ANALELE UNIVERSITII

EFTIMIE MURGU REIA

ANUL XXI, NR. 1, 2014, ISSN 1453 - 7397

Zoltan-Iosif Korka, Ion-Cornel Mituletu

A Review of Dynamic Models Used in Simulation of

Gear Transmissions

The investigation of relevant scientific literature regarding gear model-

ing enabled us to discover a significant number of papers dating back
several decades and continuing to the present. The purpose of the dy-
namic models was quite diverse, but all modeling efforts share the goal
of replicating the complex physics of power transmission through gear
interaction. This paper investigates the relevant aspects regarding the
dynamic modeling of gear transmissions, starting with the simplest
model (1DOF), then developing it into a model with three degrees of
freedom (3DOF) and finishing with six degrees of freedom model
(6DOF).

Keywords: Dynamic model, simulation, gear transmission

1. Introduction

Because the gears are critical components of any rotating machine, they have
received a considerable amount of attention regarding their dynamic modeling,
being published a significant number of papers concerning this problem [1], [2].
The objectives of dynamic modeling of gear transmissions varied past five dec-
ades, from vibration controlling and noise analysis, to the study of transmission
error and stability analyses [3], [4]. The final scopes of dynamic modeling of gears
could be summarized as follows:
- Analysis of contact and bending stress;
- Reduction of superficial wear as for example pitting;
- Study of transmission efficiency;
- Study of noise radiation;
- Influence on other parts of the transmission, particularly bearings;
- Natural frequencies of the system;
- Studies regarding the vibratory motion of the system;
- Studies of reliability and life cycle.

165
 2. Dynamic Model with One Degree of Freedom (1DOF)

Figure 1 show a typical dynamic model with one degree of freedom (1DOF)
used for the mesh investigation of a gear pair system. The gear transmission is
modeled as a pair of discs, connected along the mesh line by a spring and a
damper.
The model takes into account influences of the static transmission error which
is simulated by a displacement excitation e(t) at the mesh. This transmissions error
arises from several sources, such as tooth deflection under load, non-uniform tooth
spacing, tooth profile errors caused by machining errors as well as pitting, scuffing
of teeth flanks. The mesh stiffness cz(t) is expressed as a time-varying function.
The gear pair is assumed to operate under high torque condition with zero back-
lash. Effects of friction forces at the meshing interface are neglected on the basis
that in particular, the coefficient of friction is low (approx. 6%, according to [5]).
Furthermore, the viscous damping coefficient of the gear mesh dz is assumed to be
constant.

Figure 1. 1DOF dynamic model for a gear pair system

The differential equations of motion for this system can be expressed in the
form:
J 1&&1 + rb1c z (t )[rb11 + rb 2 2 + e(t )] + rb1d z [rb1&1 + rb 2& 2 + e&(t )] = M 1 (t ) , (1)
J 2&&2 + rb 2 c z (t )[rb11 + rb 2 2 + e(t )] + rb 2 d z [rb1&1 + rb 2& 2 + e&(t )] = M 2 (t ) , (2)
where i , & i , &&i (i= 1, 2 ) are rotation angle, angular velocity, angular

166
acceleration of the input pinion and the output wheel respectively. J1 and J2 are the
mass moments of inertia of the gears. M1(t) and M2(t) denote the external torques
load applied on the system. rb1 and rb2 represent the base radii of the gears.
By introducing the composite coordinate
q = rb11 + rb 2 2 , (3)
equations. (1), (2) yield a single differential equation in the following form:
mred q&& + c z (t )q + d z q& = F (t ) c z (t )e(t ) d z e&(t ) , (4)

J1 J 2 M 1 (t )rb1 M 2 (t )rb 2
where: m red = ,and F (t ) = m + . (5)
J 1 rb22 + J 2 rb21
red
J 1 J 2
For a specific gear-pair, the mesh stiffness cz(t) can be approximately repre-
sented by a truncated Fourier series:
K
cz (t ) = c0 + c k cos( kz t + k ) , (6)
k =1
where z is the gear meshing angular frequency and K is the number of terms of
the series.
Generally, the components of the meshing error are not identical to each gear
tooth and consequentially, they will produce excitation movements periodical to
the rotation speed of the wheel (repeated every time the respective tooth is in
contact). Therefore, the excitation function e(t) can be represented by a Fourier
series with the mains frequency corresponding to the rotation speed of the wheel.
If it is considered that the errors are located only at teeth of the pinion, e(t) can be
written as:
I
e( t ) = e i cos( i1t + i ) , (7)
i =1
where 1 is the angular speed (rotation frequency) of the pinion.
When it is assumed that:
&1 = 1 = const , & 2 = 2 = const , dz= 0, cz(t)= c0, the dynamic transmission er-
ror of the gear pair q is equal to the static deformation of the teeth under the con-
stant load qo.
Therefore:
q = r b 1 1 + r b 2 2 = q 0 , (8)
So that, Eq. (4) becomes:
m red q&& + c z ( t ) q + d z q& f ( t ) = 0 . (9)
Based on the above, the vibration of a gear pair can be written as a differen-
tial equation of the form:
m red q&& + c z ( t ) q + d z q& f ( t ) = 0 , (10)

167
where:
f ( t ) = c 0 q 0 [ c z ( t ) c 0 ] e ( t ) d z e&( t ) . (11)

Taking into account four dominant coefficients c0, c1, c2, c3 in the Fourier se-
ries of the mesh stiffness, Eq. (6) can be written as:
3 3
cz (t ) = c0 + c k cos( kz t + k ) = c 0 + ( c k cos kz t + s k sin kz t ) , (12)
k =1 k =1

where z = z11 .
If the excitation function e(t) is expressed by its first two terms of the Fourier
series, we have:
2
e( t ) = e k cos( k1t + k ) . (13)
k =1

Substituting the expressions (12) and (13) in equation (11), we obtain:

3
{ckcos [( kz1 1)1t 1] +
e1
f ( t ) = c0q0 + dze11 sin(1t + 1) + 2dze21 sin(21t + 2 )
2
k =1
+ sk sin[( kz1 1)1t 1] + +
ck cos[( kz1 + 1)1t + 1] + sk sin[( kz1 + 1)1t + 1]} , (14)
3
{ckcos [( kz1 2)1t 2] +sk sin[( kz1 2)1t 2] + ck cos[( kz1 + 2)1t + 2] +
e2

2
k =1
+ sk sin[( kz1 + 2)1t + 2 ]}

Based on the analytical form of the functions cz(t), respective f(t) and using
the harmonic balance method, the solution of the differential equation (10) can be
approximated by the expression:

3 3
q( t ) = a0 + ( ak cosk1t + bk sink1t ) + [akz12 cos(kz1 2)1t + bkz12 sin(kz1 2)1t +
k =1 k =1
+ akz11 cos(kz1 1)1t + bkz11 sin(kz1 1)1t + akz1 cos(kz1)1t + bkz1 sin(kz1)1t + . (15)
+ akz1+1 cos(kz1 + 1)1t + bkz1+1 sin(kz1 + 1)1t + akz1+2 cos(kz1 + 2)1t + bkz1+2 sin(kz1 + 2)1t ]

3. Dynamic Model with Three Degrees of Freedom (3DOF)

When the stiffness respective the elasticity of the shafts and the bearings
cannot be neglected, a dynamic model with tree degrees of freedom has to be
considered.

168
 Figure 2. Dynamic model with three degrees of freedom (3DOF)

Such a model is shown in figure 2. Same as at the 1 DOF model, the gear
mesh is modelled as a pair of discs, connected along the mesh line by a spring and
a damper. In addition to the 1 DOF model, the shafts and the bearings are consid-
ered elastic, each of the discs being supported by a spring and a damper, having
elastic constants c1 and c2, respective the viscous damping coefficients d1 and
d2.Forthermore, the backlash between the teeth of the pinion and the gear is not-
ed with 2b. The gear mesh stiffness cz(t) and the static transmission error e(t) are
considered time varying, while the viscous damping is noted with dz.
Dynamic transmission error is defined:
y d ( t ) = y 1 + r b 1 1 ( t ) y 2 r b 2 2 ( t ) . (16)
The difference between the dynamic transmission error yd(t)and the static
transmission error e(t) is given by the relation:
y ( t ) = y 1 + rb 11( t ) y 2 rb 22 ( t ) e ( t ) . (17)
The meshing force can be written as:
F y = c z ( t ) f ( y ) + d z y& ( t ) , (18)
where f(y) is a nonlinear function used for the description of the gear pair with
backlash.
Assuming an equal repartition of the gap between teeth, f(y) can be written
as:
y b y > b ,

f(y) = 0 y b, (19)
y + b y > b .

169
 The differential equations of motion can be written as follows:
m 1 y&&1 + d 1 y& 1 + d z y& +c 1y 1 + c z ( t ) f ( y ) = 0

m 2 y&&2 + d 2 y& 2 d z y& +c 2y 2 c z ( t ) f ( y ) = 0 , (20)
m
red y m red y 1+m red y 2 + d z y + c z ( t ) f ( y ) = F
&& && && &
where:
J1 J2
m red =
J 1r b 2 + J 2 r b21
2
. (21)
M 1( t ) rb 1 M 2 ( t ) rb 2
F ( t ) = m red J1
+
J2
m red e&&( t )


The system of equations (20) can be written in matrix form as follows:
m1 0 0 y&&1 d1 0 d z y&1 c 1 0 c z ( t ) y 1 0

0 m2 0 y&&2 + 0 d 2 d z y& 2 + 0 c 2 c z ( t ) y 2 = 0 . (22)
mred mred mred y&& 0 0 d z y& 0 0 c z ( t ) f ( y ) F
According to [2], a dimensionless form of the Eq. (22) can be obtained, by
assuming following simplifications. Let:
e ( t ) = e a cos( t + a ) + e b cos( t + b ) , (23)
ca
c z ( t ) = c m + c a cos( t + ) , c a = , (24)
cm
where is the main excitation frequency of the transmission error respective of
the stiffness of the gear transmission.
y (t ) ci
f ( yi ) = yi ( t ) = i ; i = i= 1, 2 (25)
b mi
y(t ) cm
y(t ) = ; n = ; t = n t (26)
b m red
d1 dz d2 dz dz
11 = ; 13 = ; 22 = ; 23 = ; 33 = ; (27)
2m1n 2m 1n 2m 2 n 2m 2n 2mredn

12 2 m m
c 11 = ; c 22 = 2 ; c 13 = red ; c 23 = red ; c 33 = 1 + k a cos( t + ) ; (28)
n
2
n2 m1 m2

with the excitation frequency vector = (29)
n
Fm e eb
Fm = ; Fa = a ( ) 2 ; Fb = ( ) 2 . (30)
m red b n2 b b
So that:

170
 F = F m + F a cos( t + a ) + F b cos( t + b ) . (31)
Eq. (22) can thus be written in dimensionless form:
1 0 0 y&&1( t ) 11 0 13 y& 1( t ) c 11 0 c 13c 33 f ( y 1 ) 0
&& &
0 1 0 y 2( t ) + 2 0 22 23 y 2( t ) + 0 c 22 c 23c 33 f ( y 2 ) = 0 , (32)
1 1 1 y&&( t ) 0 0 33 y&( t ) 0 0 c 33 f ( y ) F

with:
y 1 y > 1,
f(y)
f(y) = = 0 1 y 1, (33)
b y + 1 y > 1.

For simplifying reasons, the - sign above the variables in the Eq. (32) and
(33) will be neglected and, therefore, the equation (32) becomes:
1 0 0 y&&1( t ) 11 0 13 y 1( t ) c 11 0 c 13c 33 f ( y 1 ) 0
&&
0 1 0 y 2( t ) + 2 0 22 23 y 2( t ) + 0 c 22 c 23c 33 f ( y 2 ) = 0 (34)
1 1 1 y&& ( t ) 0 0 33 y ( t ) 0 0 c 33 f ( y ) F

As a first step, the three second order differential equation are converted in
six first order differential equations by using the Runge-Kutta method.
For this purpose, q variable is introduced as follows:
{ {
q = q 1 , q 2 , q 3 , q 4 , q 5 , q 6 }T = y 1 , y& 1 , y 2 , y& 2 , y , y& }T (35)
Thus, Eq. (34) can be written in matrix form as follows:
q& = H q + c 33 f ( q 5 ) A + B (36)
whereA is the matrix of the nonlinear coefficients:
A = {0 , c 13 , 0 , c 23 , 0 , c 13 c 23 c 33 }T (37)
B is the vector of load (forces):
B = {0 0 0 0 0 F }T (38)
and H is the matrix of the linear coefficients:
0 1 0 0 0 0

c 11 211 0 0 0 13
0 0 0 1 0 0
H = (39)
0 0 c 22 2 22 0 23
0 0 0 0 0 1

c 11 211 c 22 2 22 0 13 23 233

Eq. (36) can be solved by using the MATLAB software, with which can be al-
so performed a simulation by using the facilities offered by the SIMULINK tool.

171
 4. Dynamic Model with Six Degrees of Freedom (6DOF)

When the influence of the driving, respective driven machine cannot be ne-
glected, it has to be choosing a mathematical model with six degrees of freedom.
Such a model is shown in figure 3.

Symbols:
J- moment of inertia;
- rotation angle;
M (t)- torque moment;
D- damping constant of connecting shaft;
C- elastic constant of connecting shaft;
m- mass
z- teeth number;
rb- base radius;
d- damping constant of the bearing;
c- elastic constant of the bearing;
e(t)- displacement excitation;
cz(t)- gear mesh stiffness;
dz- damping coefficient of the gear mesh.

Figure 3. Dynamic model with six degrees of freedom (6DOF)

Generalizing those presented in the chapters 2 and 3 of this paper, as well as

on the theoretical considerations of the forced damped vibration, it can be con-
cluded that the equation of motion of a dynamic system, which includes a gear
transmission can be written in following form:
{} {} {} { }
[M ] q&& + [D ] q& + [C ] q = F (40)
where:
[M]- Matrix of masses; [D]- Matrix of dampings; [C]- Matrix of stiffness;
q - vector of deplacements; F -vector of force.
For the 6 DOF dynamic model, the elements of Eq. (40) can be written, ac-
cording to [5], as follows:

172
 4 J Mo
0 0 0 0 0
2
rb 1
4 J 1
0 0 0 0 0
r b21
4 J
0 0 2 0 0 0 , (41)
r b21
[M ] = 1
4 J Ma
m red 0 0 0 0 0
r b21

0 0 0 0 m 0
1


0 0 0 0 0 m 2

4 D 4 D Mo
Mo
0 0 0 0
r2 rb21
b1
4 D d d r dz dz
Mo
+ z z b2 0
r2 4 4 rb1 2 2
b1
4 D d r d z rb 2 ; (42)
Ma
+ z b2 4 D Ma d r
z b2
r2 4 rb1 rb21 2 rb1 2 rb1
[D] = 1

b1

c z m red 4 D Ma
0 0
rb21


symmetric d1 + d z dz


d 2 + d z


4 C 4 C Mo
Mo
0 0 0 0
r2 rb21
b 1
4 C c c z rb 2 cz
Mo
+ z 0
cz

r2 4 4 r b1 2 2
b1 ; (43)
4C c r 4 C Ma c z rb 2 c z rb 2
Ma
+ z b2
r2 4 r b1 rb21 2 r b1 2 r b1

[C ] = 1 b1

cz 4 C Ma
0 0
rb21


symmetric c1 + c z cz


c2 + c z

q = {Mo 1 2 Ma y 1 y 2 }T . (44)

173
 5. Conclusions

Mathematical models attempt to include the essential parameters of natural

phenomena in systems of equations or in systems of differential equations in order
to predict the evolution of the observed system.
The basic principle in formulating a scientific model (modeling) is to reduce
complexity, by trying to make the truth describable and understandable through
simplicity.
The present paper presented relevant aspects regarding the dynamic model-
ing of gear transmissions. Starting with the simplest model (1DOF), developing it
by considering factors as bearing, shaft, driving and driven machine, until the
mathematical model with six degrees of freedom (6 DOF) was reached.

Acknowledgment

The work has been funded by the Sectoral Operational Programme Human
Resources Development 2007-2013 of the Ministry of European Funds through the
Financial Agreement POSDRU/159/1.5/S/132395.

References

[1] Buechner S., Zschaeck S., Amthor A., Ament C., Dynamic model and
identification of a spur gear system, Proceeding of 15th IASTED Interna-
tional Conference on Control and Applications (CA 2013), pp. 43- 48,
2013.
[2] Diehl J.E., Tang J., DeSmidt H., Gear fault modeling and vibration re-
sponse analysis, ASME 2012 5th Annual Dynamic Systems and Control
Conference, Florida, 2012.
[3] Gillich G.R., Machine dynamics- Modelling of technical systems, ISBN
973-8466-51-2, Editura AGIR, Bucureti, 2003.
[4] Korka Z., An overview of mathematical models used in gear dynamics,
Romanian Journal of Acoustics and Vibration, vol. IV, No. 1, pp. 43- 50,
2007.
[5] Korka Z., Research on vibration reduction in operation of cylindrical
gearboxes, Ph.D. Thesis, Eftimie Murgu University, Reia, 2009.

Addresses:

Lect. PhD. Eng. Zoltan Iosif Korka, Eftimie Murgu University of Resita,
Piata Traian Vuia, nr. 1-4, 320085, Resita, z.korka@uem.ro
PhD. Eng. Ion-Cornel Mituletu, Eftimie Murgu University of Resita, Pi-
ata Traian Vuia, nr. 1-4, 320085, Resita, mituic@yahoo.com

174