force appears. The term - u 2 ([MdT] + [Mi] + \ [Gd]) ( p d | contains the so called Green's [6] gyroscopic stiffening effect. Finite Rotor Element. A typical finite rotor element is illustrated in Fig. 3. It should be noted that here the element time dependent cross section displacements (V, W, B, T) are also functions of position (s) along the axis of the element. The rotations (B, F) are related to the translations (V, W) by the equations B = dW (a) (9) (b)

as
dV ds

r =

Fig. 3 Typical finite rotor element and coordinates

The coordinates (qf, <?f, . , <?I) are the time dependent end point displacements (translations and rotations) of the finite rotor element and are indicated in Fig. 3. The translation of a typical point internal to the element is chosen to obey the relation

center coincident with the elastic rotor centerline is given by the expression

J!:;?)=ws'H#)}
where the spatial constraint matrix is given by

(10)

2 1W )

' mi 0 " . 0 mi_

The use of equations (1), with the retention of only second order terms, reduces equation (5a) to

{l\

iD o o o iD o Lo o i , J

(5a)

[*]

ft 0 0 ft ft 0 O f t 0 -ft - ft 0 0 ft - ft 0

(11)

&d

and is a matrix of displacement functions. In this case the individual functions represent the static displacement modes associated with a unit displacement of one of the end point coordinates with all others constrained to zero. These functions are

2\wf

L 0 md\ \WJ - 4>TBIP

2\r j

In (56)

ft = 1

3(f) 2 + 2(f)3

(a) (6) (12) (c)

The lagrangian equation of motion of the rigid disk using equation (56) and the constant spin speed restriction, <j> = Q, is

ft = s[l - 2(f) + (y)2]


*3

( [ < ] + M){q*} - nlG1]^} = W}

3(f)

2(f)'

(6)

where the individual matrices are provided in detail per Appendix A. Equation (6) is the equation of motion of the rigid disk referred to 5 with the forcing term including mass unbalance, interconnection forces, and other external effects on the disk. For the disk mass center located at (jjd, fd) relative to C, the unbalance force in ffis

h = *Hy) 2 + (y)3]
From equations (9), (10) the rotations can be expressed in the form

[r}= [*]W
with [*] = LL* JJ r
0 - f t ' ft' 0 0 -i/i 3 ' ft' 0 .ft' 0 0 ft' ft' 0 0 ft'.

d3)
(14)

W} =

cosSlt

0 0

sinQt

{Q*}cosat

{Qds}smat
T

(7)

By using equations (2, 3, 4) and premultiplying by [R] , equation (6) transforms to ([Md] + [M*}){&} u2(([Md] + u(2([MdT] + [Md]) + [MSR}) \[Gd]){&}

representing a matrix of rotation shape functions. For a differential disk located at (s) the elastic bending, axial load, and kinetic energy expressions are respectively

+X[G']){jy*} = {P*} (8a)

.-na'[fL]{a-
EI 0 0 EI.
d<S>\

For the case of a thin disk (Ip = 2Iu) equation (8a) reduces to {[Md] + [Mi])?} + co(2[M] + (1 \)[Gd])fr] (86) d3'

1 (V'\

\P 01 ~P 0"

(V'\

2W

.L 0 oP . P\

(6) (15)

rf([Md]

+ (1 - 2A)[M!])fcf} = {Pd}

Equation (8a, 6) is the equation of motion of a rigid disk referred to (R with whirl ratio A = il/oi. It is noteworthy that in the transfor-

Journal of Engineering for Industry

MAY 1976 / 5SS

By using equations (10), (13), equations (15) can be written in the matrix forms

d(?l = | E / { q W ' H * " ] M r f s


dS*A = - \ p { q e f [ * ' f [ * ' ] {q}ds rf3e =|M{q<T[*]r[*]{qe}ds

(a)
(6)

+ 1 ^ 4 , ^ + |&{q e }[*] r t*]{q e }^ - n { 4 T [ * r ] r t * B ] { q c } ^ (c) (16)


The energy of the complete element is obtained by integrating equations (16) over the length of the element to obtain <P%+(P'A + 3" = \ {q*HI*S] [K\}){^\
Fig. 4 Assemblage of subelements

+ \}l<f + <^{qeF[^]{qe} (I?)


where

[Af] = / 0 V[*r[*] M = /0^z,[*]r[*] ^


r

rfs

(A) ()
(c) (d) (e) (18)

s = 0 and s = / respectively, the equivalent unbalance force from equation (21) is presented in Appendix A. Equation (19) is transformed to whirl frame coordinates by using equations (2), (3), (4), extended to include four coordinates at each end of the element, and then premultiplying by [R]T. In addition, since .>= 2$p, the identity M r [ M / ] [ S ] = |[G] results and the transformed equation (19) is

(23)

[ # ' ] = / 0 W t * r ] [ * B ] <** [ifa']=-/0I/[*"]T[*"]^ [*i] = / > [ * ' n * ' ] *

([M] + [Mi]) {p e } + <*2[M}] + (1 - \)[G]) 8 x 1

{p} +

(([Jfj

- [K%]) - wH[M$] + (1 - 2 X ) [ M | ] ) j p 8 } = {P*} (24) The elements of the matrices of equation (18) are included in Appendix A for the case of a uniform cross-section element under with constant axial load. [M*\ = [R)T[M][S] (25) The lagrangian equation of motion for the finite rotor element using equation (17) and the constant spin speed restriction, ij> = fi, Bearings. The bearings utilized in this paper are limited to is those which obey the governing equations of the form {[Mf\ + [ M J J ) { r ? } - 0[G]{q} 6 6 + ([Kilwith [Gl= ([#][N]i) M ) { q ' } = {Q e } d9)

[c Kq } + mm

=m

(26)

2x 1
in fixed frame coordinates, where {q6}

(20)

and is referred to fixed frame coordinates. All the matrices of equation (19) are symmetric except the gyroscopic matrix [Ge] which is skew symmetric. The force vector jQ") includes mass unbalance, interconnection forces, and other element external effects. For an element with distributed mass center eccentricity (j; (s), f (s)), the equivalent unbalance force utilizing the consistent matrix approach introduced by Archer [2] is

,[cbl =

\ Cyy r"

Cyw rb

and jQ6j is the bearing external force vector. Using equation (2) in equation (26) and premultiplication by [R]T gives the transformed form [R]T[G"][R] {pb} + [RF [K"}[R] {p"} = {P>} 2 x 1 (27)

mx=/>w({^}coB
+ { ~ [ g ! | sinaA
f

= {Q}cosSM + {G&} sinSlt

(21)

in whirl frame coordinates. For nonisotropic bearings equation (27) contains periodic coefficients. This results in a parametrically excited equation of motion. For isotropic bearings, however, equation (27) reduces to an equation with constant coefficients:

For the case of a linear mass unbalance distribution over the element, as represented by the expressions, rj(s) = nL(l - j ) + %(y)

cnim + kiiw} = {p>}


2 x 1

(28)

(22)
t ( s ) = fc (1 - j) + R{j)

with (TJL, L) and (TJR, fo) denoting the mass center eccentricity at

where c and k are the isotropic bearing, damping, and stiffness coefficients respectively. Variable Cross Section Rotor Element. A typical rotor element will possess variable cross sectional properties as illustrated in Fig. 4. The equation of motion for such an element can be ob-

896 /

MAY 1976

Transactions of the ASME

tained from the previous development by either evaluating the integrals of equations (18), (21), using the variable properties or by treating the element as an assemblage of uniform sub-element equations of motion as given in Appendix A. T h e assembled set of subelelements then possesses [4 X (number of subelement stations)] coordinates which can be reduced by the following procedure. The assembled set of subelement equations, where assembly is accomplished by utilizing the appropriate geometric displacement compatibility constraints (see [7], p. 38) is of the form, in fixed frame coordinates,

([MH + [MH]) (&}a


Fig. 5 System configurationexample

= {Qe} (29)

(4 x n u m b e r of s u b s t a t i o n s ) x 1 A displacement dependency between the internal displacements, |qB)6, and the element endpoint displacements, jq e ) a and |q e | C) can be imposed by considering the static, homogeneous case of equation (29).

Whirl Speed Analysis. For an assumed solution form, \h\ = j/iol eat, for the homogeneous case of equation (34), the associated eigenvalue problem is

[0]
-[K8]-1^3]

{S[G . ] ] { * . } Q[KS]

= &,}

(35)

' [ * " ] [KU [K }ac~ [K]ba [K]bb [K]bc [K%a [Ke]cb [K%c

{0}

(30)

The internal displacement vector from the second row of equation (30) is

For orthotropic hearings (kyw ~ kwv = 6) and zero damping, the eigenvalues of equation (35) appear as pure imaginary conjugate pairs with the magnitude equal to a natural whirl speed. 3 T h e superposition of a solution with its conjugate represents an associated elliptical precession mode. Isotropic bearings produce circular precession modes for an isotropic rotating assembly. For whirl frame coordinates, with isotropic bearings, equation (33) transforms to the form [ M s ] { p s } + co(2[M s ] X[G s ]){p s }

M 6 = -[*&[*!.{}. - lK'l[K'\.{<f\.

(3D

Using equation (31) and two identity statements, the following equation of constraint can be written

+ ([Ks] - w 2 ([M s ] + X[G s ]){p s } = { P S } (36) 4M x 1 The natural circular whirl speeds and mode shapes can be obtained from the homogeneous form of equation (36). These modes are constant relative to (R and the two planes of motion are 90 deg out of phase. It is necessary, therefore, to consider only one of the two planes of motion and assume a constant solution jps) = jpo! = cnst. The associated eigenvalue problem is

rt.

a = i*}.

[Ke];l[K]ba -[K}j,{K\c

[i]

[0]

J{qeU
X 1

[0]

[I]

(32)
where the columns of the constraint matrix represent static mode shapes of the element. T h e transformation of equation (32) applied to equation (29) reduces the number of coordinates and associated force components to eight, providing the same element equation form as equation (19). The examples in this paper utilize this reduction of coordinates procedure to model elements of variable cross-section. System Equations of Motion. The assembled undamped system equation of motion, consisting of component equations of the form equation (6), (19), and (26), is of the form

[tf s ]fto}= <o2([Ms] + X[Gs]){p0} 2w x 1

(37)

The In eigenvalues are real and the positive valves, wr, with associated vectors jpo}(r) represent natural circular whirl speeds and mode shapes relative to R at the specified whirl ratio X. Unbalance Response. In fixed frame coordinates, the unbalance force for equation (33) is of the form {Q*} = {Q*} c o s at + {Qj!} s i n at Thus a steady state solution of the same form {q s } = {q|} c o s at + {q8} s i n at substituted in equation (33) yields (39) (38)

(33) [A?s]{q*} - fi[Gs]{qs} + [it s ]{q s } = {Qs} An x 1


for fixed frame coordinates. For computational purposes equation (33) is written in the first order state vector form (see [8] p. 410)

_[MS] - Q,[GS] :\}{*}

[o]

[M

f[~M] [0] L[0] [K*]


where

{h} = {H} (34) 8w x 1

([KS] - a2[Ms]) - a2[Gs]' a\Gs][([Ks] - a2[Ms])} .

' I S ) <

{o}

The solution of equation (40) and back substitution into equation (39) provides the undamped system unbalance response.

' For a discussion of the properties of gyroscopic systems, see Ziegler [9].

Journal of Engineering for Industry

MAY 1976 / 597

00

. V4

i/. - -= ~'~\~"~
^
bU

P
\/

-- - /IT"" /
/

* ---' " / / 5 1 "~"~' --- -- ^ /

\A
i

=L~ I

/ 7
^1*--

/ ^ " ^ _ ^ =

r
r

r^

\^^~~

i ! (RPM x 10 ) - WHIRL FRAME ANALYSIS, CASE {a) FIXED FRAME ANALYSIS, CASE (a) - FIXED FRAME ANALYSIS, CASE (b>

Fig. 6 Table 1
Element No. Subelement No.

System whirl speeds Rotor configuration data


A x i a l Distance t o Subelement (cm) 0.00 1.27 I.D. (cm) O.D. (cm)

e = eg eccentricity of station 3

Fig. 7

System unbalance response

0.51 1.02

5.08 7.62

0.76 2.03

For whirl frame coordinates with isotropic bearings, the unbalance response is obtained from equation (36) with to = Q (i.e., A = 1). In this case the unbalance force in equation (36) is a constant relative to (R, hence, the unbalance response is also a constant relative to (R. From either of the two planes of equation (36) the undamped unbalance response is

8.89 10.16 10-67 11.43 12.70 13.46

1.52 1.78

2.03 3.30 3.30 2.51 2.54 1.27

M2w X 1

= ([Ks] - Q2([MS]

+ [G5]))-1^8}

(41)

16.51 19.05

1.27 1.52

22.86 26.67

1.52 1.27

28.70 30.48 31.50 34.54

1.52

1.27 3.81 2.03 2.03

Table 2
Whirl Ratio

Whirl speedsCase (a)


Whirl Speed Positive

(RPM) Negative 14,758 44,695 58,424 15,470 46,612 64,752 15,858 47,520 69,640 16,060 47,957 72,737

Numerical Examples. To demonstrate the application and accuracy of the finite element model, a typical rotor bearing system as illustrated in Fig. 5 is analyzed to determine its whirl speeds and unbalance response. A density of 7806. kg/m 3 and elastic modulus 2.078 X I0 1 1 N/m 2 were used for the distributed rotor and a concentrated disk with a mass of 1.401 kg, polar inertia 0.0020 kg/m 3 and diametral inertia 0.0136 kg-m 2 was located at station 3. The distributed rotor was modeled as a six element member with each element consisting of several subelements. The geometric data of these elements and subelements is listed in Table 1. Two identical bearings, idealized as undamped and linear, were located at stations four and six. The following two cases of bearing stiffness were analyzed:
kbyy

18,148 51,430 111,455 17,159 49,983 96,457 16,700 49,204 85,552 16,481 48,800 80,649 16,267 48,384 76,382

(")

= k" = 4.378 x 10 7 N/m,/4, = kbwr = 0. = -8.756 x 10 e N/m

(b) k"vv = kbww = 3.503 x 1 0 7 N / m , ^ w = kbm

1/2

1/4

Spin Speed Forward

Whirl Speeds Backward

70,000

19,838 50,555 91,320

12815 45599 63990

This symmetric bearing is equivalent to an orthotropic bearing with principal axes oriented at (45, 135) relative to the X axis. Case (a) Isotropic Bearings. T h e undamped whirl speeds were computed from the 14th order eigenvalue problem of equation (37) for whirl ratios of 0, \ %, 1 , 2. The first three whirl speeds for each whirl ratio are listed in Table 2 and plotted in Fig. 6. The unbalance response for a disk mass center eccentricity of 0.025 in. at station three was determined from equation (41) for the speed range 0-30,000 rpm and is plotted in Fig. 7. Also by using the 56th order eigenvalue problem of equation (35) the natural whirl speeds associated with a spin speed of 70,000 rpm were computed. These speeds are also listed in Table 2 and plotted in Fig. 6. Case (6) Orthotropic Bearings. The undamped whirl speeds were computed from the 56th order eigenvalue problem of equation (35) for several spin speeds. The first three whirl speeds are

598

MAY 1976

Transactions of the ASftHE

Table 3 Spin Speed (RPM)

Whirl speedsCase ( 6 )

Whirl Speeds

(RPM)

Forward 1000 15258 48384 76387

Backward

tained by the authors for the included examples and other rotor systems indicates that the model is reliably accurate. The rotor element can be generalized to include the'effects of shear deformation, axial torque, and various forms of internal damping. It appears that the finite element approach can provide a valuable new dimension to the analytical options available for studies of rotorbearing systems.

14099 40033 66747 13762 39997 65264

20000

16606 48352 78177

Acknowledgment The authors wish to acknowledge the valuable assistance rendered by Dr. Edward Zorzi, Dynamics Specialist, AiResearch Manufacturing Co., during the preparation of this paper.

40000

17310 48293 81917

13063 39889 62414

60000

18102 48157 86177

12280 39704 59548

80000

18897 47923 90457

11497 39436 57057

100000

19655 47549 94510

10747 39077 55079

listed in Table 3 and are plotted in Fig. 6. The unbalance response with the same unbalance as for case (a) was computed for the speed range 0-30,000 rpm. The response orbits are elliptical with principal axes oriented at (45, 135) relative to the Y axis. The semimajor and semiminor axes of the orbits for station 3, are plotted in Pig. 7 versus spin speed for the above speed range. For this case (nonisotropic) the backward whirl modes are excited as well as the forward whirl modes by the unbalance force. Hence, it is interesting to note that the system precession changes from forward to backward to forward as the spin speed passes through the first backward critical speed.

References 1 Eshleman, R. L., "Critical Speeds and Response of Flexible Rotor Systems," Flexible Rotor-Bearing System Dynamics, ASME, Vol. 1,1972. 2 . Archer, J. S., "Consistent Mass Matrix for Distributed Mass Systems," Journal of the Structural Division, Proceedings of the ASCE, Vol. 89, ST4,1963, p. 161. 3 Ruhl, R. L., "Dynamics of Distributed Parameter Turborotor Systems: Transfer Matrix and Finite Element Techniques," PhD Thesis, Cornell University, Ithaca, N. Y. Jan. 1970. 4 Ruhl, R. L., and Booker, J. F., "A Finite Element Model for Distributed Parameter Turborotor Systems," JOURNAL OF ENGINEERING FOR INDUSTRY, TRANS. ASME, Feb. 1972, p. 126. 5 Lund, J. W., Rotor-Bearing Dynamics Design Technology, Parts III, IV, AFAPL-TR-65-45, May 1965. 6 Green, R. B., "Gyroscopic Effects on the Critical Speeds of Flexible Rotors," TRANS. ASME, Vol. 70,1958, pp. 369-376. 7 Hurty, W. C, and Rubenstein, M. F., Dynamics of Structures, Prentice-Hall, Inc., 1964. 8 Meirovitch, L., Analytical Methods in Vibrations, MacMillan Book Co., 1967. 9 Ziegler, H., Principles of Structural Stability, Blaisdell Publishing Co., 1968.

APPENDIX A
Rigid Disk Matrices "m 0 0 0 0 % 0 0 [M*] = 0 0 0 0 0 0 0 0

Summary and Conclusions


A finite element model including the effects of rotatory inertia, gyroscopic moments, and axial load has been developed for a rotating shaft element. The equation of motion of the element is presented in both a fixed and rotating frame of reference. The rotating frame equation is particularly useful for isotropic systems since the two planes of motion can be treated separately, while the fixed frame equations provide tha generality of handling problems with nonsymmetric bearing stiffness and damping. The major disadvantage of the fixed frame finite element formulation is that the order of the system matrices is larger thereby requiring a large computer. The finite rotor element was applied to a typical rotor bearing system to illustrate the procedure and accuracy of the model. Natural whirl speeds and modes were calculated by using both the fixed and rotating frame formulations. A separate 26 station lumped mass analysis of the case (a) example was run to obtain a comparison with the finite element results. The whirl speeds obtained using the 6 station finite element analysis were all higher than the 26 station lumped mass analysis. For the whirl speeds listed in Table 2, the largest difference between the two analyses was less than 4 percent for the third forward mode at a whirl ratio of +2. The finite element model can easily be utilized to model rotorbearing systems for purposes of determining critical speeds, stability, unbalance response, transient response, etc. The results ob-

ro o o o " 1
[Mj] =
0 0 0 0 0 0 ID 0 _0 0 0 ID_

ro o o o i
[G*] =
r

0 0 0 0 0 0 0 /p 0 0 Ip 0 0 0 0 0 0~ 0 0 0

0 ~md m 0 [<] = 0i 0 _0 0

IK1 =

"0 0 0 0

0 0 0 0

0 0 " 0 0 0 -ID ID 0

For IP = 21B, [MdR] = %[G "0 0 [&] = 0 _0 0 0 0 0 0-/, 0 0 0 ' 0 0 -Ip

For Ip = 5 ID, [G*] = - 2 M

Journal of Engineering for Industry

MAY 1976

599

Finite Shaft Element Matrices 156 sym 0 156 0 - 2 2 / 4/2 4/2 0 id 221 0 [Mfi = 420 0 13/ 156 54 0 0 0 156 0 54 - 1 3 / 22/ 4Z2 0 0 0 13/ -3/2 0 0 4/2. 3/2 - 2 2 / 0 L-13Z 0 36 sym 36 0 0 -3Z 4Z2 0 0 4Z2 _ ju"r' 3/ [Mg\ = 0 0 -3Z 36 120/ - 3 6 36 0 -36 3Z 0 0 -f 0 0 3Z 4/ 2 0 -31 0 0 Z2 - 3 / 0 0 4/2 3/ 0 skew sym 36 0 -31 0 0 2[ir2 0 -31 4/2 [G] = 120/ 0 36 - 3 / 0 -36 0 0 - 3 / 36 0 -31 0 3/ 0 L 0 -3Z -f 0 3Z 4/2 0_ 12 sym 0 12 0 - 6 / 4/2 6 / 0 0 4/2 [Ki] = ~ -12 0 0 -6/ 12 0 - 1 2 6/ 0 0 12 2 0 - 6 / 2/ 0 0 6/ 4/2 0 0 2/ 2 -61 0 0 4/2 J L 6/ 36 sym 0 36 0 31 4/2 3/ 0 0 4/2 M 30/ - 3 6 0 0 - 3 / 36 0 36 + 3 / 0 0 36 0 0 3/ 4/ 2 0 - 3 / -/2 2 2 L 31 0 0 -Z - 3 / 0 0 4/ _

\_mT\

\xl /on 420

0 0 156 0 0 -22/ 4/2 0 0 -22/ 0 13/ 0 -54 0 0 -13/ 54 0 3/ 2 0 13/ 0 0 - 1 3 / -3/2

skew sym

0 156 0 22/ 0 0 0 - 2 2 / '4/2 0

lo^ + io^
20 20 jiti1
L1

20 30

tRl

QL

QR

h^l + h **1
20
QL

fo nJ + lo v*f
20
QRL

~ QL t /2 + t 72 30 20 CR "30 ^ ' 7 ~20 ^ ~ 20 3 ~ 20 ^


R1

|o vj + fo vJ

To ^ - fo VR12
{Q1} = ^ ~ t /2 - t f 20 QL 36 QR 20
fei

20

QR

= ih

fo V + |o "^ lo ^ 2 + fo vJ2
30 ^
+

20 * ? .

600

MAY 1976

Transaction:-- of the ASME