Beruflich Dokumente
Kultur Dokumente
INA reprint
October 1999
The Dynamic Simulation of Automobile Accessory Drives
Dr.-Ing. Peter Solfrank and Dipl.-Ing. Peter Kelm
2
2 Mechanical Model Other elements representing rotational It is particularly advantageous to use a
connections between pulleys make the time series when measured values are
2.1 General model 3-dimensional in a nominal sense, available for the engine under investiga-
In this paper, it is assumed that the belt although this does not detract from the tion whose entire frequency contents are
drive is a planar system. The most impor- simplicity of the planar model: to be used. Here, a reference configu-
tant elements of the model include the rotating springs with viscous and/or ration is normally assumed, and several
following: Coulomb friction belt drive variants or tensioning system
designs are analyzed. In other applica-
belt pulleys with fixed axis and speci- one-way clutch elements tions, the aim is to evaluate a concept
fied rotation A simplification is possible as long as the without the aid of previous data, which
belt pulleys with fixed axis and one discussion is restricted to stationary often results in taking recourse to data
rotational degree of freedom as well as operating conditions, i.e. constant aver- based on similar engines. If this is the
a time-dependent or angle-dependent age speeds of the drive pulley. Calculation case, the best way to specify the cyclic
load torque can then be reduced to the deviations
irregularity of the crankshaft is to use a
from that nominal motion.
belt spans as idealized spring/damper Fourier series that is reduced to a few
elements (Kelvin-Voigt model) 2.2 Pulleys with and without harmonic components.
belt spans capable of transversal rotational degrees of freedom The mathematical representation of
vibrations pulleys having a rotational degree of free-
In practical applications, pulley rotation is
belt pulleys whose axis of rotation is on most often specified using time series or dom poses no problems either since the
a lever, which in turn rotates around a Fourier series. This is typically the case simplest form of the angular momentum
fixed point for the crankshaft since it represents the equation for these subsystems is valid for
mechanical tensioning systems: This vibration excitation and the system known torques. In addition to the
kind of system consists of a lever (with boundary simultaneously. The assump- torques transmitted from the belt to the
its own rotational degree of freedom). tion is made that a reciprocal effect of the pulley (see below), there are also load
A tension pulley is positioned on the belt drive system on this boundary torques for accessory drive belt pulleys.
lever. In addition to the torque acting condition, i.e. on the rotation of the These torques are almost exclusively
on the lever that results from belt crankshaft, is negligible. regarded as being constant with respect
forces through the tension pulley, the to time; even though this is not always
lever is also subjected to frictional the case, only insufficient information is
torque and a torque resulting from the available on their exact time behavior.
tensioning spring. Nonetheless, load torques can also be
hydraulic tensioning systems: In this specified with respect to time or position
case, the lever the tension pulley is on in the computer program developed.
is affected by a translational spring
element. The elements spring force, its
damping and the nonlinear kinematics
must all be considered. Damping,
however, affects lever rotation in one
direction only (due to a check valve).
3
Figure 1 Belt span geometry
2.3 Belts without transversal Comparing these quantities for the 2.4 Belts with transversal
vibrations nominal and current condition yields the vibrations
Regardless whether slippage between following calculation for belt elongation: Modeling the transversal vibrations of the
belt and pulley is provided for on the belt as string vibrations of the free spans
pulleys joined by a belt span, the repre- D l = lakt l0 + d1r1 (aakt a0 w1) is very similar to the representation given
sentation of belt forces for the linearly d2 r2 (aakt a0 w2) (3) in [1]. Longitudinal vibrations are disre-
elastic elongation behavior of the belt is garded here. Although the longitudinal
comparatively trivial if no transversal force of the belt in the free span is time
vibrations are allowed in modeling. Using the speeds of the belt and pulley dependent, it is regarded constant in
R R
For positions x1 and x2 of the pulley contact points the longitudinal direction. Omitting the
centers, the resulting angle a between damping terms yields a partial differential
R R R R
(4) equation that can be used to describe
the connecting line xM and the direction vk, i = vM, i + di ri vi eRie
R the forces acting on an infinitesimal belt
eRie of the belt becomes element:
calculation for the elongation speed
2 y 2y
a = arcsin
(
d1 r1 d2 r2
|xM|
R
) (1)
becomes rA
t 2
+ 2 rAv
2y
xt
4y
(7)
d
| | (F rAv2) 2 + El 4 = 0
R R
Dl = vk, 2 vk,1 (5) x x
dt
where the free belt length l is
Based on a Kelvin-Voigt model for the The symbols used in the formula signify
l = |xM| cos a
R
(2) belt with the length-specific stiffness the following:
values and damping values EA and DA rA length-specific mass
respectively, the current belt force is y amplitude of the belt element perpen-
The factor di (1) accounts for the nomi- obtained dicular to the tangent on the belt
nal rotating direction of the pulley and ri pulleys
represents the radius of the pulley EA DA d
F = max (0, F0 + Dl + Dl) (6) x longitudinal coordinate of the belt
(Figure 1). l0 l0 dt
element
v belt speed in the longitudinal direction
F belt longitudinal force
EI belt bending stiffness
4
Figure 2 Mechanical belt tensioning system
Using shape functions for the transversal Damping proportional to velocity was 2.5 Mechanical Tensioners
motion of the belt based on the method added here, and its matrix formulation As was mentioned above, a system des-
proposed by Ritz was based on the stiffness matrix. Pulley ignated as a mechanical belt tensioner
axes are assumed to be stationary. In from an application engineering stand-
y (x, t) = wT (x) q (t) (8) calculating the longitudinal force of the point represents a lever that is driven by
belt according to formula (9e), the similar- a torsion spring and damped by a
with a vector w of purely spacial shape ity to the calculation discussed in the last Coloumb frictional torque (Figure 2). The
functions and the vector q or purely time- section (formula (6)) becomes apparent: torsion spring that applies the tensioning
dependant functions, which represent The only difference is that changes in torque also introduces longitudinal force
degrees of freedom in the overall model, length and corresponding elongation in the direction of the tensioner axis. This
yields a representation transformed into speeds have been added due to the force acts on a ring surface equipped
ordinary differential equations: transversal vibrations. with a friction lining. This allows a nearly
+ (G + D)q + Kq = 0 Sine functions are used as shape func- constant frictional torque about the axis
Mq (9)
tions for the transversal with their wave- of rotation.
length adjusted in such a way that a For the representation of the overall
with whole number multiple of half of the system, this means that the number of
wavelength corresponds to the free degrees of freedom in the system can
M = w wT dx (9a) length of the belt span: change depending on whether or not the
wi (x) = sin (i x) frictional torque is sufficient to lock the
l0 lever. In the locked condition, the actual
G = 2 v w w l T dx (9b) This allows the local integral matrices to
be given. For the j th shape function, the
frictional torque compared to the maxi-
mum value achievable is reduced enough
following ordinary differential equation so that it compensates the other torques
D = d w l w l T dx (9c) results: present. Determining current condition in
the tensioning system (lever degree of
n, step 2
l0 k q k
2
+d (j) q j
qj + 4 vj freedom active or locked) as well as the
K = (F v2) w l w l T dx 2 j2 k2 2l0 current frictional torque in the locked
(9d) k = mod2 ( j ) +1
condition is comparatively trivial in such a
(10)
3 4
ll T
+ EI w w ll
dx 2 4 clear system as the one under investiga-
+ (F v2) (j) + EI (j) qj = 0
2l0 2l03 tion here.
F = F0 +
EA
l 1 12 q w w
T l lT
dx q+ l 2
(9e)
1
+ DA qT w l w l T dx q + d l
l dt 2
5
Figure 3 Hydraulic tensioning system Figure 4 Mechanical model of the hydraulic tensioning element
2.6 Hydraulic Tensioners The easiest way to represent this kind of To represent this behavior in the scope of
The hydraulic tensioner already behavior is to use a series connection of the system of differential equations to be
described (Figure 3) consists of a double the ideal damper with a spring stiffness solved, one internal degree of freedom for
lever with a rotating degree of freedom and a one-way clutch. The check valve the force element is required in addition
and a tensioning element installed be- serves to transmit forces in one direction to the position and velocity of the con-
tween a point on the lever and an inertial only (Figure 4). joined points. In contrast to the systems
point. The tensioning element collects The force applied by this model element other degrees of freedom, which corre-
forces in its longitudinal direction. There is is: spond to second order differential equa-
a tensioning force caused by a preloaded tions in the mathematical model, only a
spring, and there is a damping force that F = c (xd xc ) = dx d (11) single quantity is sufficient here since the
due to a check valve acts only in the description involves a first-order differen-
direction of the contraction of the and the change in force with respect to tial equation. This can be either a
tensioning element. time can be expressed as: displacement or the force itself.
In contrast to the theoretically resulting .
force displacement diagram for the sine- F = c (x d x c ) = c ( F x c ) (12)
shaped excitation of a unilateral damper, d
measurements deviate from the purely
speed-dependent damping effect. The
mapping approximated in this represen-
tation is a slanted semi ellipse that results
from the compressibility of the fluid in
the damper and/or the elasticity of the
damper housing or other flexible
elements in the force flow.
6
driven
driver
2.7 Slippage between Belt and 2.8 The One-way Clutch The mechanical model of the alternator
Pulley As already mentioned in the introduction, overrunning device consists of a speed-
For belt forces acting fully on the pulleys the cyclic irregularity of the crankshaft dependent damping and a parallel
in the rotational direction, a sufficient grip represents the root excitation of the arranged, nonlinear stiffness for torque
between belt and pulley would be re- accessory belt drive. The belt has the transmission, which intervenes only when
quired. In reality, however, this is not the tendency to transmit both crankshaft the speed of the driven pulley is lower
case. Instead, force transmission takes acceleration and deceleration to the than that of the driving pulley (Figure 5).
place on a more or less large portion of accessories. These alternate at a very For the nonlinear stiffness, measurements
the belt wrap. Some of these processes high frequency. The high rotating inertia suggest there is a relation between torque
are discussed at length in the literature, of the alternator in particular (increased and deflection that can be expressed as
but a great deal of research into highly by the high transmission to this acces- M = cFl wFlel . However, the parameters
dynamic operating conditions remains sory drive) leads to excessive belt forces for this must be determined experimen-
incomplete (e.g. [2]). In spite of this, the and to a greater tendency to slip in the tally for every clutch design.
resulting models are complex enough to deceleration phase of cyclic irregularity. In addition to this part of the clutch
make them unsuitable for a simple The result is an increased potential for the model, a speed-dependent damping that
dynamic simulation. part to wear. is constantly engaged is positioned in
A greatly simplified model is used for the Thus, more and more alternator pulleys order to represent the clutchs drag
task discussed in this paper by adjusting are being equipped with one-way clutch- torque in the overrunning mode. A formu-
the classic rope friction equation accord- es so that the sluggish alternator arma- lation for the torque that is proportional to
ing to Eytelwein. For the two belt forces ture can be disengaged during the decel- the relative speed usually yields satisfac-
bordering the belt wrap angle b, the eration phase. This has two advantages. tory results. In some cases though,
following auxiliary condition applies: One the one hand, the corresponding a Coloumb term was also used.
torque transmission is eliminated in the
Fmax (Fmin rAv2) eb rAv2 (13) deceleration phase, and on the other
hand, the alternators acceleration
where , is the coefficient of friction, pA requirement in the next oscillation cycle
the length-specific belt mass and v the is reduced since the deceleration of the
circumferential velocity of the belt. The crankshaft has not been. Instead, the
auxiliary condition can be implemented in alternator has only been subjected to its
such a way that an additional degree of load torque. This effect is apparently
freedom is introduced for every pulley on most obvious whenever large cyclic irreg-
which slip is to be possible. This degree ularities in the excitation are accompa-
of freedom represents the displacement nied by low alternator loads.
of belt and pulley with respect to each
other. If the auxiliary conditions are violat-
ed, there is an iterative adjustment of the
additional degrees of freedom until a
permissible condition is reached for the
system. This corresponds to the slipping
of the belt on the pulley.
7
Belt force [N]
Figure 6a Mappings for mechanical tensioning systems Figure 6b Mappings for hydraulic tensioning systems
crankshaft air conditioner compressor steering pump
idler pulley OAP alternator
water pump tensioning system
The stiffness and damping parame- 3.The third group of model parameters 5 Comparison of measure-
ters for hydraulic tensioning systems normally cannot be measured, e.g. belt ments and calculations
are also important model quantities damping parameters or the friction
that cannot be determined from values for the contact between the belt The application example used to illustrate
design data directly. For this reason, and the pulley. The reason for this diffi- the simulation program discussed here
these tensioner parameters are deter- culty is that these parameters are not is the accessory drive that is mass
mined from tests in which mappings, directly accessible from a measuring produced for a 4-cylinder diesel engine
i.e. force-displacement diagrams un- standpoint. Among other things, this (Figure 7). An air conditioner compressor,
der defined conditions (amplitude and results from the model itself. In reduc- the alternator and the water pump are
frequency of displacement excitation), ing model complexity, a small number driven from the crankshaft by means of a
are prepared (Figure 6a and 6b). The of parameters must be used to reflect single belt. The arrangement specified for
testing situation is then imitated using processes that are much more com- the accessory drive is such that an idler
a standard set of data for the simu- plicated than can be depicted for the pulley between the steering pump and
lation program. The stiffness and physical equivalent in the model. In the the alternator is necessary.
damping values for the tensioner are experiment, a generalization such as A one-way clutch is integrated in the
adjusted until the same mapping is this or a skewing of physical effects alternator pulley. It serves to increase belt
obtained for both simulation and test. cannot be understood in exactly the life and to reduce belt drive vibrations and
Unlike hydraulic systems, the deter- same way. In cases like these, the in- noise (Figure 8). The tension pulley is
mination of parameters for mechani- vestigator must rely on adjustment cal- arranged on the slack side of the drive as
cal tensioning systems is compara- culations. Starting with an application usual, and the overrunning alternator
tively simple since they can be read that matches the target system as pulley (OAP) allows a mechanical tension-
off from the mapping directly. closely as possible and for which meas- ing system to be used.
The OAP must be measured urements are available, replacement Measured values for the cyclic irregularity
separately to achieve the nonlinear parameters are adjusted for the entire of the crankshaft were available for this
stiffness function. The effort required system to reflect real-life conditions engine for some engine speeds between
for this is determined by the high more closely. This process is similar to 630 and 1700 min1. Adequate Fourier
degree of accuracy required for the that described for the hydraulic coefficients were determined from these
small twisting angles. tensioner subsystem. data to specify the motion in the
simulation.
9
measurement measurement
simulation simulation
Belt force [N]
force [N]
engine speed [min 1] engine speed [min 1]
Figure 9 Belt forces on the idler pulley Figure 10 Forces on the tensioner
5.1 Belt forces irregularity of the crankshaft) vs. engine 5.4 Slip
The level of belt forces is an important speed are not available, then the investi- Besides the natural slip component,
quantity in the design of accessory gator must rely on the interpolation of which is necessary for power transfer, no
drives. Finding the right compromise is Fourier coefficients. It has been observed significant slippage was observed in the
complicated by conflicting interests: On that a sufficient distribution of the belt drive analyzed here. This was the
the one hand, the loads on the accessory frequency content for the excitation will case for both the measurement and the
drive bearing supports and on the belt greatly simplify the recognition of belt simulation. The reason for the difference
should be reduced by lower belt forces. span vibrations. In some cases, it is not between the (low) slip values for the
At the same time, however, sufficient re- possible to recognize vibrations until this measurement and the calculation is that
serves are required to transmit the neces- distribution is made. a relative motion between belt and pulley
sary load and accelerating torque in order In the drive under investigation, belt span does not occur in the model until the belt
to avoid premature wear to the belt due vibrations between the steering pump slides through the entire belt wrap range.
to slip. and the idler pulley were observed in the In reality, slippage always occurs locally in
The maximum belt forces, particularly on experiment at approximately 800 min1. the belts input and output range. In a
the load side of the belt, greatly exceed These vibrations did not occur in the cal- series of applications it was demonstrated
the value of the static preload. Typically, culations for this speed. In order to allow that compared to other applications
preload is in the 200 to 350 N range. the simulation of the system for different any potential slippage in the drive can be
On the other hand, dynamic forces of up speeds in a range close to the given predicted (disregarding the low absolute
to 1,400 N occur on the operating drive value, the 800 min1 Fourier coefficients slip values in the simulation that result
idler pulley both in the measurements were used for those speeds as well. In from the model).
and in the simulation ( Figure 9). this way, belt vibrations in the simulation
were observed that had the same magni- 5.5 OAP
5.2 Tension-pulley forces tude as those in the test. The time series The good performance and effectiveness
An agreement similar to that found for the (Figure 12) shows highly fluctuating of the OAP of the drive under investi-
forces acting on the idler pulley can be vibration frequencies that are caused by gation are readily apparent when the an-
observed for the minimum and maximum corresponding time-variant longitudinal gular velocities of the belt pulley and the
tension-pulley forces. The maximum forces in the belt span. This effect also alternator shaft are compared. This is true
forces in the simulation, however, are evidenced experimentally is depicted for both measurement and the simula-
approx. 100 N higher than the measured correctly in the model formulation selected. tion. In the phase in which the belt
values (Figure 10).
The amplitude of the tensioning lever is
also represented correctly except for an
increase of approx. 1 mm in the lower
speed range (Figure 11).
10
OAP alternator rotor
Figure 12 Belt vibrations between steering pump and idler pulley Figure 13 Behavior of the OAP (simulation)
(simulation)
11
Sach-Nr. 005-517-192/DSP US-D 10991 Printed in Germany