You are on page 1of 8

R. S.

Beikmann
Noise and Vibration Center, Free Vibration of Serpentine
Nortti American Operations,
General Motors Corporation,
IVIilford, Ml 48380
Belt Drive Systems
The vibration of an automotive serpentine belt drive system greatly affects the per-
N. C. Perkins ceived quality and the reliability of the system. Accessory drives with unfavorable
Associate Professor, vibration characteristics transmit excessive noise and vibration to other vehicle struc-
Mem. ASME tures, to the vehicle occupants, and may also promote the fatigue and failure of
system components. Moreover, these characteristics are a consequence of decisions
made early on in the design and arrangement of the accessory drive system. The
A. G. Ulsoy present paper focuses on fundamental modeling issues that are central to predicting
Professor, accessory drive vibration. To this end, a prototypical drive is evaluated, which is
Fellow ASME composed of a driven pulley, a driving pulley, and a dynamic tensioner. The coupled
equations of free response governing the discrete and continuous elements are pre-
Meotianical Engineering sented herein. A closed-form' solution method is used to evaluate the natural frequen-
and Applied Mechanics, cies and modeshapes. Attention focuses on a key linear mechanism that couples
The University of Michigan, tensioner arm rotation and transverse vibration of the adjacent belt spans. Modal
Ann Arbor, Ml 48109 tests on an experimental drive confirm the theoretical predictions.

Introduction serpentine drive under rapid engine acceleration. Hwang et al.


(1993) model the rotational vibrations of a serpentine drive,
Automotive accessory drive systems must reliably drive en-
and apply the results to predict the onset of belt slip. The above
gine-mounted accessories with minimal noise and vibration. V-
studies, however, all neglect transverse belt dynamics. Ulsoy et
belt drives may exhibit various noise, vibration, and fatigue
al. (1985) study the transverse belt vibrations in a two span
problems (Doyle and Hornung, 1969; Bremer, 1976). Since
subsystem coupled to a dynamic tensioner. Dynamic tensions
1979, V-belts have been largely replaced by serpentine belt
in the spans are prescribed by torque variations in an adjacent
drives with dynamic tensioners (see Fig. 1). These use a
driven accessory. The dynamic tensions parametrically excite
multirib belt with a small section height compared to a V-belt,
transverse vibrations, leading to Mathieu-type instabilities.
thereby reducing belt bending rigidity and damping. Serpentine
Mockenstrum et al. (1994) evaluate the large amplitude limit
drives generally improve reliability by lowering belt stress and
cycle oscillations that may occur near the instability regions.
heat generation, and providing near constant belt tension (Cas-
sidy et al., 1979). They generally reduce noise and vibration The above models assume that, for linear response, the rota-
levels, but considerable interest still exists in (1) fully under- tional and transverse motions are uncoupled. While this is true
standing the special characteristics of the serpentine drive, and for fixed-center systems, it remains only an approximation for
(2) producing effective tools for modelling and predicting re- accessory drives containing a dynamic tensioner. The rotation
sponse. of the tensioner arm linearly couples transverse and rotational
motions (Beikmann et al., 1992; Beikmann, 1992). The present
Two distinct types of system vibration modes occur in belt study examines this coupling mechanism and leads to new con-
drive systems with fixed-center pulleys (Houser and OUver, clusions regarding linear free vibrations.
1975): (1) rotational modes^ and (2) transverse modes. In
rotational modes, the accessories rotate about their spin axes, The natural frequencies and modeshapes of an operating ser-
and the belt spans act as axial springs. In transverse modes, the pentine belt drive system are determined herein, using analytical
belt spans vibrate transversely, similar to a taut string. Both and experimental methods. Results indicate that vibration
types of vibration degrade system performance. Rotational mo- modes include both rotational and transverse vibrations, in gen-
tions induce dynamic belt tension and the attendant problem of eral. These modes are subsequently employed in a related nu-
belt fatigue, dynamic bearing reactions, structure-borne noise, merical study of nonhnear accessory drive response (Beikmann
and bearing fatigue. Transverse vibrations also induce dynamic etal., 1996).
tensions, and may directly radiate noise. The translating belt in
a serpentine drive system is an example of an axially moving System Model
material (Wickert and IVIote, 1988). Figure 1 defines a prototypical system containing the essential
The rotational modes of a serpentine drive have been evalu- components of a serpentine drive system: a driving pulley, a
ated in recent investigations. Caspar and Hawker (1989) study driven pulley, and a dynamic tensioner. Assumptions made in
a fixed-center system, including bearing and belt damping. the model are
Hawker (1991) studies a drive system with a dynamic tensioner.
Barker et al. (1991) include a dynamic tensioner in studying a (1) Belt bending stiffness is negligible (Doyle and Hor-
nung, 1969)
(2) Belt/pulley contact points are those calculated at equi-
Contributed by the Technical Committee on Vibration and Sound for publica-
tion in the JOURNAL OF VIBRATION AND ACOUSTICS. Manuscript received June
librium
1994. Associate Technical Editor: K. W. Wang. (3) The belt stretches in a quasi-static manner
' By closed-form, it is meant only that the free-vibration problem is solved (4) Belt properties (m, EA) are uniform
without discretization of the continuous portions of the system. This avoids trunca- (5) Damping is negligible
tion/discretization errors, and allows a numerical solution arbitrarily close to the
exact solution.
(6) Belt/pulley "wedging" is negligible
' Note that in the automotive field, the rotational modes are commonly referred (7) Pulleys other than the tensioner have fixed axes
to as ' 'torsional'' modes. This is a misnomer, however, since these modes do not
induce any torsional strain energy. This paper will use the term "rotational" Hamilton's principle is used to derive the nonlinear equations
instead of ' 'torsional'' for this type of vibration. of motion, which are then linearized about an equilibrium state

406 / Vol. 118, JULY 1996 Transactions of the ASME

Copyright © 1996 by ASME


Downloaded From: http://vibrationacoustics.asmedigitalcollection.asme.org/ on 11/14/2014 Terms of Use: http://asme.org/terms
where xi = ''r^/' '«; = Jilrl (see Fig. 1), and the P^, are
dynamic tensions in each span, induced by infinitesimal pulley
and tensioner arm rotations:

^^) fX Pdi = kiiXi cos lAi + X2 -- X i )

Pdi = k2(X2 cos Va + X4 --X2)


(5)

(6)

Pdi = h{X\ - XA) (7)


Here, ki = EAI{lj), l, is the length of belt span i, and ip\ and
ijj2 are the alignment angles between the tensioner arm motion
and the adjacent belt spans at equilibrium (see Fig. 1). The
linearized equation of motion for the tensioner arm is
(-P(iWi^(/i, t) + mcwi,,(lu ()) sin (/?,
+ (P(2H'2,;t(0, t) - mcw2.,(0, 0 ) sin i/zj
- A:,(X3 cos ipi + Xi - Xi) cos ipi

- kiiXi cos l/'2 + X4 - X2) cos l/f2

- (k, + kgr)Xi = 'W3X3 (8)


where k, = k^/rj. A:, is the rotational tensioner spring stiffness,
and
(P,i sin i/zi - P,2 sin i/fj)
k = (9)
r?t

is the geometric tensioner stiffness that derives from changes


"d4 - - . "3 in tensioner arm geometry with displacement (Beikmann et al.,
1991; Beikmann, 1992). Equation (8) couples the tensioner
Fig. 1 Three pulley serpentine belt drive system
arm motion to the transverse motion of the adjacent belt spans.

(Beikmann, 1992). The linear equations of transverse motion Free Vibration Analysis
for each belt span are (Wickert and Mote, 1988) The prototypical system consists of three continuous elements
(the belt spans) and four discrete elements (the three pulleys
m(wi,„ + 2cwi^,) — P„w,v„ = 0 J = 1, 2, 3 (1) and the tensioner arm). Thus, the eigenvalue problem governing
where Wi(x, t) is the transverse deflection of span /, m is the the free response consists of three ordinary differential equa-
belt mass per unit length, c is the constant belt translation speed, tions for the belt spans and four algebraic equations for the
and the subscripts ^^ and., denote partial derivatives with respect discrete elements. This eigenvalue problem is conservative and
to X and t. The term 2cw, „ is the Coriolis acceleration compo- gyroscopic, thus admitting pure imaginary eigenvalue pairs
nent, and P„ is the span tractive tension component in span ; at ±iwr and corresponding complex conjugate eigenfunction pairs
equilibrium (P„ = Pot - mc^, where P„, is the total operating (Meirovitch, 1974). The eigenvalue problem is positive semi-
tension in span i). The centripetal acceleration component, definite, admitting a rigid body mode (a; = 0) describing pure
c'Wija, is included in the term P,iWi^. The linear equations of pulley rotation (xi = X2 = XA, XS = vvi = WJ = WJ = 0).
motion for the pulleys are Response in this rigid body mode is prescribed through the belt
translation speed, c.
Pji - Pj3 = miXi (2) Holzer's method (Meirovitch, 1986) is used to evaluate the
eigensolutions for the coupled system, using two iteration loops:
Pdi - Pd\ = miXi (3)
(1) an outer loop iterating on the natural frequency, uj, and (2)
Pea - Pdi = m^XA (4) an inner loop iterating on the dynamic tension, P ^ . The method

Nomenclature
c = steady state belt speed kr = rotational spring constant of ten- t = time
c'a = phase speed for transverse vibra- sioner arm Wi {Xi, t) = transverse displacement in
tion in span 1 k, = spring stiffness of tensioner arm: k, jth belt span
c; = effective wave speed in span 1 = kjr^ Xi = local coordinate in longitudi-
c, = transverse wave speed in span i, h = length of belt span i nal direction of ith belt span
relative to belt m = belt mass per unit length, constant 9e{^) = modal energy distribution
E = modal energy throughout system function
EA = longitudinal belt modulus m, = effective mass of rotational compo- 6i (t) = rotation of ith discrete ele-
7, = mass moment of inertia for ith dis- nent i: rrii = 7,7 rf ment in system from equilib-
crete element 7,. For tensioner arm Prf; = dynamic tension component in belt rium
11 IS J arm ' f^pulley' arm span / rj = tensioner support constant
kgr = tensioner geometric spring stiff- P,i = tractive tension component in span <//], i/^2 = alignment angles (see Fig. 1)
ness i u = frequency of oscillation
ki = longitudinal stiffness of span ;' Vi = radius of (th discrete element in Xi = displacement of perimeter of
system pulley: x, = '",^i

Journal of Vibration and Acoustics JULY 1996, Vol. 1 1 8 / 4 0 7

Downloaded From: http://vibrationacoustics.asmedigitalcollection.asme.org/ on 11/14/2014 Terms of Use: http://asme.org/terms


permits a closed-form (i.e., nondiscretized) solution of the free X cos Lot for any 02 • The phase speed c'l, and the wave speed
response problem. To begin, one seeks harmonic solutions for c'2 are defined in a manner analogous to span 1. Using (19)
the dynamic tensions and the motion of the discrete elements: and (20) in the tensioner arm Eq. (8), and solving for a2 yields
Pdi = Pdi cos ujt I = 1, 2, 3 (10) LO
02 = P„«i — COS (w/i/cj) sin i/zi -f- Prfi cos i//,
and
Xi = Xi cos Lot i = 1, 2, 3, 4 (11)
+ P,a cos ip2 + {kA - miLO^)Xi \ I Pa — sin 1//2 (21)
where the natural frequency is to be determined as follows.
The inner iteration loop begins with the current value for oj
from the outer loop, and a value for X4 is assumed. Since the Using this value for aj, one checks whether the final boundary
magnitude of the eigenfunction is arbitrary, ^4 = \luj^ is se- condition, H'2(/2, 0 = 0, is satisfied. This condition becomes
lected for convenience. The inner loop then iterates P^ to satisfy
the pulley equations of motion ( 2 ) - ( 7 ) , using Eqs. (10) and W2e„or = [flz slu {uohlc'2) + Xs sin 1//2 cos (W/2/C2)] = 0 (22)
(11). The resulting equations are solved successively as fol-
lows: The frequency of oscillation, ui, is then adjusted (and the
inner iteration loop is repeated), until Eq. (22) is satisfied. The
Equation ( 4 ) : corresponding value of w is a natural frequency. The outer loop
may also be performed in a "clockwise" manner by beginning
Pd^ = Pd2 - m^uj'^xi, (12) the solution with span 2, then checking whether the last bound-
Equation ( 7 ) : ary condition in span 1 is satisfied; i.e., Wi(0, 0 = 0.
The transverse response of the fixed-fixed span (span 3) is
Xi = X4 + Pd^lh (13) not linearly coupled to the rest of the system (see assumption
1), and is analyzed separately. The general solution (Sack,
Equation ( 2 ) :
1954) is used, with the boundary conditions W3(0, t) - Ws(h,
Pdi = Pdi - miUJ^Xi (14) t) = 0 to obtain the natural frequencies:
Equation ( 3 ) : L0„ = mc'Jih) « = 1,2, 3 . . . (23)
Xi = (Pdi - Pdi)lm2Lo'^ (15) where c5 is defined in a manner analogous to cj in span 1.
Equation ( 5 ) :
Results and Discussion
X3 = {PdJki + Xi- X2)/cos 4>i (16)
Example results are presented that highlight key features of
Equation ( 6 ) : the free response. First, characteristics of the solution method
are discussed. Second, the natural frequency spectrum is pre-
Pia = hiXi cos tli2- X\ + XA) (17) sented as a function of operating speed for two prototypical
Note that in (17), the dynamic tension in span 2 is re-evalu- system designs. Then, the effect of tensioner orientation on
ated (denoted by P2<,). Successively solving Eqs. ( 1 2 ) - ( 1 7 ) natural frequencies and modal energy distribution is discussed.
maps Prf2 onto itself. The fixed-point of this linear map, Finally, experimental and analytical results are compared for
an example system.
P2a = Pd2 (18)
Error Function Characteristics. The natural frequencies
is the solution. are determined by satisfaction of the last boundary condition,
The outer iteration loop begins with the general solution of which is cast as a frequency-dependent error function; see Eq.
Eq. (1) for span 1 (Sack, 1954): (22). Frequencies at which this error is zero are natural frequen-
X — li
cies. While the solution method is straightforward, the error
Wi(x, t) = fl] sin {ujxic'i) cos \ u>t + uj (19) function exhibits singularities which complicate the search for
roots. When employing this method in the "counter-clockwise"
direction (w2(/2, 0 is evaluated last), one calculates the coeffi-
which satisfies the boundary conditions Wi (0, f) = 0 and Wi(li, cient tti, which determines the magnitude of the transverse dis-
t) = X-i sin ip\ cos ujt provided placement in span 1. In doing so, one evaluates
X-i sin 1//1 X3 sin !//,
sin (w/j/cj) sin (w/|/cl)
Here, the phase speed is defined as which is singular whenever LO = mrc\/li, where n is an integer.
C\~ C^
This singularity is inherited by the boundary condition error
Ca function (22), and occurs at the natural frequencies of a fixed-
fixed translating belt for span 1. Further, if the belt mass is
and the effective wave speed is small compared to the discrete masses (which is usually the
case), or the tensioner arm is nearly perpendicular to span 1,
c\2 - c 2 certain system natural frequencies will be nearly equal to those
of the fixed-fixed span. Thus, some roots of the error function
are in close proximity to these singularities (at 378 Hz, for
The general solution of (1) for span 2 is example), as shown in Fig. 2.
Similarly, in a ' 'clockwise'' analysis, the natural frequencies
W2{x, t) = [02 sin (WX/C2) + Xs sin 1//2 cos (u!x/c2)] of a fixed-fixed span 2 produce singularities close to the frequen-
X cos (uj{t + xlc'b)) (20) cies of a system mode (at 249 Hz, for example), as shown in
Fig. 3. In both cases, these singularities cause two types of
which satisfies the boundary condition H'2(0, t) = X3 sin i/»2 numerical problems; (1) the singularities pass from — «> to +°°

408 / Vol. 118, JULY 1996 Transactions of the ASME

Downloaded From: http://vibrationacoustics.asmedigitalcollection.asme.org/ on 11/14/2014 Terms of Use: http://asme.org/terms


Systera 2 Pivot

Fig. 4 Prototypical serpentine drive systems 1 and 2

1 tiO.9 120.8 180.7 240.6 300.S 360.4 420.3 480.2 S40.1 «00
pects the natural frequencies of the predominantly rotational
Frequency (Hz)
modes to be weak functions of speed, and those for the predomi-
Fig. 2 Boundary condition error function for system 1, 5 RPM, CCW nantly transverse modes to decrease significantly with speed.
analysis The modes are characterized by the distribution of vibration
energy among the discrete and continuous components. One
such measure is the quantity 0^ = sin"' ((E,)/{E[, + Ec))"^),
(or vice versa) without crossing zero, but numerically appear where Eo is the modal energy of the discrete components, and
to do so (false root), and (2) the error function may pass Ec is the modal energy of the continuous components. Thus, 9^
through - 00 and + oo and return to near its original value over vanishes for pure transverse modes (ED = 0) and is 90 deg for
such a small frequency range that the root could be missed, pure rotational modes (Ec = 0). Intermediate values of 6^ (e.g.,
due to inadequate frequency resolution. Moreover, the required 20 deg-70 deg) indicate a high degree of coupling between
resolution to avoid this difficulty is problem-dependent. the discrete and continuous components of the system.
Except for unusual cases (/i = {plq)l2, P and q integers), Results for two prototypical systems are discussed presently.
the singularities for the clockwise and counter-clockwise analy- The systems differ only in the location of the tensioner arm
ses do not coincide. Thus, roots near singularities in one analysis pivots; see Fig. 4. Their properties are given in Table 1. Figure
are easily recognizable zero-crossings in the other (compare 5(a) shows the lowest six natural frequencies for system 1,
Figs. 2 and 3 at 249 and 378 Hz). Performing analyses in both omitting the (decoupled) modes of span 3. This system has a
directions ensures that all natural frequencies can be found using relatively large value of the tensioner support constant, 77 =
reasonable frequency resolution. 0.88. Values of r] typically range from 0 to 1, and -q is defined
as (Beikmann, 1992; Beikmann et al., 1991)
Natural Frequencies: Speed Dependence. Two distinct
types of vibration modes exist for serpentine drive systems, in 77 = kh/(ki, + kg + k,) (24)
the limiting case of infinite tensioner spring stiffness: ( I ) the
rotational modes, and (2) the transverse modes. Because the where k,, = (EA/L)(cos ipi + cos i/fj)^, kg = Pr(^, + ^2). C'
stiffness of the rotational modes derives from the axial stiffness = d(cos il/i)/dxi, and k, = k^/rl. The quantities k,,, kg, and kg
of the belt spans, their natural frequencies are nearly indepen- are components of the tensioner arm stiffness used to solve the
dent of speed (they are slightly affected by speed-induced ge- equilibrium problem: (1) k^, the "belt stiffness," derives from
ometry changes, Beikmann et al., 1991; Beikmann, 1992). By the belt material property, EA, (2) kg, the "geometric stiff-
contrast, the natural frequencies of the transverse modes de- ness,' ' derives from changes in alignment between the direction
crease monotonically with speed (Sack, 1954); see, for exam- of tensioner pulley center motion and the adjacent belt spans,
ple, Eq. (23). and (3) fe,, the "spring stiffness," derives from the tensioner
spring. The tractive tension in span i may then be approximated
When the tensioner stiffness is finite, the sharp distinction by
between rotational and transverse modes is lost; all modes in-
volving transverse belt response of spans 1 and 2 also include Pr + (v- i)mc^ (25)
rotations of the tensioner arm and pulleys. The vibration energy
of most modes, however, is dominated by either the energy of where P, is the system tension at rest. Thus, the tractive tensions
the discrete elements or the transverse belt response. One ex- in systems with lower values of r/ decrease more rapidly with
increasing speed. The lowest six natural frequencies and values
of dg for system 1 are shown in Table 2 ( c ) , for zero speed.
Figure 5(a) shows that as the engine speed increases (quasi-
statically) from zero, the natural frequencies of the predomi-
nantly rotational modes remain nearly constant, while those of
the predominantly transverse modes decrease with speed; refer
to Table 2(a). Similar results for system 2 (77 = 0.73) are listed
/ \ in Table 2(b) and illustrated in Fig. 5(b).
At some operating speeds, e.g., 3700 RPM in Fig. 5(b), a
transversely dominant mode frequency approaches that of a

Table 1 Component properties for the prototypical systems


: Moments of inertia:, Ji=0.00351 kg-m^, J2=J3=0.000293 kg-m^, ]^= 0.0585 kg-m^, and
^ \ m= 0.08929 Icg/m. Radii: ri=0.1016 m, 12=0.0762 m, i3=0.0508 m, and r4=0.1524 m.
1 60.9 120.8 180.7 240.6 300.S 360.4 420.3 480.2 540.1 600 Spin axis coordinates (meters): (Xi,Zi)=(0.3048,0.0), (X2,Z2)=(0.1778, -0.0127), and
Frequency (Hz) (X4,Z4)={0.0,0.0). Tensioner pivot coordinates (meters): for System 1; (X3,Z3)=
(0.127,-0.0127), for System 2; (X3,Z3)=(0.1419, -0.0486). Belt modulus: EA=88964 N.
Fig. 3 Boundary condition error function for system 1,5 RPM, CW analy- Tensioner spring constant: kr=13.S9 N-m/rad.

Journal of Vibration and Acoustics JULY 1996, Vol. 1 1 8 / 4 0 9

Downloaded From: http://vibrationacoustics.asmedigitalcollection.asme.org/ on 11/14/2014 Terms of Use: http://asme.org/terms


600 226

600

300

200 216

100

0 210
2 3 4 6 6 7 8 3300 3600 3700 3000 4100
Engine RPH
Engine RPM (Thousands)
Fig. 6(a) Veering of natural frequencies in system 2
Fig 5(a) Natural frequencies for system 1

300

u. 200

3700 4100
4 6 6 7
Engine RPM
Engine RPM (Thousands)
Fig. 6(b) Modal energy distribution functions during curve veering In
Fig. 5(b) Natural frequencies for system 2 system 2

rotationally dominant mode. Upon close examination, one sees


Fig. 6(b). Thus, the rotationally dominant mode gains trans-
that the two frequencies do not intersect, but approach closely
verse vibration energy as engine speed increases, and vice versa.
and curve away sharply; see Fig. 6(a). This behavior is termed
In the range of the curve veering, values for 9^ indicate a high
"curve veering," and indicates large and complementary
degree of coupling between the discrete and continuous compo-
changes in the associated modes (Perkins and Mote, 1986).
nents. Thus, in this speed interval, transverse belt motions may
Values of 9^ in this operating range shows that during the veer-
be readily excited by the belt stretching induced by dynamic
ing, the modal energy distributions are interchanged; refer to
moments applied at the pulleys. This linear coupling is distinct
from the nonlinear coupling considered by Mockenstrum et al.
(1994) for a parametrically excited belt.
Table 2(a) Natural frequencies and modal energy distribution for Sys-
tem 1, zero speed
Effects of Tensioner Orientation. Tensioner orientation
affects the vibration of a serpentine drive system in three ways:
Mode# NatFreq. rHz) Be fdeprees)
(1) controlling the speed-dependent span tensions through the
1 74.4 87.91 tensioner support constant, rj [Eq. (25)], (2) controlling how
2 249.0 29.96 the stiffness terms deriving from the belt modulus, EA, act on
the tensioner [terms involving fci and k2 in Eq. ( 8 ) ] , and (3)
3 265.1 59.44 controlling the degree of linear coupling between the tensioner
4 378.1 4.52 arm motion and the transverse motion of the adjacent belt spans
5 473.1 86.54 [terms involving sin tjjt and sin tpx in Eq. ( 8 ) ] .
6 506.5 3.92 The first effect is highlighted by comparing two systems
having identical component properties and different tensioner
angles (see Fig. 4). For system I, rj = 0.88, while for system
Table 2(b) Natural frequencies and modal energy distribution for Sys-
2, T} = 0.73. The corresponding natural frequencies are shown
tem 2, zero speed in Figs. 5(a) and 5(b), as functions of speed. The natural
frequencies of transversely dominant modes in system 2 (lower
Mode# Nat Freq. (Hz) Oj fdeprees) value of rj) decrease at a faster rate than those of system 1
1 70.8 (higher value of 77). This trend reflects the more rapid decrease
87.68
in tractive tension in system 2, as expected from Eq. (25).
2 215.5 84.19 The second effect, the contribution of belt stiffness to ten-
3 245.8 0.72 sioner arm stiffness, is highlighted by comparing the natural
4 375.1 5.58 frequencies of the mode dominated by tensioner arm motion.
For system 1 (tj/i = 40.1 deg, i/f2 = 46.0 deg, at rest) this
5 479.8 87.93 frequency is 265.1 Hz, at zero engine speed [see Table 2(a)].
6 491.6 0.56 For system 2, (tp, = 85.1 deg, i/f2 = 1-0 deg, at rest) this

410 / Vol. 118, JULY 1996 Transactions of the ASME

Downloaded From: http://vibrationacoustics.asmedigitalcollection.asme.org/ on 11/14/2014 Terms of Use: http://asme.org/terms


Table 3 Component properties for the test systems

Metric Units;
Rotational inertias: Ji=0.07248 kg-m^, J2=0.000293 Icg-m2, J3=0.001165 kg-m^, J4=
0.000293 kg-m2, tensioner pulley mass: 0.302 kg (baseline), 0.378 kg (modified), and
ni=0.1029 kg/m. Radii: ri=0.0889 m, r2=0.0452 m, r3=0.097 m, and r4=0.02697 m.
Spin axis coordinates (meters): (Xi,Zi)=(0.5525,0.0556), (X2,Z2)=(0.3477, 0.05715),
and (X4,Z4)=(0.0, 0.0). Tensioner pivot coordinates (meters): (X3,Z3)= (0.2508,
Fig. 8 Transverse mode of span 3 (33.0 Hz), baseline and modified
00635). Belt modulus: EA=170000 N. Tensioner spring constant: kr=54.37 N-m/rad. systems

frequency is 215.5 Hz [see Table 2(b)]. The alignment angles rotational dominant modes. The baseline system has a rotational
(ipi and i/fa) for system 2 result in smaller tensioner arm stiffness dominant mode natural frequency nearly twice that of the first
and consequently the lower natural frequency for the tensioner transverse mode of span 3. The modified system is identical to
arm dominant mode. This tensioner stiffness component due to the first, except for the addition of a 75.8 gram (2.67 ounce)
belt modulus is given by (k, cos^ ijf, + k^ cos^ ifi2). mass to the tensioner arm, at the tensioner pulley axis. This
reduces the natural frequency of the rotational dominant mode,
The third effect, the coupling of rotational and transverse
but does not alter that of the first transverse mode of span 3.
vibration, is directly responsible for the curve veering illustrated
in Figs. 5 and 6. Consider system 2 at rest, for which sin i//i = The prototypical system is excited by driving pulley 4 tangen-
0.996 and sin i/'2 = 0.017. For system 2, the terms coupling the tially at its outer edge with an electro-mechanical shaker (see
tensioner arm and transverse vibrations of span 1 are much Fig. 7 ) . At zero speed, the response of the pulleys and tensioner
larger than those for span 2 [see Eq. ( 8 ) ] . The effect is evident arm are measured using (tangentially oriented) accelerometers.
in Fig. 5{b). At 3700 RPM, the natural frequency of a tensioner These are removed when the pulleys rotate while powered by
arm dominated mode nears that of a mode dominated by trans- a frequency-modulated AC motor. The transverse response of
verse vibration of span 2, at approximately 220 Hz. This occurs the belt spans (stationary and translating) are measured using
again between the tensioner dominated mode and a span 1 noncontacting ultrasonic displacement transducers. The mea-
dominant mode at 8100 RPM, 220 Hz. The veering involving sured mode shapes are scaled so that the maximum displacement
span 2 is much sharper (greater curvature) than that for span in each mode is unity. The modeshapes of the first three modes
1. Sharper curve veering indicates weaker coupling between the are shown in Figs. 8-10, for the baseline and modified systems
participating modes (Perkins and Mote, 1986). Thus span 2 is at rest. The dashed lines show the equilibrium position of the
seen to couple with the tensioner arm significantly less than system, and the solid lines denote the displaced position for
span 1. that particular mode.
The first observation is that the transverse mode of span 3
Experimental Results. Results from the theoretical model (see Fig. 8) is excited despite the fact that pulley 4 is driven
are presently compared to those measured using the test stand solely in the tangential direction. This demonstrates a degree
described in Beikmann (1992). Attention will focus on two of linear coupling between pulley rotations and the transverse
systems (see Table 3 for physical properties) having different motion of span 3, which is omitted in the present linear theory.
The present model ignores belt bending stiffness, pulley bearing
compliance, and belt/pulley wedging, all of which could lead
Motor to the observed response. Note, however, that the coupling is
'Enclosure truly small. This fact is demonstrated by the mode shapes in
Figs. 8-10: modes involving span 3 vibration induce negligible
motion of the other components, and vice versa. The fundamen-
tal span 3 modes are thus identical for the baseline and modified
systems.
The second mode, for both systems, is dominated by the
transverse vibration of span 2. This mode has a natural fre-
quency of 51.75 Hz for the baseline system, and 51.5 Hz for
the modified system. As predicted by the linear theory herein,
there exists significant coupling between the span 2 transverse

Bedplate
Amplifier
Amplifier
Oscilloscope Gri^
Fig. 9(a) Transverse mode of span 2, baseline system (51.75 Hz)

Fourier Analyzer

Fig. 7 Experimental test setup Fig. 9(b) Transverse mode of span 2, modified system (51.5 Hz)

Journal of Vibration and Acoustics JULY 1996, Vol. 1 1 8 / 4 1 1

Downloaded From: http://vibrationacoustics.asmedigitalcollection.asme.org/ on 11/14/2014 Terms of Use: http://asme.org/terms


36 • ' • 1

32.6 -

30
ExpcrlmanUI
Fig. 10(a) Rotational mode, baseline system (62.5 Hz)
N 27.6 i/1\y;
Thioratloal
^ ;_
26

fr^^ 22.5
\ \

Fig. 10(6) Rotational mode, modified system (S8.0 Hz) \ \ ^


20

1 \ \ I
vibration and the rotation of the tensioner arm. This coupling 17.6
is apparent in Figs. 9(a) and 9(b).
The third mode, for both systems, is dominated by tensioner
arm rotation. This mode has a natural frequency of 62.5 Hz for 1 i i i 1 1
16
the baseline system, but only 58 Hz for the modified system 1 2 3 4 6 6 1
(see Fig. 10). The small mass added to the tensioner for the
Motor RPM (Thousands)
modified system has a pronounced effect on this mode.
The theoretical and experimental natural frequencies for the Fig. 11 Span 3 transverse mode frequency vs. motor speed
baseline and modified systems are shown in Tables 4(a) and
4(b), respectively. Inspection reveals excellent correlation,
with discrepancies remaining less than 3 percent. Correlation
between the experimental and theoretical mode shapes (not the theoretically predicted modal frequency remains a lower
shown) is equally good (Beikmann, 1992). bound to the experimental results.
The natural frequency of the transverse mode for span 3 was
measured as a function of belt speed. This frequency is expected Summary and Conclusions
to decrease with increasing axial belt speed, due to the reduction The linear free response of the serpentine drive involves cou-
in tractive tension; refer to Eq. (23). Experimental measurement pling between discrete and continuous system components. This
of this natural frequency was obtained by comparing the trans- paper presents an exact solution procedure to determine the
verse vibration response spectra while (1) operating with no natural frequencies and modeshapes of a prototypical three-
external excitation, vs. (2) striking the middle of span 3 with pulley system. This procedure may be readily extended to an
a round pipe. accessory drive with an arbitrary number of pulleys. The major
The natural frequencies for the fundamental mode of trans- conclusions of this study include:
verse vibration of span 3 (thefixed-fixedspan) of the baseline
system is shown in Fig. 11, for motor (pulley 1) speeds between 1. Coupling between the rotational and transverse vibrations
zero and 6000 RPM. Observe that the correlation between ex- can produce qualitatively different linear dynamic response than
perimental and theoretical results is generally good, although that predicted by ignoring this coupling.
the percentage error in the frequency prediction increases with 2. The natural frequency spectrum of the serpentine drive sys-
increasing axial speed. It is likely that this error derives from tem is speed-dependent, due to speed-dependent (1) equilibrium
neglecting belt bending stiffness, which constitutes a larger por- tension, (2) equilibrium tensioner position, and (3) Coriolis
tion of the belt restoring force as belt speed increases. Thus, and centripetal belt acceleration.
3. Tensioner arm orientation influences both rotationally and
transversely dominant modes, through (i) tension-speed depen-
Table 4(8) Natural frequencies (Hz) for the baseline system, zero speed dence, (ii) tensioner stiffness, and (iii) coupling between the
tensioner arm and transverse belt motions.
Mode# Experimental Theoretical % Error
1 33.0 32.03 -2.93 Acknowledgment
2 51.75 50.52 -2.38 The authors thank the Noise and Vibration Center of North
American Operations, General Motors Corporation, for support
3 62.5 62.22 +0.23 of this research.

Table 4(b) Natural frequencies (Hz) for the modified system, zero References
speed Barker, C. R., Oliver, L. R., and Breig, W. F., 1991, "Dynamic Analysis of
Belt Drive Tension Forces During Rapid Engine Acceleration," SAE Paper No.
Mode# Experimental Theoretical % Error 910687.
Beikmann, R. S., 1992, "Static and Dynamic Behavior of Serpentine Belt
Drive Systems: Theory and Experiment," Ph.D. Dissertation, The University of
33.0 32.03 -2.93 Michigan, Ann Arbor, MI.
Beikmann, R. S., Perkins, N. C, and Ulsoy, A. G., 1991, "Equilibrium Analysis
51.5 50.25 -2.43 of Automotive Serpentine Belt Drive Systems Under Steady Operating Condi-
tions," Proceedings of the ASME Midwestern Mechanics Conference, Rolla, MI,
58.0 58.81 +1.39 October 6-8, pp. 533-534.

412 / Vol. 118, JULY 1996 Transactions of the ASME

Downloaded From: http://vibrationacoustics.asmedigitalcollection.asme.org/ on 11/14/2014 Terms of Use: http://asme.org/terms


Beikmann, R. S., Perkins, N. C , and Ulsoy, A. G., 1992, "Free Vibration Hwang, S. J., Perkins, N, C , Ulsoy, A. G., and Meck.stroth, R., "Rotational
Analysis of Automotive Serpentine Belt Accessory Drive Systems," Proceedings Response and Slip Prediction of Serpentine Belt Drive Systems," ASME JOURNAL
of the CSME Forum, "Transport 1992+," Montreal, Canada, June 1-4. OF VIBRATION AND ACOUSTICS, Vol. 116, No. 1, pp. 7 1 - 7 8 .
Beikmann, R. S., Perkins, N. C and Ulsoy, A. G., 1996, "Nonlinear Coupled Meirovitch, L., 1974, "A New Method of Solution of the Eigenvalue Problem
Response of Serpentine Belt Drive Systems," ASME JOURNAL OF VIBRATION for Gyroscopic Systems," AlAA Journal, October, pp. 1337-1342.
AND ACOUSTICS, July. Meirovitch, L., 1986, Elements of Vibration Analysis, 2nd Edition, MacGraw
Bremer, R. C , 1976, " A Theory on the Relationship Between Drive-Train Hill, Inc., New York, NY, pp. 290-296.
Vibration and Belt-Driven Engine Cooling Fan Fatigue Failure," SAE paper No. Mockenstrum, E. M., Perkins, N. C , and Ulsoy, A. G., 1994, "Limit Cycles and
760842. Stability of a Parametrically Excited Axially Moving String," ASME JOURNAL OF
VIBRATION AND ACOUSTICS, in press.
Cassidy, R. L., Fan, S. K., MacDonald, R. S., and Samson, W. F., 1979,
Perkins, N. C , and Mote, C. D., Jr., 1986, "Comments on Curve Veering in
"Serpentine-Extended Life Accessory Drive," SAE Paper No. 790699.
Eigenvalue Problems," Journal of Sound and Vibration, Vol. 106, No. 3, pp.
Doyle, E., and Homung, K. G., 1969, "Lateral Vibration of V-Belts," ASME 451-463.
Paper No. 69-VIBR-24. Sack, R. A., 1954, "Transverse Oscillations in Travelling Strings," British
Gaspar, R. G. S., and Hawker, L. E., 1989, "Resonance Frequency Prediction Journal of Applied Physics, Vol. 5, pp. 224-226.
of Automotive Serpentine Belt Drive Systems By Computer Modeling," Proc. Ulsoy, A. G., Whitesell, J. E., and Hooven, M. D., 1985, "Design of Belt-
ASME Conf. Mechanical Vibration and Noise, DE Vol. 18-2, pp. 13-16. Tensioner Systems for Dynamic Stability," ASME JOURNAL OF VIBRATION,
Hawker, L. E., 1991, "A Vibration Analysis of Automotive Serpentine Acces- ACOUSTICS, STRESS, AND RELIABILITY IN DESIGN, Vol. 107, No. 3, July, pp. 2 8 2 -
sory Drive Systems," Ph.D. Dissertation, University of Windsor, Ontario, Canada. 290.
Houser, D. R., and Oliver, L., 1975, "Vibration of V-Belt Drives Excited by Wickert, J. A., and Mote, C. D., Jr., 1988, "Current Research on the Vibration
Lateral and Torsional Inputs," Proc. Fourth World Congress on the Theory of and Stability of Axially-Moving Materials," Shock and Vibration Digest, Vol.
Machines and Mechanisms, Vol. 4, Newcastle Upon Tyne-England, Sept. 8-13. 20, No. 5, May, pp. 3-13.

Journal of Vibration and Acoustics JULY 1996, Vol. 1 1 8 / 4 1 3

Downloaded From: http://vibrationacoustics.asmedigitalcollection.asme.org/ on 11/14/2014 Terms of Use: http://asme.org/terms

You might also like