Mechanism and Machine Theory 45 (2010) 15831598

Contents lists available at ScienceDirect

Mechanism and Machine Theory

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / m e c h m t

Torsional vibration model based optimization of gearbox geometric design

parameters to reduce rattle noise in an automotive transmission
Mehmet Bozca
Yildiz Technical University, Mechanical Engineering Faculty, Machine Design Division, 34349, Yildiz, Istanbul, Turkey

a r t i c l e

i n f o

Article history:
Received 23 November 2009
Received in revised form 8 June 2010
Accepted 22 June 2010
Available online 24 July 2010
Keywords:
Optimization
Gearbox
Rattle noise
Torsional vibration
Automotive transmission

a b s t r a c t
The optimization of gearbox geometric design parameters to reduce rattle noise in an automotive
transmission based on a torsional vibration model approach is studied. Rattle noise is calculated
and simulated based on the design parameters of a 5-speed gearbox, and all pinion gears and
wheel gears are helical. The effect of the design parameters on rattle noise is analyzed. The
observed rattle noise proles are obtained depending on the design parameters. During the
optimization, a four-degree-of-freedom torsional vibration model of the pinion gearwheel gear
system is obtained and the minimum singular value of the transfer matrix is considered as the
objective functions and design parameters are optimized under several constraints that include
bending stress, contact stress and a constant distance between gear centers. Therefore, by
optimizing the geometric parameters of the gearbox such as, the module, number of teeth, axial
clearance, and backlash, it is possible to obtain a light-weight-gearbox structure and minimize the
rattling noise. It is concluded that the optimized geometric design parameters lower the rattle
noise by 10% compared to the calculated rattle noise for sample gearbox. All optimized geometric
design parameters also satisfy all constraints.
2010 Elsevier Ltd. All rights reserved.

1. Introduction
The optimization of gearbox geometric design parameters to reduce rattle noise in automotive transmissions based on a
torsional vibration model approach is studied.
Gear motion causes rattling and clattering noise, and noise level is considered to be a comfort factor in the automotive
industry. Therefore, reducing rattling and clattering noise in the gearbox is important in the automotive transmission for a
comfortable car design.
Gears are widely used in automotive transmissions to transmit mechanical power from one shaft to another. The purpose of the
gears is to couple two shafts together such that the rotation of the output, or driven, shaft is a function of the rotation of the input,
or driving, shaft [4].
Rattling and clattering noise in automotive transmissions are caused by torsional vibration that is transmitted from the
internal combustion engine to the transmission input shaft. This noise is known as rattling when the transmission is in
neutral, and as clattering when the gear is engaged under power or in overrun [914].
Rattling and clattering are caused by torsional vibration of loose parts, i.e. parts, such as idler gears, synchronizer rings
and sliding sleeves, which are not under load and therefore can move within their functional clearances [9,14].
Gear rattling noise is one of the major problems facing the industry, and the car industry in particular, because cars
spend so much time idling under no load or very light loads [1].

Tel.: +90 212 383 27 93.

E-mail address: mbozca@yildiz.edu.tr.
0094-114X/$ see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mechmachtheory.2010.06.014

1584

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

The parameters that are responsible for rattle and clatter are classied as geometric parameters and operational
parameter [914]. The geometrical parameters include the module m, number of teeth z, helix angle , axial clearance s
1, and excitation frequency
and backlash sv as shown in Fig. 3. The operational parameters include the angular acceleration
an [914].
Transmission uid as an engineering design parameter also has a considerable inuence on rattling and clattering noises. The
important factors include the type of oil, the additives used and the viscosity and the level of oil in the transmission which together
act on a gear pair as drag torque, resulting in a reduction in rattling and clattering noises, especially at low speeds and when cold
[914].
The analysis suggests that in order to reduce the gear rattle noise, rather than to increase the oil level in a gearbox and thus
decreasing the mechanical efciency, it would be opportune to guarantee the oil presence between the meshing teeth by means of
a suitable lubrication device feeding the lubricant only in the meshing zone [15].
Sliding friction between meshing teeth is one of the primary excitations for noise and vibration in geared systems. Among the
different kinds of nonlinearities in gear system, such as clearance, spatial variations and sliding friction, the effect of friction is the least
understood. Certain unique characteristics of gear tooth sliding make it a potentially dominant factor. For instance, due to the reversal
in direction during meshing action, friction is associated with a large oscillatory component, which causes a higher bandwidth in the
system response. Furthermore, it becomes more signicant at high values of torque and lower speeds, due to the tribological
characteristics as well as due to higher force transmissibility in the sliding direction [17].
Optimization of the gear's macro-geometry i.e. the use of high contact ratio gears that has lead to minor noise emissions with
higher transmitted power levels. Optimization of the gear's micro-geometry i.e. trying to balance load-induced teeth deections
with prole corrections that has generally lead to less noisy transmission effects. This is not a suitable solution for an overall
working range; therefore prole corrections must be determined statistically to take into account manufacturing deviations which
will overlap their effect [16].
1.1. Gearbox mechanism
The gearbox mechanism includes pinion gears, wheel gears, an input shaft, an output shaft, a lay shaft, a bearing support, and
synchronizers, as shown in Fig. 1.
1.1.1. Pinion gears and gear wheels
All pinion gears and wheel gears are helical, and all gears are made of 16MnCr5.
1.1.2. Input shaft
The constant pinion gear and rear wheel gear are engaged on the input shaft. The rear wheel gear is the idler gear and runs in
the needle bearing on the input shaft. The synchronizer of the rear wheel gear is connected to the input shaft.
1.1.3. Output shaft
The 1st, 2nd and 3rd wheel gears and 4th pinion gear are engaged on the output shaft. The 1st, 2nd and 3rd wheel gears are
idler gears and run in the needle bearings on the output shaft. The synchronizer of the 1st and 2nd wheel gears is connected to the
output shaft.
1.1.4. Lay shaft
The 1st, 2nd, and 3rd pinion gears, the rear pinion gear and the 4th wheel gear, and the constant wheel gears are engaged on
the lay shaft. The 4th wheel gear is the idler gear and runs in the needle bearing on the lay shaft. The synchronizer of the 3rd and
4th wheel gears is connected to the lay shaft.

Fig. 1. 5-Speed gearbox for automotive transmission.

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

1585

1.1.5. Synchronizer
All synchronizers run in the needle bearings to maximize smoothness.
2. Calculation of rattle noise
Rattle noise is calculated and simulated based on the gearbox design parameters for a 5-speed gearbox for an automotive
transmission, which is shown in Figs. 1 and 2.
The rattle noise level of a complete automotive transmission is calculated as follows [913]:
n

LpComp = 10log

i=1


0;1L
10 p;i :

The rattle noise Lp, is calculated as follows by correlating the computed noise value and the measured noise level [913]:



0;1L
Lp = 10log kIm + 10 basic

where k is the calibration factor [] and Im is the average impact intensity [N]. The average impact intensity Im, is written as follows
[913]:
1 rb1 CIm
Im = m2

1 is the angular acceleration [rad/s2], rb1 is the pitch circle radius [mm] and CIm is the related average
where m2 is a loose part [kg],
impact intensity []. The average impact intensity CIm, is written as follows [913]:
CIm

p
= Csv 1; 462

0; 714Cfa Csa
0; 016Cfa + 0; 12Csv

!
4

where Csv is the non-dimensional circumferential backlash []. The non-dimensional circumferential backlash Csv [] is dened as
follows [913]:
Csv =

sv 2an
1
rb1

where sv is backlash [mm], an is the excitation frequency [rad/s] and Csa is the non-dimensional axial clearance []. The nondimensional axial clearance Csa [] is dened as follows [913]:
Csa =

sa 2an tan
1
rb1

where sa is the axial clearance [mm], is the helix angle [], Cfa is the related axial friction force [] and Lbasic is the basic noise level
[dB].

Fig. 2. View of 5-speed gearbox for automotive transmission.

1586

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

Fig. 3. Design parameters for a gearbox.

3. Calculating the load capacity of helical gears

Determining the service life and/or determining the strength of gears is crucial for gear manufacturers. Gear strength is dened
by the bending and contact strengths [22].
3.1. Tooth bending stress
The tooth bending stress is calculated as follows, Fig. 4 [2,3,5,7]. According to the ISO 6336, shear stresses due to lateral forces
were not taken into account when determining the loading capacity of the gear [3,22,23]. A tooth-root bending fatigue fracture
usually starts at the 30 tangent of the root [3,8,22,24].
The real tooth-root stress F is calculated as follows [2,3,5,7]:
F =

Ft
Y YY Y K K K K
bmn F S A V F F

where Ft is the nominal tangential load [N], b is the face width [mm], mn is the normal module [mm], YF is the form factor [], YS is
the stress correction factor [], Y is the contact ratio factor [], KA is the application factor [], KV is the internal dynamic factor [],
KF is the face load factor for tooth-root stress [] and KF is the transverse load factor for tooth-root stress [].
The permissible bending stress Fp is calculated as follows [2,3,5,7]:
Fp = F lim YST YN Y YR YX

Fig. 4. Bending stress at the tooth root.

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

1587

where Flim is the nominal stress number (bending) [N/mm2], YST is the stress correction factor [], YN is the life factor for the
tooth-root stress [], Y is the relative notch sensitivity factor [], YR is the relative surface factor [] and YX is the size factor that
represents the tooth-root strength [].
The safety factor for bending stress SF is calculated as follows [2,3,5,7]:
SF =

Fp
F

3.2. Tooth contact stress

The tooth contact stress HC is calculated as follows, as seen in Fig. 5 [2,3,5,6].
The real contact stress H is calculated as follows [2,3,5,6]:
s
q
Ft u + 1
Z H Z E Z Z K A K V K H K H
H =
bm n u

10

where
d1 is the reference diameter of the pinion [mm], u is gear ratio [], Z H is the zone factor [], Z E is the elasticity factor
q
N=mm2 , Z is the contact ratio factor [], Z is the helix angle factor [], KH is the face load factor for contact stress [] and KH
is the transverse load factor for contact stress [].
The permissible contact stress Hp is calculated as follows [2,3,5,6]:
Hp = H lim Z N Z L Z V Z R Z W Z X

11

where H lim is the allowable stress numbers (contact) [N/mm2], Z N is the life factor for contact stress [], Z L is the lubrication
factor [], Z V is the velocity factor [], Z R is the roughness factor [], Z W is the work hardening factor [] and Z X is the size factor for
contact stress [].
The safety factor for contact stress SH is calculated as follows [2,3,5,6]:
SH =

Hp
H

12

4. Calculating the torsional vibration

The torsional vibration transmitted from the combustion engine to the transmission excites idler components such as the idler
gears, synchronizer rings and sliding sleeves, to vibrate within their functional clearances [11].
A dynamic gear vibration model is a useful tool to study the vibration response of a geared system with various gear parameters
and operating conditions [27].

Fig. 5. Contact stress at the tooth ank.

1588

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

Torsional vibration models of gears system are classied according to their rigidity and elasticity such as purely torsional multibody models, rigid multi-body models, exible multi- body models and semi-rigid-elastic multi-body models [2530].
Torsional vibration models of gears system are also classied according to time-invariant and time-variant such as linear timeinvariant (LTI) models with stiffness, linear time-varying (LTV) models with stiffness, time-varying models with backlash, as well
as time-invariant average stiffness and time-invariant models in both backlash and stiffness simultaneously [31].
4.1. Equation of motion of gears system
A four-degree-of-freedom model of the pinion gearwheel gear system is considered for simplicity. The pinion gear body and
wheel gear body are assumed to be rigid. The teeth are assumed to be elastic and parallel springdamper combinations are
assumed to exist between the teeth and the gear body. A four-degree-of-freedom model is shown in Fig. 6.
The equations of motion of the pinion gearwheel gear system are written in terms of the four-degree-of-freedom model as
follows [2528]: Sp, p, Sw and w.
Eq. (13) is written for tooth i of a pinion gear as follows:
::
2
2
Jpth S p = De rdp p S p + K e rdp p S p + T p

13

where Jpth is the moment of inertia of tooth i of the pinion gear [kg.mm2], r dp is the pitch circle radius of the pinion gear [mm], p is
the rotational position of the pinion gear body [rad], Sp is the rotational position of tooth i of the pinion gear [rad], p is the
::
rotational velocity of the pinion gear body [rad/s], S p is the rotational velocity of the tooth i of the pinion gear [rad/s], S p is the
2

rotational acceleration of the tooth i of the pinion gear [rad/s ], and Tp is the contact torque applied to tooth i [N.mm].
Eq. (14) is written for a pinion gear body as follows:
::
Jpb b = 0

14

::
where Jpb is the moment of inertia of the pinion gear body [kg.mm2], b is the rotational acceleration of the pinion gear body
[rad/s2].
Eq. (15) is written for tooth i of the wheel gear as follows:
::
2
2
J wth S w = De rdw
w S w + Ke rdw w Sw + Tw

15

where Jwth is the moment of inertia of tooth i of the wheel gear [kg.mm2], rdw is the pitch circle radius of the wheel gear [mm], w is
the rotational position of the wheel gear body [rad], Sw is the rotational position of tooth i of the wheel gear [rad], :: w is the
rotational velocity of the wheel gear body [rad/s], Sw is the rotational velocity of the tooth i of the wheel gear [rad/s], S w is the
rotational acceleration of tooth i of the wheel gear [rad/s2], and Tw is the contact torque applied to tooth i [N.mm].
Eq. (16) is written for the wheel gear body as follows:
::
J wb w = 0

16

::
where Jwb is the moment of inertia of the wheel gear body [kg.mm2], w is the rotational acceleration of the wheel gear body
[rad/s2].

Fig. 6. 4-Degree-of-freedom model.

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

1589

Ke is the stiffness coefcient [N/mm] and is assumed to be time-invariant. The stiffness coefcients Ke resulting from the tooth
surface contact are written as follows [29,32]:
ES0
412

Ke =

17

where E is Young's modulus [N/mm2], is the Poisson's ratio [] and S 0 is the thickness of gear [mm] which is written as follows
[3,5]:
S0 = m n = 2

18

De are the viscous damping coefcients [N.s/mm] and are assumed to be time-invariant. Assuming viscous damping, Rayleigh
damping is written as follows [20,21].
De = Ke

19

where is the damping ratio [].

The gear system equation of motion is written in matrix form as follows [2528]
::
:
J + D + K = T

20

where
J is the moment of inertia matrix, :D is the viscous damping matrix, K is the stiffness matrix, T is the vector of applied torques,
::
is the rotational acceleration vector, is the rotational velocity vector, and is the rotational position vector.
The moment of inertia matrix J is written as follows:
2

J pth
6 0
6
J=6
4 0
0

0
J pb
0
0

0
0
J wth
0

0
0
0

3
7
7
7
5

J wb

21

4x4

The viscous damping matrix D is written as follows:

2
6
6
6
D=6
6
4

2
De rdp

2
De rdp

2
De rdw

7
7
7
7
2 7
De rdw 5
0
4x4
0

22

The stiffness matrix K is written as follows:

2
6
6
6
K=6
6
4

Ke rdp

Ke rdp

Ke rdw

7
7
7
:
7
2
Ke rdw 7
5
0
4x4
0

23

The vector of applied torque vector T is written as follows:

2

3
Tp
6 0 7
7
T=6
4 Tw 5
0 4x1

24

::
The rotational acceleration vector is written as follows:

::
S
::p
::
p
= ::
Sw
::
w

4x1

25

1590

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

:
The rotational velocity vector is written as follows:
2 : 3
Sp
6 : 7
:
6 p 7
=6 : 7
4S 5
:w
w 4 x 1

26

The rotational position vector is written as follows:

2

3
Sp
6 p 7
7
=6
4 Sw 5
w 4 x 1

27

5. Singular value decomposition (SVD)

Some basic properties of SVD are revisited below [1821].
Let A F mxn where F is the eld (the eld may be real or complex C). There exist unitary matrices
U = u1 ; u2 ; :::; u m  F
such that

mxm

and V = v1 ; v2 ; :::; vn  F

mxm

28

29

A = UV*
where

=

1
0


0
; 1 = Diagf 1 ; 2 ; ::: p g
0

1 2 3 ::: p 0; p = minfm; ng:

30
31

In the above equations, i is the ith singular value of A, and the vectors u i and v i are, respectively, the ith left and right singular
vectors dened by the following eigenvalue problems
Avi = i u i

or A*ui = i vi

32

where a superposed asterisk denotes conjugated transpose. The following notations for singular values are adopted:

A = max A = 1 A = the largest singular value of A

33

A = min A = p A = the smallest singular value of A



34

Suppose that A and are arbitrary square matrices. Then, the following equations are valid:
1: max A

1
if A is invertible:
min A

35

2: i A i A 1 for any i:

36

3: i+j1 A + i A + j for any i and j:

37

5.1. Structural singular value

Assuming zero initial conditions, one gets the following harmonic response of a structure by taking the Laplace transform of the
transfer matrix (20) as follows [20,21]:
2

Js s + Dss + Ks = Ts

s = Js + Ds + K

T:

38

39

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

1591

The complex Laplace transform variable s is substituted by s = j, where is the excitation frequency, and j is the imaginary
unit. Then, Eq. (39) is written in the frequency domain as follows [16,17]:
2

= J + Dj + K

40

Note that singular values of the transfer matrix (J 2 + Dj + K) 1 in Eq. (40) are called structural singular values in this
article, and are function of .
6. Optimization of gearbox design parameters
Constrained optimization is a very useful tool for light-weight-structure design of machine elements with constraints such as
stress, deformation and vibration.
In optimization, the goal is usually to minimize the cost of a structure while satisfying the design specications [21]. By
optimizing the responsible parameters, it is possible to obtain a light-weight-gearbox structure and minimize the rattling noise
[13].
Let F(X) denote the objective function to be minimized, where X is the design parameter (variable) vector to be determined.
Then, to nd the constrained minimum of F(X), the following optimization problem is solved [20,21]:
minFX

41

subject to : LB X UB and G X 0

42

Fig. 7. Flow chart to optimize gearbox design parameters.

1592

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

where, LB and UB dene, the sets of lower and upper bounds on the design parameter (variable) X. The iterations start with the
initial design parameter vector X0 and a solution vector X is found that minimizes the objective function F(X)subject to the
nonlinear inequalities G(X) 0 [20,21].
7. Numerical example
Constrained optimization approaches are applied to the 5-speed gearbox for automotive transmission. All programs are
developed using the MATLAB program. In all optimization studies, the sequential quadratic programming (SQP) method is
employed.
To nd the optimum design parameter, the initial design parameters of the 5-speed gearbox for automotive transmission such
as m, z, , b, sa, and sv are varied. Thirty-six design parameters are optimized simultaneously using the developed programs. During
optimization, different initial value vectors are used to identify the global minimum solution of the objective function T (m, z, , b,
sa, sv).
7.1. Objectives function
Singular values of the gear system transfer matrix are considered as objective functions, and the design parameters are
optimized considering bending stress, contact stress and distance between gear center constraints. The owchart of the design
parameter optimization procedure is shown in Fig. 7.
The following objective function is employed:
F = min T:

43

The minimum singular values of the transfer matrix min(T) are dened as follows
2

min T = min J + Dj + K

44

The minimum singular values of the transfer matrix are considered to be the objective functions to be minimized, where
module m, the number of teeth z, helix angle , axial clearance sa and backlash sv are the design parameters (variables) to be
determined. Then, to nd the constrained minimum of the transfer matrix T (m, z, , b, sa, sv), the following optimization problem
is solved.
minTm; z; ; b; sa ; sv

45

subject to : LB m; z; ; b; sa ; sv UB and G X 0

46

where, LB and UB dene, the sets of lower and upper bounds on the design parameter (variables) vector such as m, z, , b, sa, and sv.
The iteration begins with the initial design parameter vector, which include, e.g., m0, z0, 0, b0, sa0 and sv0 and a solution
vector with m, z, , b, sa, and sv is found that minimizes the objective function T (m, z, , b, sa, sv) subject to the nonlinear
inequalities G (X) 0.
Although the number of DOF of structures is very large in practice, the computational cost of the associated singular value
problems is quite low for the objective function, because it is only necessary to compute the largest and smallest singular values
( max and min) that can be achieved by using selective eigenvalue solvers; the other singular values are not needed [20,21].
Table 1
Tooth bending stress parameters.
Parameter

Unit

1st pinion

2nd pinion

3rd pinion

4th pinion

Constant pinion

Rear pinion

Torque TL
Gear ratio u
Stress correction factor YST
Form factor YF
Stress correction factor YS
Transverse contact ratio
Overlap ratio
Application factor KA
Internal dynamic factor KV
Transverse load factor for tooth-root stress KF
Nominal stress number (bending) Flim
Life factor for tooth-root stress YN
Relative notch sensitivity factor Y
Relative surface factor YR
Size factor relevant to tooth-root strength YX

[N.mm]
[]
[]
[]
[]
[]
[]
[]
[]
[]
[N/mm2]
[]
[]
[]
[]

392.103
1.814
2
2.75
1.60
1
1
1.25
1.14
1.2
300
1
1
1
1

392.103
1.147
2
2.75
1.60
1
1
1.25
1.14
1.2
300
1
1
1
1

316.103
1.242
2
2.75
1.60
1
1
1.25
1.14
1.2
300
1
1
1
1

252.103
1.560
2
2.75
1.60
1
1
1.25
1.14
1.2
300
1
1
1
1

200.103
1
2
2.75
1.60
1
1
1.25
1.14
1.2
300
1
1
1
1

1148.103
2.84
2
2.75
1.60
1
1
1.25
1.14
1.2
300
1
1
1
1

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

1593

Table 2
Tooth contact stress parameters.
Parameter

Unit

Reference diameter of pinion d1

Gear ratio u
Zone factor ZH
Elasticity factor ZE

[mm]
[]
[]
q
N=mm2

1st pinion

189.8

189.8

189.8

189.8

189.8

189.8

Transverse load factor for contact stress KH

Allowable stress numbers (contact) Hlim
Life factor for contact stress ZN
Velocity factor ZV
Roughness factor ZR
Work hardening factor ZW
Size factor for contact stress ZX

[]
[N/mm2]
[]
[]
[]
[]
[]

1.2
800
1
1
1
1
1

1.2
800
1
1
1
1
1

1.2
800
1
1
1
1
1

1.2
800
1
1
1
1
1

1.2
800
1
1
1
1
1

1.2
800
1
1
1
1
1

61.116
1.814
1

2nd pinion
80.124
1.147
1

3rd pinion
76.716
1.242
1

4th pinion

Constant pinion

67.199
1.560
1

58.151
1
1

Rear pinion
41.319
2.84
1

7.2. Constraint functions

Tooth bending stress, contact stress and the distance between gear centers are considered to be the constraint functions in the
optimization. The tooth bending stress parameters, tooth contact stress parameters and torsional vibration parameters are shown
in Tables 13 respectively.
The following constraints are considered to be constraint functions
1: F Fp 0

47

where F is the real tooth-root stress [N/mm2] and Fp is the permissible bending stress [N/mm2].
2: H Hp 0

48

where H is the real contact stress [N/mm2] and Hp is the permissible contact stress [N/mm2].
49

3: a1 = a2 = a3 = a4 = a5 = aR = constant

where a1 is the center distance of the 1st speed, a2 is the center distance of 2nd speed, a3 is the center distance of the 3rd speed, a4
is the center distance of the 4th speed, a5 is the center distance of the 5th speed and aR is the center distance of rear speed.
7.3. Optimization results
The optimization results using objective function F are presented in Table 4. Because of the limited space, only the important
results are presented.
It is observed in solution 1 that the obtained optimum module changes between 3.3178 [mm] and 4.6740 [mm]. The optimum
number of teeth varies between 14 [] and 17 []. In addition, the helix angle varies between 26.3342 [] and 26.0866 [] while face
width varies between 25 [mm] and 28 [mm].
It is observed in solution 2 that the optimum module ranges from 2.9926 [mm] to 4.3290 [mm]. The numbers of teeth vary
between 14 [] and 19 []. Moreover, the optimum helix angle varies between 26 [] and 26.4142 [] and the optimum face width
varies between 28 [mm] and 32 [mm].
The results from solution 3 show that the optimum module ranges between 3.2862 [mm] and 4.3290 [mm]; the optimum
number of teeth vary between 14 [] and 19 []; the optimum helix angle is between 30.1647 [0] and 32 [0]; and the optimum face
width varies between 30 [mm] and 32 [mm].
The solution 4 results indicate that the optimum module varies between 3.1241 [mm] and 4.3290 [mm]; the optimum
number of teeth is between 14 [] and 20 []; the optimum helix angle varies between 30.1411 [] and 32 []; and the
optimum face width ranges from 30 [mm] and 31.0004 [mm].
Although the results given above represent the optimum solution, the standard design parameter values used by gear
manufacturers do not necessarily reect these results because some of the solutions are impossible in practice.
Table 3
Torsional vibration parameters.
Parameter

Unit

1st pinion

2nd pinion

3rd pinion

4th pinion

Constant pinion

Rear pinion

Gear ratio u
Young's modulusE
Poisson's ratio
Damping ratio

[]
[N/mm2]
[]
[]

1.814
21.104
0.3
0.1

1.147
21.104
0.3
0.1

1.242
21.104
0.3
0.1

1.560
21.104
0.3
0.1

1
21.104
0.3
0.1

2.84
21.104
0.3
0.1

1594

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

Table 4
Optimization results by using objective function F.
No 1
Lb = [2 2 2 2 2 2 14 14 14 14 14 14 20 20 20 20 20 20 20 20 20 20 20 20 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2]
Ub = [6 6 6 6 6 6 24 24 24 24 24 24 32 32 32 32 32 32 28 28 28 28 28 28 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5]
x0 = [5 5 5 5 5 5 18 18 18 18 18 18 26 26 26 26 26 25 25 25 25 25 25 25 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4]
Solution

m
z

b
sa
sv
SF
SH
a
Lp
LpComp
Lpaverage
CPU
SV

1st pinion

2nd pinion

3rd pinion

4th pinion

Constant pinion

Rear pinion

3.3178
17.0000
26.0866
25.0026
0.4000
0.4000
1.0196
1.2653
79.3590
72.9004
85.3506
84.3903
4.4665
0.0134

4.3494
17.0000
26.0000
25.0000
0.4000
0.4000
1.5550
1.4276
79.3735
72.1098
85.3916

4.1650
17.0000
26.0000
25.0000
0.4000
0.4000
1.7689
1.5506
79.3720
71.9973
85.3968

3.6474
17.0000
26.0000
25.0000
0.4000
0.4000
1.7012
1.5949
79.3669
71.6494
85.4120

4.6740
17.0000
26.0000
25.0000
0.4000
0.4000
3.5160
2.0757
79.4575
72.2932
85.5909

4.3290
14.0000
20.3342
28.0000
0.4000
0.4000
1.0000
2.1206
80.0000
84.4584
79.1996

No 2
Lb = [2 2 2 2 2 2 14 14 14 14 14 14 20 20 20 20 20 20 24 24 24 24 24 24 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2]
Ub = [7 7 7 7 7 7 24 24 24 24 24 24 32 32 32 32 32 32 32 32 32 32 32 32 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5]
x0 = [6 6 6 6 6 6 20 20 20 20 20 20 26 26 26 26 26 26 28 28 28 28 28 28 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5]
Solution

m
z

b
sa
sv
SF
SH
a
Lp
LpComp
Lpaverage
CPU
SV

1st pinion

2nd pinion

3rd pinion

4th pinion

Constant pinion

Rear pinion

2.9926
19.0000
26.1083
28.0031
0.5000
0.5000
1.0625
1.3602
79.9999
73.8058
81.7958
81.6667
3.3858
0.0012

3.9175
19.0000
26.0000
28.0000
0.5000
0.5000
1.5687
1.5109
79.9036
72.6827
81.9502

3.7509
19.0000
26.0000
28.0000
0.5000
0.5000
1.7842
1.6410
79.8912
72.5724
81.9631

3.2833
19.0000
26.0000
28.0000
0.5000
0.5000
1.7152
1.6879
79.8500
72.2320
82.0006

4.2065
19.0000
26.0000
28.0000
0.5000
0.5000
3.5440
2.1967
79.9226
72.8624
82.4361

4.3290
14.0000
26.4142
32.0000
0.5000
0.5000
1.0000
1.2453
80.0000
78.9505
79.8542

No 3
Lb = [2 2 2 2 2 2 14 14 14 14 14 14 20 20 20 20 20 20 24 24 24 24 24 24 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2]
Ub = [7 7 7 7 7 7 28 28 28 28 28 28 32 32 32 32 32 32 32 32 32 32 32 32 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6]
x0 = [6 6 6 6 6 6 20 20 20 20 20 20 32 32 32 32 32 30 30 30 30 30 30 32 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5]
Solution

m
z

b
sa
v
SF
SH
a
Lp
LpComp
Lpaverage
CPU
SV

1st pinion

2nd pinion

3rd pinion

4th pinion

Constant pinion

Rear pinion

4.0593
14.0000
32.0000
30.0000
0.5000
0.5000
1.0368
1.1542
79.9604
61.8075
80.4780
79.8926
2.2650
0.0010

3.9201
19.0000
32. 0000
30. 0000
0.5000
0.5000
1.3123
1.3772
79.9561
63.4120
80.4515

3.7537
19. 0000
32.0000
30.0000
0.5000
0.5000
1.4927
1.4958
79.9505
63.1788
80.4560

3.2867
19.0000
32.0000
30.0000
0.5000
0.5000
1.4354
1.5385
79.9317
62.4124
80.4692

4.2087
19. 0000
32.0000
30.0000
0.5000
0.5000
2.9643
2.0024
79.9648
63.7788
80.5366

4.3290
14.0000
30.1647
32.0000
0.5000
0.5000
1.0000
1.1867
80.0000
80.1381
76.9647

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

1595

Table 4 (continued)
Table 4 (continued)
No 4
Lb = [2 2 2 2 2 2 14 14 14 14 14 14 20 20 20 20 20 20 24 24 24 24 24 24 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2]
Ub = [7 7 7 7 7 7 28 28 28 28 28 28 32 32 32 32 32 32 32 32 32 32 32 32 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6]
x0 = [7 7 7 7 7 7 21 21 21 21 21 21 32 31 31 31 31 30 30 30 30 30 30 31 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6]
Solution

m
z

b
sa
sv
SF
SH
a
Lp
LpComp
Lpaverage
CPU
SV

1st pinion

2nd pinion

3rd pinion

4th pinion

Constant pinion

Rear pinion

4.0607
14.0000
32.0000
30.0000
0.6000
0.6000
1.0371
1.1542
79.9869
62.9285
84.1556
83.4316
2.0074
0.0012

3.7254
20.0000
31.0000
30.0000
0.6000
0.6000
1.0847
1.3152
79.9841
75.4959
83.5577

3.5675
20.0000
31.0000
30.0000
0.6000
0.6000
1.2339
1.4284
79.9826
75.4104
83.5709

3.1241
20.0000
31.0000
30.0000
0.6000
0.6000
1.1867
1.4693
79.9776
75.1493
83.6094

3.9993
20.0000
31.0000
30.0000
0.6000
0.6000
2.4500
1.9122
79.9864
75.6364
84.1882

4.3290
14.0000
30.1411
31.0004
0.6000
0.6000
1.0000
1.2122
80.0000
80.8208
81.5079

The rattle noise values vary between 76 [dB] and 80 [dB] during optimization. Thus, by optimizing the design
parameters, the rattle noise values are reduced to 76 [dB]. These rattle noise values are between 7% and 10% lower than the
calculated rattle noise values for the sample gearbox.
The safety factor for bending stress SF ranges between 1.00 and 3.54 during optimization. In addition, the safety factor
for contact stress SH varies between 1.15 and 2.19 during optimization. Thus, all optimized design parameters satisfy all
constraints.
The CPU time varies between 2 [s] and 4 [s] using objective function F during optimization.

8. Comparison of rattle noise levels

The sample gearbox rattle noise level and optimized gearbox rattle noise levels using objective function F for the 1st,
2nd, 3rd, 4th, 5th and rear speed are shown below.
A comparison between the rattle noise level of the sample gearbox and the optimized gearbox for the 1st speed is shown
in Fig. 8. While the rattle noise of the sample gearbox for the 1st speed is 88.1524 [dB], the rattle noise of the optimized
gearbox for 1st speed is 80.4780 [dB].
A comparison between the rattle noise level of the sample gearbox and the optimized gearbox for the 2nd speed is
shown in Fig. 9. While the rattle noise of the sample gearbox for 2st speed is 88.1893 [dB], rattle noise of the optimized
gearbox for the 2nd speed is 80.4515 [dB].
A comparison between the rattle noise level of the sample gearbox and the optimized gearbox for the 3rd speed is
shown in Fig. 10. While the rattle noise of the sample gearbox for the 3rd speed is 86.3327 [dB], the rattle noise of the
optimized gearbox for the 3rd speed is 80.4560 [dB].

Fig. 8. Comparison between the rattle noise level of the sample gearbox and the optimized gearbox for the 1st speed.

1596

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

Fig. 9. Comparison between the rattle noise level of the sample gearbox and the optimized gearbox for the for 2nd speed.

Fig. 10. Comparison between the rattle noise level of the sample gearbox and the optimized gearbox for the 3rd speed.

A comparison between the rattle noise level of the sample gearbox and optimized gearbox for the 4th speed is shown in
Fig. 11. While the rattle noise of the sample gearbox for the 4th speed is 86.3327 [dB], the rattle noise of the optimized
gearbox for the 4th speed is 80.4692 [dB].
The rattle noise level of the sample gearbox and the optimized gearbox for the 5th speed is compared in Fig. 12. While
the rattle noise of the sample gearbox for the 5th speed is 88.4915 [dB], the rattle noise of the optimized gearbox for the
5th speed is 80.5366 [dB].
The rattle noise level of the sample gearbox and optimized gearbox for rear speed is compared in Fig. 13. While the
rattle noise of the sample gearbox for the rear speed is 87.7589 [dB], the rattle noise of the optimized gearbox for the rear
speed is 76.9647 [dB].

Fig. 11. Comparison between the rattle noise level of the sample gearbox and the optimized gearbox for the 4th speed.

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

1597

Fig. 12. Comparison between the rattle noise level of the sample gearbox and the optimized gearbox for the 5th speed.

Fig. 13. Comparison between the rattle noise level of the sample gearbox and the optimized gearbox for the rear speed.

It is shown that the rattle noise values of the optimized gearbox are lower than the calculated rattle noise values for the
sample gearbox for each speed.
9. Conclusion
It is concluded through simulation that increasing the geometric parameters of the gearbox, such as the module and number of
teeth results in increased rattle noise. In addition, increased axial clearance results in increased rattle noise until the axial clearance
reaches its maximum value, then the rattle noise decreases. Moreover, increased backlash causes decreased rattle noise until the
backlash reaches its maximum value, then the rattle noise increases. Changing the helix angle resulted in different levels of rattle
noise, while changing the face width resulted in a constant level of rattle noise. Furthermore, increasing the gearbox operational
parameters, such as the angular acceleration and excitation frequency, caused increased gearbox rattle noise [33].
Some geometric design parameters, such as the module, and number of teeth must satisfy desired safety protocols, and
some backlash are necessary to allow room for an oil lm for all conditions of thermal expansion and contraction. Although,
there is no relationship between face width and rattle noise, face width is necessary to satisfy the desired contact safety
requirement. Therefore, by optimizing the geometric parameters of the gearbox, including the module, number of teeth,
axial clearance, and backlash, it is possible to obtain a light-weight-gearbox structure and minimize the rattling noise.
Optimized geometric design parameters lower the rattle noise by 10% compared to the calculated rattle noise values for
the sample gearbox. All optimized geometric design parameters also satisfy all constraints. Optimizing the geometric design
parameters not only reduces the rattle noise but also increases the desirable bending stress and contact stress level.
While geometric parameters, such as the module, number of teeth, helix angle, face width, backlash and axial clearance
are optimized, the operational parameters, such as angular acceleration and excitation frequency are not optimized because
these operational parameters are given by the automotive manufacturer as input values.
Acknowledgements
This research study is supported by the Institute of Machine Elements (IMA), University of Stuttgart. The author thanks Prof.
Dr.-Ing. B. Bertsche, Dipl.-Ing. P. Fietkau, Dipl.-Ing. A. Baumann, Dipl.-Ing. W. Novak, Dipl.-Ing. S. Nebel and Dipl.-Ing. S.
Sanzenbacher for their helpful co-operation.

1598

M. Bozca / Mechanism and Machine Theory 45 (2010) 15831598

References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]

J.D. Smith, Gear Noise and Vibration, Second Edition, Markel Dekker, Inc., Newyork, 2003.
H. Naunheimer, B. Bertsche, G. Lechner, Fahrzeuggetriebe, 2. Auage, Springer Verlag, Berlin Heidelberg, 2007.
Roloff, Matek, Maschinenelemente, Vieweg & Sohn Verlag/Fachverlage GmbH, Wiesbaden, 2005.
H.D. Eckhardt, Kinematic Design of Machines and Mechanisms, McGraw-Hill, U.S.A, 1998.
ISO 63361: calculation of load capacity of spur and helical gears, part 1: basic principle, introduction and general inuence factors.
ISO 63362: calculation of load capacity of spur and helical gears, part 2: calculation of surface durability (pitting).
ISO 63363: calculation of load capacity of spur and helical gears, part 3: calculation of tooth bending strength.
ISO 63365: calculation of load capacity of spur and helical gears, part 5: strength and quality of materials.
S.N. Dogan, J. Ryborz, B. Bertsche, Design of low-noise manual automotive transmissions, Proc. IMechE 220 Part K: J. Multi-body Dynamics, 2006, pp. 7995.
S.N. Dogan, J. Ryborz, B. Bertsche, Rattling and clattering noise in automotive transmission-simulation of drag torque and noise.
B. Bertsche, W. Nowak, M. Stockmeier, S.N. Dogan, J. Ryborz, Development Process of Clatter and Rattle Noise Free (CARF-) Transmissions, Proceedings of
IDETC/CIE 2005, ASME, Long Beach, California, USA, September 2005, pp. 19.
S.N. Dogan, J. Ryborz, B. Bertsche, Low-noise Automotive Transmission Investigations of Rattling and Clattering, 3rd International Symposium on Multi-body
Dynamics: Monitoring & Simulation Techniques, Loughborough University, Loughborough, Leicestershire, UK, July 2004, pp. 118.
S.N. Dogan, Loose part vibration in vehicle transmission-gear rattle, (TBITAK), Tr. J. of Engineering and Environmental Series 23 (1999) 439454.
W. Novak, M. Stockmeier, B. Bertsche, Improvement of Power Train Efciency Via CARF- Transmissions, International Conference on Engineering Design,
ICED, Melbourne, August 2005.
R. Brancati, E. Rocca, R. Russo, A gear rattle model accounting for oil squeeze between the meshing gear teeth, Proc. IMechE 219 Part D, J. Automobile
Engineering (2005) 10751083.
P. Bellomo, F. Cricenti, N. De Vito, C.H. Lang, D. Minervini, Innovative Vehicle Power Train System Engineering: Beating the Noisy Offenders in Vehicle
Transmissions, Society of Automotive Engineers, ASE 2000 World Congress, Detroit, Michigan, USA, March 2000.
M. Vaishya, R. Singh, Strategies for modeling friction in gear dynamics, Journal of Mechanical Design (2003) 383393.
G.H. Golub, C.F.V. Loan, Matrix Computation, Johns Hopkins University Press, Baltimore, 1983.
R.A. Horn, C.R. Johnson, Topics in Matrix Analysis, Cambridge University Press, Cambridge, 1995.
M. Bozca, Integrated optimum design of structures and controllers, Ph.D. dissertation submitted to Istanbul Technical University, Istanbul, Turkey, 2005
M. Bozca, A. Mugan, H. Temeltas, Decoupled approach to integrated optimum design of structures and robust control systems, Structural and
Multidisciplinary Optimization, Springer 36 (2007) 169191.
M. Bozca, Inuence of Core Hardness on Bending Strength of Tooth in Cylindrical Gears under Quasistatic Loading Conditions, Fatigue & Fracture of
Engineering Materials & Structures, vol. 31, Blackwell Publishing Ltd, U.K., 31 (2008), 902910.
G.P. Cavallaro, T.P. Wilks, C. Subramanian, C. Straffortd, K.N.P. French, J.E. Allison, Bending fatigue and contact fatigue characteristics of carburized gears,
Surface and Coatings Technology 71 (1995) 182192.
M. Mackaldener, M. Olsson, Interior fatigue of gear teeth, Fatigue and Fracture of Engineering Materials and Structures 23 (2000) 283292.
Hans Dresig, Franz Holzweiig, Maschinendynamik, 6. Auage, Springer-Verlag, Berlin Heidelberg, 2005.
Saeed Ebrahimi, A contribution to computational contact procedures in exible multibody systems, Ph.D. dissertation submitted to Institute of Engineering
and Computational Mechanics, University of Stuttgart, Stuttgart, Germany, 2007
Saeed Ebrahimi, Peter Eberhard, Rigid-elastic modeling of meshing gear wheels in multibody systems, Multibody System Dynamics 16 (2006).
Saeed Ebrahimi and Peter Eberhard, Contact modeling of meshing gear wheels using tangentially movable teeth, SIMPACK User Meeting, Baden Baden
(2006).
A. Parey, M. El Badaoui, F. Guillet, N. Tandon, Dynamic modeling of spur gear pair and application of empirical mode decomposition-based statistical analysis
for early detection of localized tooth defect, Journal of Sound and Vibration 294 (2006) 547561.
J. Peters, D. Vandepitte, P. Sas, Flexible multi body model of a three-stage planetary gearbox in wind turbine, Proceeding of ISMA, 2004, pp. 39233941.
A. Kahraman, R. Singh, Interactions between time-varying mesh stiffness and clearance non-linearities in a geared system, Journal of Sound and Vibration 146
(1990) 135156.
N. Sarkar, R.E. Ellis, T.N. Moore, Backlash detection in geared mechanism: modelling, simulation, and experiment, Mechanical System and Signal Processing 3
(1997) 391408.
M. Bozca, P. Fietkau, Empirical model based optimization of gearbox geometric design parameters to reduce rattle noise in an automotive transmission,
Mechanism and Machine Theory, in press.