Journal of Sound and Vibration (1988) 121(2), 201-210

FLEXURAL M O T I O N OF A RADIALLY ROTATING BEAM

ATTACHED T O A RIGID BODY

A. YIGIT, R. A. Scox-r AND A. GALIP ULSOY

Department of Afechanical Engineering and Applied Mechanics, The University of Michigan,
Ann Arbor, Michigan 48109-2125, U.S.A.

(Received 3 January 1987, attd in revised form 25 March 1987)

Flexural motion of a radially rotating beam attached to a rigid body is investigated.

Fully coupled non-linear equations of motion are derived by using the extended ttamihon's
Principle. Spatial dependence is suppressed by using the extended Galerkin method. A
torque profile is used to drive the rotating body so that the rigid body motion is not known
a priori. The effect of the coupling terms upon the vibration waveforms are investigated
by using both a linearized analysis and numerical solution of the differential equations.
It is found that for small values of the ratio of the flexible beam and rigid shaft inertia
uncoupled equations can lead to substantially incorrect results, particularly with regard
to frequencies.

1. INTRODUCTION
In mechanical systems operating at high speeds, some oscillatory elastic motion is
inevitable. This motion becomes a major concern at high speeds and when performance
requirements are such that high precision is important. Because of increasing operating
speeds and reduced weights it may be inaccurate to treat certain links in a system as
rigid. The effect of flexibility on the dynamic behavior of mechanical systems has been
the subject of numerous investigations (see, e.g., references [1-3]). Among these studies
the dynamics of a beam attached to a moving body deserves some special attention, for
it can represent a variety of technological problems, such as a helicopter rotor [4], a
turbine blade [5], a space satellite with flexible appendages [6], etc.
Natural frequencies of rotating beams have been investigated by various researchers.
Putter and Manor [7] used the finite element method to find the natural frequencies o f
a rotating beam. Hoa did a similar study, but included a tip mass [8]. Wright et al. [9],
studied the vibration modes and natural frequencies using the Frobenius method. Kojima
has investigated the transient vibrations o f a beam-mass system fixed to a rotating body
[10]. However he used a prescribed velocity profile for the rigid body motion resulting
in a linear system of equations to describe the beam vibrations.
There are some studies in which account has been taken o f the coupling between the
rigid body motion and elastic motion. Bakr and Shabana used coupled equations for
describing the dynamics of mechanisms [ l l ] . Dubowsky and Gardner gave the fully
coupled non-linear equations of motion for a mechanism link [ 12]. In spacecraft dynamics,
this approach is quite common and the efl'ect of coupling is generally taken into con-
sideration. Ryan [13] introduced a general approach for modeling flexible systems.
In their very recent paper, Christensen and Lee [14] developed a finite element model
for modeling flexible structures undergoing large elastic deflections coupled with "rigid
b o d y " translational and rotational motions. Very soft materials were considered and so
indeed the elastic deformation could be large. One of their numerical examples involved
201
0022-460x/88/050201 + 10 S03.00/0 (~) 1988 Academic Press Limited
202 A. YIGIT, R. A. SCO'l'q" A N D A. G A L I P ULSOY

a beam attached to a rotating rigid shaft just as in the present case. However, their major
concern was in comparing the linear and non-linear results. The current work differs from
theirs in that the major focus is on the difference between the coupled and uncoupled
simulations. In addition, a study is given of how these differences change as the ratio of
the flexible beam and rigid shaft inertia changes.
The governing equations are derived by using the extended Hamilton's Principle, and
ordinary differential equations in time are obtained by using the extended Galerkin
method. The etl'ects o f the coupling terms upon the vibration wave forms are investigated
by using both a linearized analysis and numerical solution of the differential equations.
The results presented show the importance o f using coupled equations; prescribing the
motion of the rigid body gives bad results for the frequencies.

2. EQUATIONS OF MOTION
The system to be considered consists of a slender flexible beam cantilevered onto a
rigid rotating shaft and is shown in Figure 1. Longitudinal deformations are neglected
and it is assumed that Euler-Bernoulli theory is adequate to describe the flexural motions.

p/•/• wIxot)

I I I I " I

I 9
l I l l

t t

MU)

Figure 1. Sketch of the system.

To describe the kinematics, a frame moving with the shaft is introduced. This frame
rotates with the beam as if the beam were rigid and is so oriented that one of its axes
coincides with the elastic axis of the undeformed beam. The general problem consists of
calculating the motion, both of the rigid shaft and the flexible beam, when a prescribed
torque M(t) is applied to the shaft. Inspection of Figure ! shows that the position vector
r, relative to an inertial frame, of a point on the beam's deformed qenterline may be
written as
r = (a +x)e~ + we,., (1)
where a is the length o f the rigid shaft, x gives the location of the undeformed point, w
is the transverse elastic displacement and e~ and e r are unit vectors in the x and y
directions respectively. Note that axial deformations are not considered in this work. For
the beams considered here calculations show that the first axial natural frequency is o f
 FLEXURAL MOTION OF A ROTATING BEAM 203

the order of 100 times greater than the first bending natural frequency. Consequently, it
is felt that the bending motions will not induce significant axial motions. This is in
agreement with the treatments in references [10, 12] in which the axial motions are also
neglected. Upon letting (') denote the time derivative and noting that
~. = Oe,., 6,. = -Oex (2)
the velocity of the point may be written as
t = (a + x) 0e,. + v;,ey w0e~.
- (3)
Then for small deflections the kinetic energy of the system is given by

7"=89
IoO{~2+[w2+(x+a)2lO2+2(x+a)~.'O}dx+~l, O2+89 b "b, (4)

where I is the length of the beam, IR is the moment of inertia of the rigid shaft about
the axis of rotation, mt is a tip masst, p is the mass density per unit length and
~l = ( a + I) Oey + ~,( I, t )e~ - w( I, t ) Oex. (5)
By using Euler-Bernoulli theory, the potential energy may be written, where ( )' denotes
O/Ox, as

U =89 [EI(w") 2] d x + P(x, t) ( d s - d x ) , (6)

where EI denotes the flexural rigidity of the beam and s the arc length along the deformed
centerline of the beam. The last term in equation (6) requires some elaboration. P is an
axial force arising from centrifugal ettects and the term represents work done by this
force. In many applications, such as rotation of helicopter blades [15] the inclusion of
such a term has been found to be important and that practice and modeling is followed
here.
Considering the equilibrium of the differential element shown in Figure 2, one can see
that A P = p A x ( a + x ) O 2 so that

P(x, t) = I] p(a + x)~2 dx + (a + I)m,O 2. (7)

For a uniform beam this reduces to

P(x, t) = lpO~(i ~- x 2) + pO2a(i - x ) + ( a + l)m,O 2. (8)

O+X ~,

Z~P~ pAxto§ z

" I

Figure 2. A differential beam clement.

tThis term is included for generality. It could represent, for example, the mass of an aecelerometer, or a
payload carried by a robot arm.
204 A. Y I G I T , R. A. S C O T T A N D A. G A L I P ULSOY

From geometry ds - dx = {(dx)2 + (w' dx)2} ~/2 - d x , which, by using the binomial theorem,
can be approximated by d s - d x =89 2 dx, so that equation (6) may be written as

U=~ [ E l ( w " ) 2] dx+ 89 P(x, t)(w')2dx, (9)

where P is given by equation (8).

The equations of motion can now be derived by applying the extended Hamilton's
Principle

8 ( T - U) d t + 8wdt=0, (10)
tI tl

where 8W denotes the virtual work done by the external forces, inserting equations (4)
and (9) into equation (lO) yields

~ 8 p[~2+(w2+(x+a)2)O2+2(x+a)~'O] d x + IR~}2+m,i'a 9 i'l

Ii

-
;o [El(w")2]dx -
fo P(x,t)(w')2dx
} f:
dr+ , MSOdt=O. (11)

Integrating by parts and noting that the variations 8w and ~0 are independent, one
obtains, after some algebra and use of equation (8), the following equations of motion
and the associated boundary conditions:

p[ fi, - wO2 + (x + a) 0"] + E I w " " - [~pO2(l2 _ x 2) + pO2 a (i - x) + (a + 1)m,02] w"

-[-p~2x-p~2a]w'= 0, (12)

2 p{(w2)O'+ 2wv;'0 + ~"(a +x)} d x + ( J + IR)O'-~ I f O [ 1 2 - x 2 + 2 a l - 2 a x l ( w ' ) 2 0 "

(13)
where

J = p ( a + x) 2 dx, w(0, t) = 0, w'(0, t) = 0,

w"(i, t ) = 0 , Elw'"(l, t ) = m,[(a+l)O'+fi~(I, t ) - w ( l , t)02]. (14)

Equations (12) and (13) constitute a highly non-linear, coupled set of differential
equations. Some comments should be made on the terms w 2, w~i,, (w') 2, w'v:.'. From the
point of view of numerical simulation, they do not impose any particular computational
burden. In the results to be presented later, no perceptible difference between dropping
them or not was found. However, for large values of torque, these terms were found to
lead to instabilities. Physically, it seems clear that the system does not have instabilities
and so it is the modeling that is incorrect. Here attention is restricted to torque levels for
which the elastic motions are small, so that the terms may be dropped. The authors did
not pursue the issue of accurate modeling for large torque values, since the main purpose
of the paper is to demonstrate the inadequacies of the uncoupled model. We feel, as will
emerge, that we have achieved this for parameter ranges for which the modeling is accurate.
 FLEXURAL MOTION OF A ROTATING BEAM 205
Upon dropping terms involving w", w,i', w '2, w',,7, equation (13) becomes

fo
'P(w(a+x) d x + ( J + IR)}0"+ (a + l)m,i:t = .~//(t). (15)

The system is still non-linear due to the presence of/j2 in equation (12). It should be
noted that a considerable simplification occurs if one assumes the motion O(t) to be
known (which is not the usual situation in practice) in that then equation (12) becomes
linear. An analytic solution to equation (12) cannot be found and approximate means
must be used. Here the extended Galerkin method is to be used. One assumes that
N
w(x, t)= ~ q~,(x)q,(t), (16)
i=1

where the unknowns q~(t) are the generalized co-ordinates and the ~bi(x) are a set of
admissible functions: that is, they satisfy the geometric boundary conditions. In the sequel
they will be chosen as the mode shapes for the non-rotating cantilever beam.
The Galerkin procedure is now well-known, and since the details are tedious they will
not be reproduced here. The technique yields the following highly non-linear coupled
set of ordinary differential equations:
N
Y. {ijj(mo+m,Q, Qj)+O2qj(co-mo)+qj(ko+k*)}+[S,+m,(l+a)Q~]O'=O,
j=!

r = 1 , 2 , . . . , N, (17)

(IR+J)O'+ ~ ii~S;+,n, ~ il,Qj(i+a)+(l+a) ~ O'=M(t). (18)

j=l j i
Here:

mo = pb,qSjdx, Qj = qsj(I), (19,20)

[~p(12- x2 + 2al- 2ax)qb ~'q~i- p(x + a)b',qSi] dx, (21)

IO tttt
ko = EI~, 6j dx, k* = -Et~7~j(t), S, = o ( x + a ) 4 , , dx. (22,23,24)

3. UNCOUPLED EQUATIONS
By uncoupled equations is meant here equations in which the elastic motions do not
influence the rigid body motion. However, the effect of the rigid body motion on the
elastic motion is retained. Thus equation (17) remains unaltered and equation (18) gives,
on setting the elastic terms (q, t]) to zero,
(lt~ + J)0"+ m,(l+ t/)20 "-- 1~,/(l). (25)

Once a torque M(t) is specified, O(t) can be computed from equation (25). Then
substitution into equation (17) gives a set of linear differential equations with time varying
coefficients for the determination of the qj.
206 A. Y I G I T . R. A. S C O T T ANt) A. G A L I P ULSOY

4. LINEARIZED EQUATIONS AND NATURAL FREQUENCIES

In order to gain some insight into the behavior o f the system, a simple linear model
was constructed by retaining only one mode ( N = 1) and discarding the term involving
~J: in equations (17) and (18). On further setting the tip mass m, =0, one obtains
(Y+l~)b'+S,~=M(t), Sab'+m,,ij,+kz,q,=O. (26,27)
The natural frequencies of this system are
(o I =0, to2 = {J*ktl/(J*t~ll~-S~)} I/2, (28)
where J* = Y + IR. Note that if one were to adopt the uncoupled point of view and assume
a prescribed velocity profile 0(t), then in equation (27) S~0" would appear as a forcing
term and one would be led to a natural frequency of ~o~,= {ktx/m,I} I/2. This is the
fundamental frequency of a cantilever beam and it is less than ~o2 given by equation (28).
thus, as regards natural frequencies, rotation produces a stiflening effect. The effect is
demonstrated graphically in Figure 3, in which ~o_,/~o'~is plotted as a function of'8 = la/J.

50

-,'J 4 . 2
N
3
_~34
2
c
~2.6

1.0 I t 1 t
0 2 4 6 8 10
Inertia raho, B = IRIJ

Figure 3. Variation of the frequency ratio with/3.

5. NUMERICAL RESULTS
The torque pulse shown in Figure 4(a) was chosen as an input to the system. This
input profile is typical of many practical situations, such as moving an inertial object
from one location to another. In a decoupled mathematical model in which the effect of
elastic motions on the rigid body motion is neglected, this pulse would lead to a trapezoidal
velocity profile (see Figure 4), such as was studied by Kojima [10].
The differential equations of motion are integrated by using a backward differentiation
formula (BDF) which is an implicit muhistep integration algorithm suitable for stiff
problems. This integration package (Episode) was developed at Lawrence-Livermore
Laboratories. (See reference [16].)
Plots of the tip elastic vibration, obtained from a three-mode approximation to the
fully non-linear system are shown in Figures 5 and 6. The tip mass is assumed to be zero
for simplicity. The values of the parameters involved are as follows: El = 5.50 Nm 2,
m , = 0 , L = 0 . 5 m , a = 0 . 0 5 m , p = 0 . 0 8 5 8 k g / m , M I = I Nm, h = 0 . 0 5 s , h = 0 . 1 s , t3=
0.15 s. Both plots are for ,8 = la/J =0.5, so that the moments of inertia of the rigid and
flexible pans are comparable.
 F L E X U R A L M O T I O N OF A R O T A T I N G BEAM 207

2
I r3
,Ttme (s)

J
tl J f2

-M
(o)

I I --Time (s)
tl t2 f3
(hi

Figure 4. (a) Prescribed torque, and (b) angu'tar ~etochy profi;es.

-25f
0 O0
I
006
I
012 0t8
I
024
I
050
TLme (S)

Figure 5. Beam tip deflection, coupled equations,/3 = 0-5, M = I Nm.

The results show that the coupling can radically alter the nature of the response. Figure
5 is for the coupled case and the wave form is dominated by a fundamental mode with
a small second mode component superimposed on it (actually this is more evident in ~,,
results not shown here). It should be noted that the frequency of the dominant component
is 40.67 Hz, consistent with the prediction of linearized analysis (equation (28)) and this
is larger than that o f t h e fundamental cantilever mode o f t h e flexible beam which is 18 Hz.
Figure 6 gives the results for the uncoupled model. Now the fundamental frequency
involved is thal of the cantilever beam and clearly the response is substantially in error.
Note that a similar stiffening effect can be seen on comparing Figures 4 and 6, in reference
[14], even though not explicitly commented upon there. As the inertia of the rigid link
is increased (/3 increases) one would intuitively expect that the effects of coupling would
decrease. This is confirmed, at least as regards frequencies, by the results shown in Figures
7 and 8 which are for/3 = 4. Now both fundamental response frequencies are essentially
208 A. Y I G I T , R. A. S C O T T AND A. G A L I P ULSOY

2 5 - -

~5
o.

Eo-15

-25 - - I l l I
000 006 0 12 0 18 0 24 0 30
Time (s)

Figure 6. Beam tip deflection, uncoupled equations, p =0-5, At' = I Nm.

!'iJ A

V , , , ,

0 O0 0 06 0 12 018 0 ::)4 0 30
Time (s)

Figure 7. Beam tip deflection, coupled eqautions,/3 = 4, At' = 2 Nm.

15

~
E-9

-15
0 00
I
0 06
I
O12
I
0-18
I
024 030
Time (s)

Figure 8. Beam tip deflr uncoupled equations, fl = 4, M = 2 Nm.

 FLEXURAL MOTION OF A ROTATING BEAM 209
12 0

v
9% 64

~ 0'8

-20 I I
000 006 0 12 0 18 0 24 030
Time (s)

Figure 9 . A n g u l a r velocity o f the rigid shaft, coupled equations,/3 = 0.5, M = I Nm.

6000

4 790

3 580

2 370

1.160

-0 05C __.-.I 4-,..-,.. . . .

0 O0 0 06 0 12 018 0 24 0 30
Time (S)

Figure 10. Angular velocityof the rigid shaft, coupled equations,/3 = 4, At = 2 Nm.

the first cantilever frequency and differ only by about 16%. In passing it should be noted
that all of the above results, which involved three modes, were also obtained when using
only a one mode approximation and excellent agreement was found.
Figures 9 and 10 show the solution for 0(t), corresponding to Figures 5 and 7 (low
and high values of fl) respectively. It is evident from these results that the effect of the
elastic motion on the rigid body motion, as regards to 0 becomes more significant as the
value o f / 3 decreases, and that the uncoupled analysis introduces significant errors. It
should be noted that, though not shown here, the effect on O(t) is minor with respect to
that of 0(t).
The results presented in this section show two important effects which must be con-
sidered: (1) the increase in response frequencies, due to coupling, with decreasing/3 is
illustrated in Figures 3 and 5-8; (2) the effect of the elastic motion on the rigid body
motion becomes more significant with decreasing /3, as illustrated in Figures 5, 7, 9
and 10.

6. SUMMARY AND CONCLUSIONS

A study has been made of the flexural motion of a radially rotating flexible beam fixed
to a rigid body. The rigid body is driven by a prescribed torque profile, so that the motion
210 A. YIGIT, R. A. SCOTT AND A. GALIP ULSOY
is not known a priori. Governing equations have been derived by using the extended
Hamilton's principle. Spatial dependence has been suppressed by utilizing the extended
Galerkin approach. These equations are coupled: i.e., the elastic and rigid body motions
are interdependent.
A linearized analysis o f the coupled equations o f motion and the numerical solution
of the fully coupled non-linear equations show that the uncoupled equation can lead to
substantially incorrect results, particularly with regard to the frequencies. There are two
significant effects to be noted: (i) the response frequencies increase with decreasing/3;
(ii) the effect of the elastic motion on the rigid body motion becomes more significant
for decreasing f t .

REFERENCES
1. B. V. VISCOM1and R. J. AYRE 1971 Journal of Engineering for Industry 93B, 251-262. Nonlinear
dynamic response of elastic slider-crank mechanism.
2. J. O. SONG and E. J. HAUG 1980 Computer Methods in Applied Mechanics and Engineering
24, 359-381. Dynamic analysis of planar flexible mechanisms.
3. A. A. SHABANA and R. A. WEHAGE 1983 Journal of Structural Mechanics I1, 401-431. A
coordinate reduction technique for dynamic analysis of spatial substructures with large angular
rotations.
4. R. STAFFORD and V. GIURGIUTIU 1975 International Journal of Mechanical Science 17,
719-727. Semi-analytic methods for rotating Timoshenko beams.
5. J. S. RAO and W. CARNEGIE 1970 Aeronautical Journal of the Royal Aeronautical Society 74,
161-165. Nonlinear vibrations of rotating cantilever beams.
6. P. W. LIKINS 1972 International Journal of Solids and Structures 8, 709-731. Dynamic analysis
of a system of hinge-connected rigid beams with nonrigid appendages.
7. S. PU'I~ER and H. MANOR 1978 Journal of Sound and Vibration 56, 175-185 Natural frequencies
of radial rotating beams.
8. S. V. HOA 1979 Journal of Sound and Vibration 67, 369-381. Vibration of a rotating beam with
tip mass.
9. A. D. WRIGHT, G. E. SMITH, R. W. THRESHER and J. L. C. WANG 1982 Journal of Applied
Mechanics 49, 197-202. Vibration modes of centrifugally stiffened beams.
10. H. KOJIMA 1986 Journal of Sound and Vibration 107, 149-154. Transient vibrations of a
beam/mass system fixed to a rotating body.
11. E. M. BAKR and A. A. SHABANA 1986 Structures and Computers 23, 739-751. Geometrically
nonlinear analysis of multibody systems.
12. S. DUBOWSKYand T. N. GARDNER 1975 Journal of Engineering for Industry 97B, 652-671.
Dynamic interactions of link elasticity and clearance connections in planar mechanical systems.
13. R. R. RYAN 1985 Ph.D. Thesis, Stanford University. Flexibility modeling methods in multibody
dynamics.
14. E. R. CHRISTENSEN and S. W. LEE 1986 Computers and Structures 23, 819-829. Nonlinear
finite element modeling of the dynamics of unrestrained flexible structures.
15. L. MEIROVI"i-CH1980 Computational Methods in Structural Dynamics. Alphen aan den Rijn:
Sijthoff & Noordhoff.
16. G. D. BYRNE and A. C. H1NDMARSH 1975 A C M Transactions on Mathematical Software 1,
71-96. A polyalgorithm for the numerical solution of ordinary differential equations.