Downloaded from SAE International by Purdue University, Tuesday, August 21, 2018

Modeling of Gearbox Whining Noise 2014-01-2090

Published 06/30/2014

Joël Perret-Liaudet
Ecole Centrale de Lyon

Alexandre Carbonelli
Vibratec

Emmanuel Rigaud
Ecole Centrale de Lyon

Brice Nelain and Pascal Bouvet

Vibratec

C. Jacques Vialonga
Renault SA

CITATION: Perret-Liaudet, J., Carbonelli, A., Rigaud, E., Nelain, B. et al., "Modeling of Gearbox Whining Noise," SAE
Technical Paper 2014-01-2090, 2014, doi:10.4271/2014-01-2090.
Copyright © 2014 SAE International

Abstract finite element model of the gearbox (including housings, shafts

and bearings) is used to extract the modal basis necessary for
The main source of excitation in gearboxes is generated by the
this method.
meshing process, which generates vibration transmitted to the
casings through shafts and bearings. Casing vibration
Measurements carried out on a specific gearbox test bench
generates leads to acoustic radiation (whining noise). It is
allow comparison of simulation and experimental results.
usually assumed that the transmission error and variation of
Accelerometers and optical encoders were used to measure
the gear mesh stiffness are the dominant excitation
static and dynamic errors of transmission and the vibration
mechanisms. These excitations result from tooth deflection and
response of the casings.
tooth micro-geometries (voluntary profile modifications and
manufacturing errors). For real cases, the prediction of noise
The correlation between calculations and measurements at
induced by the Static Transmission Error (STE) remains a
each stage of the computation chain is used to validate the
difficult problem.
relevance of these numerical approaches.
In this work, an original calculation procedure is implemented
by using a finite element method and taking into account the Introduction
parametric excitations and their coupling (Spectral Iterative
A gearbox can be one of the main sources of noise and
Method, developed by the Ecole Centrale de Lyon). The
vibration in automotives. The gearbox internal sources of
procedure is based on a modal approach developed in the
excitation are various and depend on the state of the gearbox.
frequency domain, particularly efficient to analyze systems
For instance the whining noise arises from loaded shafts,
having many degrees of freedom.
especially in reverse running. The source of excitation is a
parametric excitation and is transmitted to the gearbox housing
In a first step, the static transmission errors and tooth
via the crankshafts and bearings, as presented Figure 1. The
stiffness's are calculated from the knowledge of the tooth
gearbox housing will then transmit the vibration and noise in
macro and micro geometry. In a second step, these data are
the passenger compartment directly, via noise transparency of
used to calculate the dynamic transmission errors, the teeth
the car panels, and indirectly via structure-borne vibration
dynamic loads, and the dynamic response of the gearbox. A
propagation.
 Downloaded from SAE International by Purdue University, Tuesday, August 21, 2018

The following data are needed to perform this calculation:

• initial gaps e between the teeth : they are function of

the geometry defects and the teeth micro-geometry
modifications,
• compliance matrix C, of the teeth obtained from interpolation
functions calculated by a Finite Element model of elasto-
static deformations,
• Hertz deformations hertz, calculated according to Hertz's
theory.

The calculation of the actual approach of distant teeth on the

Figure 1. Generation and transmission of whining noise in a gearbox. contact line for each position θ permits to access the time
1: parametric excitation between teeth, 2: propagation in the gearbox, variation of STE, as a function of the applied torque (or the
3: vibration of housing applied load P), and the teeth modifications.

The excitation is divided in two phenomena: the transmission

The parametric mesh stiffness k(t) is thus directly related to the
error and the stiffness mesh fluctuations [1]. The transmission
applied torque T = P · Rb and the static transmission error δ(t)
error is mainly due to voluntary (tooth modifications) and
with:
involuntary (manufacturing errors) geometrical deviations of the
teeth, at a micrometric scale. The flexibility of the teeth, the
pinion and the shafts result in additional fluctuations of the
transmission error. The contact between teeth can be primarily
modeled by mesh stiffness (see Figure 2), which varies rapidly where Rb stands for the base radius of the driven gear and e(t)
in time depending on the contact conditions (number of teeth in for the static transmission error expressed in meters (projected
contact, position of the contact lines between meshing on the action line).
teeth…).
The complete computation scheme is summarized Figure 3

Figure 2. Equivalent model of the contact between two pinions: the

pinions are modeled as 3D body or as concentrated inertia while the
teeth contact is modeled with a “mesh” stiffness [2].

Transmission Error Computation

For geared system, the STE under load [3] is one of the main
noise sources. It corresponds to the difference between the
actual position of the driven gear and its theoretical position for
a very slow rotation velocity and for a given applied torque. Its
characteristics depend on the instantaneous situations of the
meshing tooth pairs. STE results from teeth deflections, teeth Figure 3. Complete transmission error computation procedure.
surface modifications and manufacturing errors. The
calculation of STE is relatively classical [4]. For each position θ Studied System
of the driving gear, a kinematical analysis of the mesh allows
The studied gearbox is presented in Figure 4, where a part of
determination of the theoretical contact line on the mating
the housing has been removed. The gearbox is composed of
surfaces of gearing teeth within the plane of action.
four gears, three shafts and 2 meshes. The number of teeth Z
of each meshing gear is indicated. Accelerometers and optical
Equation system which describes the elasto-static
encoders were used to measure static and dynamic errors of
deformations of the teeth can be written as follows [4]:
transmission and the vibration response of the casings.
 Downloaded from SAE International by Purdue University, Tuesday, August 21, 2018

All the computational scheme steps are summarized in Figure

5.

Figure 4. Picture of the studied gearbox with the two considered

meshes and the instrumentation.

Development of a Scheme to Calculate the

Dynamic Response
This computation scheme requires a finite element model of
the complete gearbox in order to obtain its modal basis. The
contact between the gears is modeled with a stiffness matrix
linking the degrees of freedom of each pair of meshing gears. Figure 5. Overview of the computational scheme.
To achieve that, the mean value of the mesh stiffness is taken,

i.e. (T is the mesh period), leading to mean modal
characteristics. The scheme uses then a powerful resolution
algorithm in frequency domain to solve the dynamic equations
with an iterative procedure [5-6]. The original spectral iterative
method has been extended in order to take into account
several parametric excitations. The complete description of the
spectral iterative method can be found in [7]. In that case, there
is a coupling between the excitations due to the stiffness's
fluctuations. The corresponding set of equations is then:
Validation of the Computation Scheme
The computation scheme has been validated step by step by
comparisons with extensive and complex measurements on a
modified but representative automotive gearbox as presented
in Figure 6.
Four quantities have been measured: the static transmission
error fluctuation, the dynamic transmission error, housing
vibration and whining noise. Accelerometers, microphones and
optical encoders are used in this way. In this paper, the results
are mainly focused on the housing vibration.

The measurements were performed at RENAULT's workshop

M, C, and K are respectively the global mass, damping and
stiffness matrices of the system.
x the vector of the generalized coordinates of the system,
stands for the time derivative.
in Lardy in France, on the BACY acyclism test bench. An
electrical motor drives the gearbox, while a braking torque
simulates the reaction of the wheel. The rotation speed and the
torque are also measured.

The superscript m corresponds to the number of mesh (i.e. the

number of excitations).

Rj are the macro-geometric coupling vectors of the degrees of

freedom of two meshing gears.

The final outputs are thus the dynamic transmission error

(DTE), the teeth dynamic loads and the housing vibration as a
function of the frequency. The operating speeds corresponding
to resonance peaks and the amplitude of the housing vibration
characterize the whining noise severity. The process can be
repeated for several applied torques. The first part of the
computation scheme, i.e. STE computation, can be used to
optimize the teeth geometry in order to minimize the excitation
[8]). The dynamic calculation can be used to optimize other Figure 6. Step by step validation of the computation scheme.
gearbox components (for instance the geometry of the housing Intermediate quantities such as transmission errors, as well as housing
to minimize its vibration, stiffness of gear bearings…). dynamic vibration are compared.
 Downloaded from SAE International by Purdue University, Tuesday, August 21, 2018
The tuning of the assembled gearbox model has not been
done ideally because the experimental modal analysis was not
possible on the BACY test bench when a static torque is
applied. Preloads effect on mesh and bearings stiffness's are
thus not taken into account and may not be negligible [9].
Results from a former experimental modal analysis performed
by Vibratec have instead been used. Measurements have been
done with a static torque applied but the clamping conditions of
the gearbox are different from the ones in the test bench BACY
on which the vibration measurements have been done.
As some parts of the gearbox can show non negligible
discrepancies over some frequency range, the final assembly
is not completely able to describe accurately the dynamic
behavior of the measured gearbox. But the results obtained
are precise enough to validate the computational scheme.
Figure 7 displays the comparison between measured and
computed transmission errors. The mean value cannot be
obtained by measurements, but the fluctuations, which are the
most important data, can be compared. The peak-to-peak
amplitude is correctly estimated. The measurement of the
transmission error is particularly complicated. The dispersion
due to manufacturing errors and assembling errors can be
large. Moreover, the micro-geometry should be accurately
measured tooth by tooth in order to have real tooth topologies.
The agreement between measurements and computations is
really satisfying. The measurement curve is a reconstitution
from the spectra analysis of the experimental data.
Figure 7. Measured (left) and computed (right) static error
transmission. Peak-to-peak values are compared for the validation
(both scales are the same).
Figure 8 shows a comparison of the housing vibration (dynamic
acceleration) as function of the operating speed. The
comparison is based on predominance of orders and modes, in
terms of frequency and amplitude.
Figure 8. Comparison of measured (left) and simulated (right) housing
vibration as a function of the motor regime and frequency. The analysis
highlights the principal orders and the predominant frequencies.
An order tracking has also been done in order to compare
properly the vibration measurements with the computations.
The meshing order and its first harmonic orders of the two
meshes have been considered (the first mesh corresponds to a
gear ratio of Z1/Z2=35/39. The second mesh corresponds to a
gear ratio Z3/Z4=16/69 as specified in Figure 4). The
acceleration of one housing point for the second order of the
second mesh is displayed in Figure 9.
Figure 9. Tracking of the second order of the second mesh. Dynamic
response of a housing point in the normal direction. Red solid:
measurements. Purple dashed: simulations.
The dynamic model has been tuned in different operating
conditions explaining some non negligible frequency shifts and
modal response discrepancies. However the agreement
between the measurements and the computations remains
satisfying enough for a predicting tool.
The measurements cannot take into account the variability of
the results, but the simulation can. Extracted from teeth
metrology, a dispersion study has been performed to determine
the envelope of the dynamic response. As the teeth micro-
geometric dispersion observed in metrology reports doesn't
follow a Gaussian law, the gear faults repartition over the
tolerance range has been considered uniform, which is a rather
a pessimistic case. It is well known that the dispersion due to
manufacturing and assembling errors can lead strong
variability of the dynamic behavior and noise radiated for
geared systems (sometimes up to 10 dB [10 - 11]). One
hundred excitations have thus been computed and the
corresponding responses are plotted in Figure 10 for both
meshes and for the two first orders. Discrepancies can be
observed due to a bad modal behavior representation at some
frequencies. Nevertheless, the order of magnitude of the
response is in a good agreement with the measurements. The
dispersion for the first mesh is much higher than for the second
mesh, and second orders are more sensitive to the dispersion
than the first orders.
One of the main industrial concerns is to build a source noise
hierarchy to determinate for instance which housing point has
the highest vibration, and which order is dominating.
 Downloaded from SAE International by Purdue University, Tuesday, August 21, 2018
Figure 10. Quadratic mean value over 10 housing points order by order
and mesh by mesh. Comparison between measurements and 100
dynamic responses generated randomly according the gear faults
dispersion.
Figure 11 displays the RMS value of the acceleration of some
strategically chosen points on the housing, for both simulation
and measurements. Considering these results, the
computations and the measurements advice the point Nr. 2 as
the less vibrating, and should therefore be used as an
attachment point for to the rest of the structure. Once again,
the frequency shifts due to a model tuning in different operating
conditions introduce some level difference and some low
frequency excitations may have not been taken into account
(e.g. shafts eccentricity) explaining the discrepancies at lowest
frequencies. Nevertheless the hierarchy between the different
points remains acceptable as a first order analysis tool.
Figure 11. RMS value of the acceleration of some housing points. Left:
simulations. Right: measurements.
Summary/Conclusions
The computation scheme presented in this paper is a complete
method to predict whining noise severity, accounting for the
scattering of the manufacturing data. The scheme is globally
validated and can be used to optimize the current studied
gearbox. From the knowledge of micro and macro geometry of
the gears the system excitations (STE and mesh stiffness
fluctuations) can be correctly estimated (and even minimized
by teeth modifications optimization). Then with a FE model of
the complete system, the computations permit to identify the
key parameters in order to minimize the whining noise for given
operating configurations by minimizing the excitation source,
i.e. the static transmission error fluctuations. A particular effort
should be done to ensure a good numerical model tuning. The
prediction remains satisfying despite some bad tuned model
parameters. The dynamic response can give a accurate
enough information about the parietal velocity of the housing
and thus to the acoustic power of the whining noise induces by
geared systems.
References
1. Remond, D., Velex, P., Sabot, J., “Comportement
dynamique et acoustique des transmissions par
engrenages : synthèse bibliographique”. 1993 hal-
00694869.
2. Welbourn, D. B., “Fundamental knowledge of gear noise-a
survey,” in Proceedings of the Conference on Noise and
Vibrations of Engines and Transmissions, vol. C177/79, pp.
9-29, Cranfield Institute of Technology, July 1979.
3. Harris, L.S., “Dynamic loads on the teeth of spur gears”,
Proceedings of the Institution of Mechanical Engineers 172,
(1958), pp.87-112.
4. Rigaud, E, Barday, D. “Modeling and analysis of static
transmission error of gears: effect of wheel body
deformation and interactions between adjacent loaded
teeth”, Mécanique Industrielle et Matériaux Vol. 51 Issue 2
(1998): pp. 58-60.
5. Perret-Liaudet, J., “Etude des Mécanismes de Transfert
entre l'Erreur de Transmission et la Réponse Dynamique
des Boîtes de Vitesses Automobiles”, Thèse de doctorat de
l'Ecole Centrale de Lyon N°9207,” 1992.
6. Perret-Liaudet J., “An original method for computing the
response of a parametrically excitated forced system,”
Journal of Sound and Vibration, vol. 196, pp. 165-177,
1996.
7. Carbonelli, A., “Caractérisation vibro-acoustique d'un
cascade de distribution poids lourd”,Thèse de doctorat de
l'Ecole Centrale de Lyon N°2012-34”, 2008.
8. Carbonelli, A., Rigaud, E., Le Bot, A. et al. “Particle swarm
optimization as an efficient computational method in order
to minimize vibrations of multimesh gears transmission.”
Advances in Acoustics and Vibration, 2011, 2011.
9. Åkerblom, M., Sellgren, U., “Gearbox Noise and Vibration:
Influence of Beraing Preload”, TRITA-MMK 2008:17 / ISSN
1400-1179 / ISRN/KTH/MMK/R-08/17-SE, Stockholm,
2008.
10. Nonaka, T., Kubo, A., Kato, S., Ohmori, T., “Silent gear
design for mass produced gears with scratters in tooth
accuracy”, ASME Proceedings of the International Power
Transmission and Gearing Conference, Scottsdale, USA,
Vol. 2, pp. 589-595, 1992.
11. Driot, N., “Etude de la dispersion vibro-acoustique des
transmissions par engrenages”, Thèse de doctorat de
l'Ecole Centrale de Lyon N°2002-34”, 2002.
 Downloaded from SAE International by Purdue University, Tuesday, August 21, 2018

Acknowledgments Definitions/Abbreviations
This work has been performed within the project MABCA, TE - Transmission Error
funded by ADEME and ANR (contract number: ANR-08- STE - Static Transmission Error
VTT-007-02). The authors acknowledge gratefully this support.
DTE - Dynamic Transmission Error
The partners involved are Ecole Centrale de Lyon, RENAULT
and RENAULT TRUCKS. FE - Finite Elements

The authors want to thank especially J. Vialonga from Renault

- Technical center of Lardy (France) for his support.

The Engineering Meetings Board has approved this paper for publication. It has successfully completed SAE’s peer review process under the supervision of the session
organizer. The process requires a minimum of three (3) reviews by industry experts.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical,
photocopying, recording, or otherwise, without the prior written permission of SAE International.

Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE International. The author is solely responsible for the content of the
paper.

ISSN 0148-7191

http://papers.sae.org/2014-01-2090