Você está na página 1de 9

THE TORSIONAL STIFFNESS OF AN

EQUIVALENT TORSIONAL DYNAMIC MODEL OF


A CRANKSHAFT-CONTRIBUTION TO DEFINITION
Aleksandar Milasinovic
Professor
Department of IC Engine
University of Banjaluka
78000 Banjaluka
Bosnia and Herzegovina
Tel.: +387 65 566 328
Fax: +387 51 431 152
Email: acom@blic.net

Ivan Filipovic
Professor
Department of IC Engine
University of Sarajevo
71000 Sarajevo
Bosnia and Herzegovina
Email: ifilipovic@mef.unsa.ba

Ales Hribernik
Professor
Department of IC Engine
University of Maribor
2000 Maribor
Slovenia
Email: ales.hribernik@uni-mb.si

In order to mathematically-describe the torsional vibration


of a reciprocating engine, it is necessary to replace the actual
system consisting of the crankshaft, connection rods, and reciprocating elements by an equivalent system, usually modeled as a lumped-parameter system. A new approach for determining the torsional stiffness coefficient of an equivalent
mechanical systems crankshaft section, which can be used
to study torsional vibrations is presented. The real boundary
conditions of the crankshaft are emphasized as a problem,
which makes crankshaft-section torsional stiffness coefficient
determination even more difficult. An indirect method for
this determination is explained in this paper. By using this
method, the real boundary conditions of the crankshaft are
considered and the obtained results reliable, and confirmed
by experiment.[? ]
Nomenclature
c stiffness coefficients
C stiffness matrix
D diameter
e eccentricity
G shear modulus
J polar moment of inertia matrix

Address all correspondence related to ASME style format and figures


to this author.

k absolute damping
K damping matrix
l lenght
n number of lumped masses
q amplitude vector
t time
T torque vector
angular position vector
angular step
eccentricity ratio or relative damping
0 average (constant) angular speed
c j , s j harmonic components of the crankshafts speed
crankshafts speed
natural frequencies of the system
vector of the crankshafts torsional vibration
c installation clearance between the bearing and the main
journal
m installation clearance between the bearing and the
crankpin journal
Subscript
f flywheel
i, j indices
k crank
p pulley
Abbreviations
CAD Computer Aided Design
FEM Finit Element Method

IC Internal Combustion

a)
1 Introduction
Research into the torsional vibration of a crankshaft
have become more relevant because of the increasing load of
IC engine parts due to increasing engine torque and power.
On the one hand, intensity of torsional vibration has a crucial influence on crankshaft dimension and, on the other hand
the motion of the crankshaft, which is directly connected to
torsional vibration, is the base for combustion diagnostics
[1, 2]. This important role of IC engines torsional vibration
requires the development of both experimental methods for
measuring and mathematical models for predicting torsional
vibration. For the successful application of a mathematical
model describing the torsional vibration of a crankshaft, it is
necessary to know the parameters of the mathematical model
(moment of inertia, stiffness coefficient, dumping coefficient
and torque), and the boundary conditions. This paper contributes to a more realistic description of the torsional vibration systems parameters, especially the crankshaft torsional
stiffness coefficient.
In general, the crankshaft, as well as other parts of the engine, can be considered as structural components with distributed mass and elasticity. This leads to a system with an
infinite number of degrees of freedom, and requires the solving of partial differential equations. This approach has only
a theoretical aspect [3, 4]. The next approach is the Finite
Element Method (FEM), where the structure is divided into
a large number of connected elements. Another approach is
to divide the continuous system into a finite set of rigid bodies interconnected by springs and dampers (the lumped-mass
model). Such models can also be extracted from FEM models by reducing the models order. In this paper, a lumpedmass model [3, 58] with reasonable accurate results validated by experiments was used.
The motion of the crankshaft is determined by the action
of the external torque (gas-pressure and reciprocating inertia
torques), and by the dynamic response of the elastic structure to this excitation. Under steady-state operating conditions, the external torque and the crankshaft speed are periodic functions of the crank-angle, and may be expressed as
Fourier series. The dynamic response of the crankshaft to
the external torque may be determined by summing up the
response of each harmonic component of the torque. Even
when operating under steady-state conditions, two consecutive cycles of the same cylinder are never identical, due to
the stochastic nature of the combustion process. It is wellknown, that it is necessary to provide measurements of a
number of cycles in order to establish a mean reference cycle. The lumped mass-model of the crankshaft and the corresponding linear differential equations of motion for each
mass, simulate the actual dynamics of the crankshaft fairly
accurately.
For torsional-vibration analysis of the reciprocating machines, it is necessary to replace the real system with an
equivalent dynamic model of the crankshaft, in the most realistic way. An example of substituting an actual system, com-

b)

Fig. 1: Actual (a) and equivalent (b) torsional dynamic


system.
posed of flywheel on the right side and pulley on the left side
Fig. 1a, with the equivalent dynamic model of the crankshaft,
is shown in Fig. 1b. The real system is a four-stroke, fourcylinder turbo-charged engine. The general scheme of the
crankshafts equivalent dynamic model with n number of
lumped masses, is shown in Fig. 2. On the basis of the equivalent system, it is possible to create a mathematical model
describing torsional vibration. In order to achieve matching
between the dynamic models of the real system and equivalent system, it is necessary to precisely define the parameters
of the equivalent system [9].
The equations of the dynamic models motion as shown in

Jn
J2

J1

k1

J4

J3

Jn-1

c1

c2

c3

cn

e1
k2

e2
k3

e3
k4

en
kn-1

kn

Fig. 2: Scheme of the torsional dynamic model of the


crankshaft.

Fig. 2, can be written in matrix form as


J + K + C = T ()

(1)

2 METHODS FOR CRANK TORSIONAL STIFFNESS COEFFICIENT DETERMINATION


The most widely used approaches for determining
crank torsional stiffness coefficient are semi-empirical equations, and the Finite Element Method. The total twistangle consists of the torsional twist-angle of the main and
crankpin journals, and the bending angle of the crankshaft
cranks web. A simplified crankshafts crank model, with
cylindrically-shaped main and crankpin journals and with
the web of a parallelepiped shape, is used to derive semiempirical equations. Many authors have tried to consider a
real shape for the crank by implementing their equations for
the torsional stiffness coefficient of a simplified crankshafts
crank with some correction factors. In this paper, the authors have analyzed the crankshafts crank of a four-cylinder
diesel engine, type: TAM BF 4L 515 C. Fig. 3 shows predictions for the torsional stiffness coefficient of a simplified
single-crank of the crankshaft, determined by semi-empirical
equations [5], and by Finite Element Method using two different boundary conditions.
A solid model of a crankshafts crank has been made

x 10

Semiempirical equations
FEA method
3
2

FEA
supported

Tuplin

Kritzer

Ker Wilson

Jackson

Heldt

Carter

FEA
unsupported

Steffness coefficient Nm/rad

where is the angular position vector, J is the polar moment


of inertia matrix, K is the dumping matrix, and C is the
stiffness matrix. These matrixes are of n n order, where n
is the number of masses. The system dumping is modeled
as viscous dumping, e.g. the dumping torque is proportional
to angular velocity. On the right side of the equation Eqn. 1
is the torque vector T. The torque vector consists of torque
due to combustion in the cylinder and torque due to motion
of the masses, as well as the resistant torque.
Equation 1 is a non-linear differential equation of the second
order. The polar moment of inertia of lumped masses, and
the instantaneous angular velocity of the crankshaft are
functions of the angular position of corresponding mass.
The error made by assuming constant inertia and constant
angular velocity is moderate, and does not exceed 2 3%
[8, 10]. It has been shown that the use of a linear model
to determine crankshaft torsional-vibration is completely
acceptable, if there are no significant non-linear elements in
the system, [3, 5, 8, 11].
In order to determine natural frequencies, neglecting
dumping is allowed [3, 5]. In this case the parameters of
the system are moments of inertia and torsional stiffness
coefficient.
Using modern CAD software, it is possible to derive
the polar moment of inertia of lumped masses from the
geometric representation. Based on drawings of all those
parts of IC engines influencing equivalent dynamic model
of the crankshaft, solid models were created and moments
of inertia computed. This approach is more accurate than
the approximate methods given in [5], or even experimental
methods, and thus the error has been minimized. Determination of crankshaft torsional stiffness coefficient can be
reduced to determine the torsional stiffness coefficient of
simple cylindrical parts, and determining the torsional stiffness coefficient for a single crank of the crankshaft [3, 5, 12].
Determination of the torsional stiffness coefficient of the
crank is a demanding task, due to its complex geometry and
the difficulties in determining its boundary conditions. As
is well-known, there are three approaches to evaluating the
torsional stiffness coefficient of a crank: experimental evaluation, application of semi-empirical methods and application
of the Finite Element Method. The experimental method
is expensive and a very demanding task. The problems are
accurate measurement of the crank-twist angle and correct
boundary conditions. Therefore, the experimental method is
rarely used. The semi-empirical method is based on semiempirical equations given by many authors [3, 5]. These
equations are based on the empirical data and experimental
results, and used for stiffness coefficient approximations.
Today, the Finite Element Method is the most widely used
method for determining the torsional stiffness coefficient.
Determination of the torsional stiffness coefficient using any
of the previously-mentioned methods is a very demanding
task, because of boundary condition complexity. This
paper presents a new approach for determining the torsional
stiffness coefficient of a crank.

Fig. 3: Torsional stiffness coefficients of the crank predicted


by the traditional approach (semi-empirical equations) and
(FEM) method.

in order to compare the Finite Element Method and


semi-empirical equations. The unsupported crank of the
crankshaft was analyzed first (Fig. 4). The obtained results
agree well with those obtained by semi-empirical equations,
as can be seen from Fig. 3. However, the predicted torsional
stiffness coefficient does not match the actual one, since
the boundary conditions are unrealistic. Movement of the
right main journal axes, which is unsupported, takes place
(Fig. 4). The crank fixed at the left side and supported without any clearance at the right side was analyzed next (Fig. 5).
In this case the torsional stiffness coefficient is about 30%
higher than the stiffness coefficient of the crank with unsupported right journal, which is very close to the predic-

y
Fixed
y
B x
A

A xAA
yA
j yB B x
xB B

a)

b)

c)

Fig. 4: The unsupported crank of the crankshaft, a) sketch of the crank fixed at the left side and loaded by torsional moment
T on the right side, b) the movement of the highlighted surface in x-y plane due to action of the moment T, c) corresponding
Finite Element Method analysis.

y
Fixed

y
B x
A

A=A

xB

B x
yB j
B

Fixed

Fig. 5: The supported crank of the crankshaft without clearance, a) sketch of the crank fixed on the left side and supported
without clearance at the right side and loaded by moment T, b) the movement of the highlighted surface in x-y plane due to
action of the moment T, c) corresponding Finite Element Method analysis.
tion of semi-empirical models (maximal 8% dissipation was
found-Fig. 3). It can, therefore, be concluded that the semiempirical equations given in the literature and corrected according to actual engine data, match the boundary conditions
of the unsupported right main journal, as presented in Fig. 4.
Two cases have been studied by the presented FEM analysis
of the crank torsional stiffness coefficient:
1. the crank fixed at the left side with unsupported main
journal at the right side,
2. the crank fixed at the left side with supported main journal without clearance at the right side.
These two cases represent theoretically-extreme cases of
the crankshafts actual boundary conditions. The actual
boundary conditions of the crankshaft are between these
two extremes and they change during the working cycle. In
addition to the dimensions of the crankshaft, the working
conditions of the engine have the greatest influence on the
actual boundary conditions.

2.1 The actual boundary conditions of the crankshaft


In order to analyze the boundary conditions of the
crankshaft, it is necessary to consider:
- construction and dimensions of the crankshaft (the number
of the main journals);
- installation clearance between the journals of the
crankshaft and bearings. In the case from Fig. 6a the installation clearance of the main journal is m = 80 m,
and the installation clearance of the crankpin journal is
c = 48 m;
- load of the crankshaft journals and the load variation with
engine speed;
- properties of the engine lubrication system (oil-flow, oil
pressure and oil temperature).
Determination of the journals central position relative to the
bearings is possible if the previously-mentioned parameters
are known. Fig. 6a shows a polar diagram of the crankpin
journal load and Fig. 6b shows the trajectory of the journals
center as a function of eccentricity ratio = ec (e-bearings
center-eccentricity). The results shown in Fig. 6were obtained using computed bearing oil-pressure and its balance

with the active forces on the crankpin journal.


According to Fig. 6b movement of the crankshaft journal is
very limited, thus the FEM approach with free right-journal
(Fig. 4a) is quite unrealistic. The same can be concluded
for the FEM approach with fixed right-journal (Fig. 4b) Application of an indirect method for determining the actual
torsional stiffness coefficient of a crank is, therefore, more
promising. The actual torsional stiffness coefficient of the
crank can be calculated using measured torsional vibrations
of the crankshaft. This approach and its results are presented
next.

[kN]
120 370
365
360 100

Enkoder

Pressure in cylinder

Pulley
Cylinders

Flywheel
Cardan shaft
Hydraulics
Dynamometer

IC Engine

Engine
speed

Torque
PC

Crankshafts speed variations

Fig. 7: Scheme of experimental setup for measurement of


the engine torsional vibration.
380

360
Axex of the web
of the crank (y)

80

Axes of the web


60 of the crank (y)
40

400

20
-40 -20 0 20 40 [kN]
290 70
180 -20
480
0
a)

0,5 0,75 1

90
270 630
180

450

Table 1: Tested engine characteristics (engine type: TAM


BF 4L 515C)

540
b)

Fig. 6: The load of crankpin journal of the tested engine at


n = 2150 min 1 and oil temperature t = 88 0C a) polar load
diagram of the crankpin journal b) the journal center
trajectory as a function of eccentricity ratio = e/ c .

3 ANALYSIS OF THE EXPERIMENTAL RESULTS


A four-stroke four cylinder turbo-charged diesel engine
with air cooling (engine type: TAM BF 4l 515 C) was experimentally investigated. Those engine parameters important for torsional vibration analysis of the crankshaft were
measured. These parameters are: crankshaft instantaneous
rotational speed at the free-end of the crankshaft (pulley), incylinder combustion pressure and engine torque. All these
parameters were measured simultaneously at different engine speeds and loads. The crankshaft speed at the free-end
was measured by the optical encoder. A detailed description
of the experimental setup and analyzed results, is given in
[13]. A sketch of the experimental setup, is shown in Fig. 7,
and the characteristics of the tested engine, are presented in.
Table 1.
The polar moments of inertia of the equivalent torsional
vibration systems lumped masses (Fig. 1b) had to be known
in order to perform torsional vibration analysis of the tested
engine. The polar moments of inertia of the lumped masses
were determined (with high accuracy) using CAD software
(Solid Works), and are given in Table 2.
Measurements were performed at various engine operating

1500
1
2150 kW min
787
1
1500 N m min
125
1
145 mm mm

Bore diameter/stroke

P
n
T
n
D
s

Crank radius

r = 0.0725 m

Length of the connection rod

l = 0.237 m

Firing orde

1342

Reciprocating mass

ma = 4.6 kg

Rotating mass

mb = 2.6 kg

Engine power/engine speed


Engine torque/engine speed

=
=

Table 2: The polar moments of inertia for the equivalent


dynamic systems lumped masses (engine type: TAM BF
4L 515 C)
Moment of inertia of the pulley (J 1 )

0.096 kg m 2

Moment of inertia of the crankshafts


crank (J2 , J3 , J4 , J5 )

0.1144 kg m 2

Moment of inertia of the flywheel (J 6 )

1.77 kg m2

regimes under maximum engine load, by gradually increasing engine speed from 1000 to 2250 rpm using 50 rpm steps.
The angular speed variation of the crankshafts free-end was
calculated on the basis of the known angular step of the optical encoder, and the measured time necessary to run the
distance, i.e. the angular step, as:
=

.
t

(2)

j0t
j0 t
+ s j sin
),
= 0 + (c j cos
2
2
j=1

(3)

where 0 is the average (constant) angular speed, c j and s j


are the jth-order harmonic components of the crankshafts
speed. The angular position of the crankshafts free-end is
obtained by integrating equation Eqn. 3

= 0 t +

j=1


2c j
j0 t 2s j
j0 t

sin
cos
.
j0
2
j0
2

(4)

The second term of the Eqn. 4 consists of contributions due


to the rigid body motion and due to the torsional vibration
of the crankshaft [11, 13, 14].
Fig. 8 and Fig. 9 show, only as examples, the harmonic

Speed [rad/sec]

108

Speed [rad/sec]

200

a)

195
190
185

Amplitude [rad/sec]

The crankshaft speed variation may be expressed in Fourier


series

90

180 270 360 450 540 630 720


Crank angle [degrees]

b)

1.5
1
0.5
0

8 10 12 14 16 18 20
Harmonic order

Fig. 9: Harmonic synthesis (a) and analysis (b) of speed


variation of the crankshafts free-end at 1850 rpm.

a)

105.5
103
100.5

Amplitude [rad/sec]

98

90

180 270 360 450 540 630 720


Crank angle [degrees]

3
b)
2
1
0

8 10 12 14 16 18 20
Harmonic order

Fig. 8: Harmonic synthesis (a) and analysis (b) of speed


variation of the crankshafts free-end at 1000 rpm.

analysis and the synthesis of the measured speed variation of


the crankshafts free-end for two engine operating regimes
(n=1000 rpm and n=1850 rpm).
The main harmonic components of the gas pressure torque
act in the same direction for all cylinders and it is possible
to algebraically-sum their amplitudes. The main harmonic
of 10th order is dominant at 1850 rpm, and it excites the
torsional vibrations of the crankshaft (Fig. 9). At 1000 rpm
(Fig. 9), however, the 2nd and 4th order main harmonics are
dominant. The frequencies of these harmonics are lower
than the natural frequency of the system. These harmonics
are caused by mechanical excitation due to combustion and
inertial forces and represent the rigid body motion of the

crankshaft.
An amplitude spectrum of the crankshaft free-ends speed
variation, is shown in Fig. 10. The main harmonics 10, 12
and 14, and only partly harmonic 8, are within the engines
operating range. Other harmonics (just the most important
8.5, 9, 9.5, 10.5 and 11.5 are shown), are also present within
the engines operating range. These harmonic are caused by
crankshaft torsional vibration.
It is possible to determine the natural frequency of the
torsional vibration system, by analyzing the results shown in
Fig. 10. However, taking into account the fact that the engine
speed was varied using 50 rpm steps, and by considering
experimental error, it is more convenient to predict the
natural frequency interval rather than its exact value.
Some main characteristics of a few more important
harmonics of instantaneous angular speed variation, are
given in Table 3. Based on these results, it is possible to
state that the maximum amplitude of harmonics is within a
frequency interval between: 1922.7 and 1959 rad s 1 with
an average value of 1937 rad s 1 . The average value will be
used as a first natural frequency of the experimental engines
torsional-vibration system. The mode shapes of the second,
third and higher-mode natural frequencies are analyzed
very rarely. These higher-mode natural frequencies are very
high and, consequently, the frequency of actual excitation
torque can not be within their resonant range. The obtained
natural torsional vibration frequency may be applied in the
real torsional-vibration model of the investigated engine
and used to determine actual engine crankshaft stiffness
coefficient.

Table 3: Harmonics of the angular speed variations of the crankshafts free-end within the engine operating range

8.5

9.5

10

10.5

11.5

12

14

2200

2050

1950

1850

1750

1600

1550

1300

1.308

0.3908

0.9064

1.6250

0.6205

0.4017

0.6963

0.5184

1958

1932.1

1939.9

1937.3

1924.2

1926.8

1922.7

1959

Harmonic order
Engine speed (rpm)
Maximum of Amplitude (rad s1 )
Frequency of harmonic (rad

s1 )

tem Eqn. 6 is equal to 0; thus



det 2 J (J1 , J2 , ..., Jn ) + C (c1 , c2 , ..., cn1 ) = 0.

Amplitude [rad/sec]

10
8

1.5

(7)

8.5
1

9.5
12

0.5

0
1200

10.5
11.5

14

1400

1600
1800
2000
Engine speed [rpm]

2200

Fig. 10: The amplitude spectrum of speed variation of the


crankshafts free-end as a function of engine speed within
the engines operating range (engine type TAM BF 4L 515
C).

3.1 The indirect method for crank stiffness coefficient


determination
The stiffness coefficient of the crank may be determined
if the first natural frequency is known, which is the basic
concept of the suggested indirect method. A mathematical
model of the systems free vibration is given by equation:
J (J1 , J2 , ..., Jn ) + C (c1 , c2 , ..., cn ) = 0,

(5)

where is the vector of the crankshafts torsional vibration.


The research was performed purposely on the engine without the torsional vibration dumper. The stiffness coefficient
of the torsional vibration dumper is non-linear and could induce the wrong conclusions.
The solution for the ordinary differential equation system
Eqn. 5 is of the type = qe t , where is the natural frequencies of the system, and q is the amplitudes vector.
The characteristic equation of the relevant eigenproblem is
then

 2
J (J1 , J2 , ..., Jn ) + C (c1 , c2 , ..., cn1 ) q = 0.

(6)

Specifically, nontrivial solutions for are possible if, and only


if, the determinant of the linear homogeneous equation sys-

Eqn. 7 is known as the characteristic equation of


the system.
When all parameters of the system
(J1 , J2 , ..., Jn , c1 , c2 , ..., cn1 ) are known, then the solution of
the characteristic Eqn. 7 yields the n 1 natural frequencies
of the system ( = 0 is also a solution), which is a common
approach to this problem. In our case, however, the parameters of the system are not completely known, thus an indirect
solving method was applied [10, 15].
The first natural frequency is known from the harmonic analysis of the speed variation of the crankshafts free-end. Such
a set of parameters can now be found, so that the first natural
frequency will solve Eqn. 7.
Not all the parameters are unknown. The moments of the
lumped masses inertia were computed using CAD software,
and are presented in Table 2. Thus only n 1 unknown torsional stiffness coefficients of the crankshaft section have to
be found. If it were possible to determine all the n 1 natural frequencies, then we could obtain a system of n 1 nonlinear equations which would yield the n 1 unknown stiffness coefficients.
However, determining the natural frequencies of the
torsional-vibration systems of the second, third and higher
modes is impossible due to the following:
- the natural frequencies of the second, third, and higher
modes are very high (e.g. the second frequency of
the analyzed torsional-vibration system exceeds 4730
rad s1 ), therefore, only very high orders of excitation
torque (torque orders 20 and higher in our case) are
within the engines operational range;
- the excitation torque orders 20 and higher have a very low
amplitude, which is very difficult to obtain accurately.
The differences in the stiffness coefficients of the crank are
very low, therefore, it is possible to assume that crank stiffness coefficients are equal (c k1 = ck2 = ... = ckl = ck ) in most
cases. The stiffness coefficients of all other sections of the
crankshaft ( j), which are cylindrical, can be calculated using
the following equation
cj =

GD4
,
32l

(8)

3
2
1
0

cc2 c

c3

1
2

cc1

cc4 cc5

cc6

FEA
supported

After some rearrangements using Eqn. 8 and Eqn. 9 the characteristic Eqn. 7 becomes an n 1 order polynomial with
the ck as a variable. The solutions of this polynomial are
n 1 ck values. In the mathematical sense, each of these
values is the solution of Eqn. 7. However, by considering
their physical meaning, it is possible to determine a single
solution, which satisfies the nature of the process and, therefore, represents the stiffness coefficient of the system. It has
already been shown, that the correct stiffness coefficient lies
within a range of theoretically predicted extreme values. The
lower extreme is the stiffness coefficient of the crank with
an unsupported main journal and the upper extreme is the
stiffness coefficient of the crank with a fixed main journal.
As can be seen from Fig. 11, only one of the solutions of
Eqn. 7 is within this range. In our particular case it is, therefore, the correct stiffness coefficient. This value is about 8
% lower than the stiffness coefficient of the supported main
journal without clearance (Fig. 5), and almost 20 % higher
than the stiffness coefficient of the unsupported main journal
(see Fig. 4). The actual stiffness coefficient of the crank is
closer to the stiffness coefficient of supported main journal
without clearance, which may be expected due to the very
small clearance of the main journal. The suggestion that follows is, that the stiffness coefficient predicted by FEM analysis of the crank with the supported main journal without
clearance should be used when there is a lack of experimental data.
In order to analyze the influence of the equivalent torsional
vibration systems parameters in regard to possible error of
the first natural frequency, analysis of the influences of each
single parameter was carried out. The results of that analysis
are shown in Fig. 12 as a function of the anticipated error
within the range -10 % to +10 %, for each particular parameter of the torsional vibration system.
Prediction error of the first natural frequency does not exceed 2,5 % when the anticipated error of any of the lumped
masses is within 10 %. These errors, in particular, depend
on the equivalent torsional vibration system itself.
Anticipated error of the crank torsional stiffness coefficient
influences the prediction error of the first natural frequency
more significantly. In our particular case, 4,8 % prediction error was observed with 10 % stiffness coefficient error. Hence, it is possible to conclude that it is very important
to accurately determine all of the parameters of the torsional
vibration system, especially the torsional stiffness coefficient
of the crank.

x 10

FEA
unsupported

(9)

Steffness coefficient Nm/rad

1
1
= .
ci
j cj

Fig. 11: The torsional stiffness coefficients of crank


determined by the indirect method and FEM.

2050
First natural frequvency [rad/s]

where G is the shear modulus, D is a diameter and l is a


length. The torsional stiffness coefficient of the crankshaft
section (i) which is between two lumped masses and consists
of ( j) sub-sections with different stiffness coefficients, can
be calculated from the equation

2000

1950

1900

1850
10

0
5
Parameters error [%]

10

Fig. 12: The influence of the anticipated errors:  J f


-flywheel moment of inertia, J p-pulley moment of
inertia, J1 -moment of inertia of the crankshafts crank,
and ck -torsional stiffness coefficient of the crankshafts
crank on the predicted first natural frequency.

4 CONCLUSIONS
Based on the results of the crankshaft torsional vibration
research, the following may be concluded:
- Accurate determination of the systems parameters, in
particular the torsional stiffness coefficient of crank is
necessary, in order to perform reliable quality and quantity analysis of the crankshafts torsional vibration system.
- Accuracy of traditional empirical methods used for determining crank torsional stiffness coefficient is limited
due to the complex shape of the crank and the sophisticated boundary conditions (loads and supports of the
journals).

- It is difficult to determine actual boundary conditions


and to correctly predict crank stiffness coefficient by
FEM, therefore, the indirect method presented in this
paper is suggested for determining crank torsional stiffness coefficient.
- Stiffness coefficient predicted by FEM analysis of a
crank with supported main journal without clearance,
should be used.

References
[1] Kinkle, U., and Nielsen, L., 2005. Automotive Control
System for Engine, Driveline and Vehicle. SpringerVerlag, Berlin, Germany.
[2] Milasinovic, A., Filipovic, I., and Hribernik, A., 2007.
Mathematical model for choosing flywheel of ic engine. In DEMI 2007, Faculty of Mechanical Engineering, Banja Luka, Bosnia and Herzegovina, pp. 767
775.
[3] Genta, G., 1993. Vibrations of Structures and Machine.
Springer-Verlag, Berlin, Germany.
[4] Shabana, A., 2005. Dynamics of Multibody Systems.
Cambridge University Press, New York, USA.
[5] Hafner, K. E., and Maass, H., 1985.
Torsionsschwingungen in der Verbrennungskraftmaschine.
Springer-Verlag, Wien, Austria.
[6] Taraza, D., 2001. Quantifying relationships between
the crankshafts speed variation and the gas pressure
torque. SAE Paper 2001-01-1007.
[7] Taraza, D., Henein, N. A., and Bryzik, W., 2001. The
frequency analysis of the crankshafts speed variation:
A reliable tool for diesel engine diagnosis. ASME
Journal of Engineering for Gas Turbines and Power,
123, April, pp. 428431.
[8] Schagerberg, S., and McKelvey, T., 2003. Instantaneous crankshaft torque measurements-modeling and
validation. SAE Paper 2003-01-0713.
[9] Filipovic, I., Dolecek, V., and Bibic, D., 2005.
Modeling and the analysis of parameter in the
torsional-oscillatory system equivalent to the diesel engines in transport vehicles. Journal of Mechanical
Engineering-Strojniski vestnik, 51, December, pp. 786
797.
[10] Milasinovic, A., 2007. Mathematical modeling and
experimental investigation of nonlinear torsional vibration of crankshaft of ic engine. PhD Thesis, University
of Banja Luka, Bosnia and Herzegovina, April.
[11] Taraza, D., Henein, N. A., and Bryzik, W., 1998.
Determination of the gas-pressure torque of a multicylinder engine from measurements of the crankshafts
speed variation. SAE Paper 980164.
[12] Filipovic, I., 2007. IC Engine Dynamic and vibration.
Faculty of Mechanical Engineering, Sarajevo, Bosnia
and Herzegovina.
[13] Bombek, G., Milainovic, A., Filipovic, I., and
Hribernik, A., 2005. Determination of torsional vibrations of a diesel engine crankshaft. In Proceedings

of the Inovative Automotive Technology, University of


Ljubljana, Slovenia, pp. 236242.
[14] Filipovic, I., and Jankov, R., 2001. Modern approach
to measurement and analysis of the ic engines parameters. In International scientific symposium Retracing the Puch Track, University of Maribor, Slovenia,
pp. 126132.
[15] Serban, R., and Freeman, J. S., 2001. Identification
and identifiability of unknown parameters in multibody
dynamic systems. Multibody System Dynamics Journal, 5, pp. 335350.

Você também pode gostar