You are on page 1of 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/266458660

Gear noise evaluation through multibody TE-based simulations

Article · January 2010

CITATIONS READS

9 298

6 authors, including:

Antonio Palermo Domenico Mundo


Ferrari Università della Calabria
14 PUBLICATIONS   88 CITATIONS    106 PUBLICATIONS   807 CITATIONS   

SEE PROFILE SEE PROFILE

Peter Mas
Siemens
42 PUBLICATIONS   470 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

ARRAYCON - application of distributed control on smart structures View project

Rocket acoustic loads View project

All content following this page was uploaded by Wim Desmet on 03 April 2016.

The user has requested enhancement of the downloaded file.


Gear noise evaluation through multibody TE-based
simulations

A.Palermo1, D. Mundo1, A.S. Lentini2, R. Hadjit2, P. Mas2, W. Desmet3


1
University of Calabria, Department Mechanical Engineering
Ponte Pietro Bucci, 87036, Rende, Italy
email: palermo.antonio@unical.it
2
LMS International,
Interleuvenlaan 68, B-3001, Leuven, Belgium
2
K.U.Leuven, Department Mechanical Engineering
Celestijnenlaan 300 B, B-3001, Heverlee, Belgium

Abstract
The possibility of estimating noise radiating from a gearbox is essential to achieve valid design solutions
in shorter timeframes and to limit the testing phase, especially in those industrial fields, such as
automotive, helicopter and wind turbine industry, with a strong demand for gear noise reduction. This
paper presents a methodology for the calculation of gear bearing forces, useful for the acoustic analysis of
gearboxes and applicable to spur as well as helical parallel gear systems. The methodology is based on the
implementation of a procedure for the computation of the dynamic transmission error (DTE) in a
multibody environment. The DTE is obtained from the static transmission error (STE), i.e. the static
relative displacement between meshing teeth, which is variable along the mesh cycle. The adopted
multibody technique enables to overcome the principal drawbacks of FEM, achieving good computational
efficiencies, and of analytical models, avoiding to lump the system in one or few degrees of freedom.
These goals are reached by means of a user-defined force element, acting as teeth meshing force, which
stems from the integration of the multibody software, LMS Virtual.Lab Motion, with an external program
specialized in gear meshing analysis. The multibody software captures the system dynamics and includes
the nonlinear effects such as gear backlash, bearing clearances and stiffness; the specialized software
enables to consider tooth microgeometry, assembly errors, global and contact tooth stiffness and also shaft
deflections. The new feature introduced by the proposed technique is the ability to take into account the
instantaneous torque with good computational efficiency.

1 Introduction

Gears are extensively employed in mechanical systems since they allow the transfer of motion in a wide
range of working conditions, with a variety of gear ratios, and at reasonable production costs. The gear
meshing is a complex process because it involves moving and multiple contact points, variable load
sharing on the meshing teeth, contact mechanics (which is nonlinear), and all of them from a dynamic
standpoint. Furthermore tooth microgeometry, manufacturing imperfections and assembly errors have
relevant effects on the behavior of gear systems and cannot be ignored [1-3]. This complexity has to be
faced in the design phase, which must address both endurance and noise requirements. For these reasons, a
considerable amount of research works on gear dynamics is available in literature, but still many aspects
remain unresolved. Moreover, today’s markets are highly competitive and therefore reaching valid
solutions in shorter timeframes represents a clear advantage. In this context, numerical models and
simulations allow to achieve solutions to improve the dynamic behavior of gear systems, and to limit the
testing phase saving time and money. This explains why efforts continue to be spent in the gear dynamics
research field, with applications especially in helicopter [1], wind turbine [4] and automotive [5]

3033
3034 P ROCEEDINGS OF ISMA2010 INCLUDING USD2010

industries. Gear dynamics are mentioned because, at the occurrence, the proposed methodology enables to
evaluate the dynamic meshing loads, which in transients can be several times higher than the static ones.
From this perspective, the proposed technique will also allow the simulation of load sharing in planetary
gear trains, which is currently a major issue for this kind of transmissions [6-8]. Coming back to gear
noise purposes, the proposed methodology takes into account the dynamic transmission error (DTE),
which is defined for a gear pair as the dynamic relative displacement between meshing teeth.
The transmission error is widely regarded as one of the main causes of gear noise [10, 11]. According to
Munro [12], the transmission error was defined first by Harris [13] in 1958, who started its analytical
investigation. Nevertheless references [14, 15] prove that this concept was already applied before, but
using an empirical approach. The two main factors affecting the TE are the mesh stiffness, which accounts
for tooth flexibility and number of meshing tooth pairs, and the tooth microgeometry, in terms of
intentional modifications [10, 14-16] and manufacturing errors. Variations in the TE, during the gear
meshing, trigger vibrations and then airborne noise [17, 18]. The present analysis will be focused on the
case of involute parallel spur and helical gears, which are the most common ones.
Several methodologies are available in literature to simulate and estimate meshing vibrations, using
analytical lumped parameter, Finite Element (FE), and multibody approaches. The analytical model
proposed by Umezawa [19-22], later corrected by Cai [23, 24] to consider the influence on tooth stiffness
of the gear tooth number, describes the gear meshing with a single degree of freedom (SDOF) system
aligned along the line of action. Assuming a time-varying function for the mesh stiffness, defined within
one mesh period (or, adimensionally, within one mesh cycle), and the damping, the equation of motion
can be solved. These models allow to consider the effects of tooth microgeometry, assembly and
manufacturing errors, lumping them on the line of action with a displacement-driven excitation for the
SDOF system. With these assumptions it is impossible to consider the three-dimensionality of the contact
problem, and the quality of the results that can be obtained depends on how realistic are the mesh stiffness
function for the given gear pair and the displacement excitation. Different analytical models improved the
accuracy of the results using a FE model to take into account shaft deflections and three-dimensional
geometry [25-28], but the lumped parameters description is still suitable to analyze simple cases. Full FE
models [29] allow more accurate representations of gear systems and avoid a-priori assumptions on the
TE, but since tooth contact happens in a very small area, and it spans the teeth from root to tip, highly
refined mesh or contact detection followed by remeshing is needed along the whole tooth face. This
causes high computational costs to run a simulation. Moreover, it is also an issue to correctly describe the
tooth three-dimensional microgeometry. In the FE field, an interesting technique is proposed by Parker et
al. [9], who use a semi-analytical finite element formulation specifically devised for contact problems
[30]. The tooth is divided in a contact zone (extending beneath the tooth surface) and a FE zone, separated
by a matching interface. The contact zone is analytically solved by means of the Boussinesq’s solution.
This solution is evaluated at the matching FE nodes and the obtained nodal parameters are used to solve
the remaining FE part. However, FEM techniques are not able to consider with good computational
efficiencies nonlinear entities such as gear lashes, assembly clearances, bearings, clutches, and other
nonlinear effects which arise from large angle rotations. Such nonlinear effects require time-domain
integration, which is typical of the multibody environment. Moreover, to assess the noise and vibration
performances of a geared system it is usually desirable to test a wide range of working conditions (e.g.,
torques, regime run-up, …). With such demands, the use of large scale finite element models in time
domain becomes computationally expensive and maybe impractical.
The technique proposed in this paper is an improvement of the Static Transmission Error method
described by Morgan et al. in [31]. The basic idea of the method is to let a specialized, thus highly
efficient, software for gear contact analysis (abbreviated GCAS from now on) execute the calculation of
the static mesh stiffness, which can then be used in the dynamic multibody simulation. The gear meshing,
in the multibody simulation, is in fact governed by the dynamic equilibrium of the contact forces applied
to the gears, rather than the ideal kinematic contact ratio. Considering a single spur or helical gear pair
described by the standard gear parameters, the static mesh stiffness can be calculated using GCAS which
enables to take into account three-dimensional teeth microgeometric modifications and manufacturing
errors, teeth global and contact stiffness, shaft deflections and assembly misalignments. This mesh
stiffness is obtained, for one static working condition, as a function of the position along the mesh cycle.
M ULTI - BODY DYNAMICS AND CONTROL 3035

Once the static mesh stiffness is imported in the multibody software, the contact forces are calculated and
applied to the gears by a user-defined force element which reads the instantaneous value of the mesh
stiffness based on the actual position along the mesh cycle. In this way, the meshing complexity is
captured avoiding the high computational cost related to the full-scale model, since the multibody gears
and shafts models are rigid (while the bearings are compliant). The improvement brought by the current
work is the capability to import and use the static mesh stiffness as a function of the instantaneous values
of torque. In this paper, first the multibody model adopted for the gear system is described, then the Static
Transmission Error (STE) and the Dynamic Transmission Error (DTE) are defined. Subsequently, the
static mesh stiffness sensitivity to the main assembly errors is evaluated, in order to identify the most
influent ones. A description of the new technique to consider the variable torque follows. Finally, the
obtained results are discussed and compared to the ones obtained with the previous technique and the
static GCAS values.

2 Multibody simulation of gear noise

2.1 System modeling

Since the contact analysis is captured in the instantaneous static mesh stiffness by GCAS, including three-
dimensional teeth microgeometric modifications and manufacturing errors, teeth global and contact
stiffness, shaft deflections and assembly errors, the gear system can be modeled as rigid in the multibody
environment, achieving a good computational efficiency. The term “static” mesh stiffness indicates that
GCAS calculation is based on the assumption that the gears reach the equilibrium under static torque.
Nevertheless the static mesh stiffness is variable along the mesh cycle, for example due to a different
number of contacting tooth pairs or due to a profile modification. With more detail, GCAS returns as an
output the STE, which is defined along the line of action as the difference between the real and the ideal
displacement of the driven gear (Figure 1).

Figure 1 : Transmission Error definition.

The static mesh stiffness can be easily obtained dividing the nominal contact force value by the static TE.
Since the mesh stiffness can be calculated as described, in every gear train, single gear pairs can be
identified and their meshing can be analyzed independently in GCAS. Once the STE is calculated by
GCAS for every meshing gear pair, a user-defined force element can then be defined for each pair in the
multibody model. Since this procedure is repeated for every gear pair, it will be described and discussed
hereafter for one example.
The analyzed gear pair has the specifications reported in Table 1 and is shown in Figure 2. Addendum and
dedendum are standard, and respectively equal to the module and 1.25 times the module. No
microgeometry is considered at this stage, since it does not add to the purpose of showing the procedure.
3036 P ROCEEDINGS OF ISMA2010 INCLUDING USD2010

Parameter name Value


Driving gear tooth number 20
Driven gear tooth number 20
Helix angle 0°
Pressure angle 20°
Module 4 mm
Facewidth 25 mm
Center line 80 mm
Contact ratio 1.557

Table 1: Gears specifications. Figure 2: Three-dimensional view of the


gearing in the multibody software.

The gears are rigidly connected to the shafts, 200 mm long, which are supported by compliant bushings.
The axial and radial stiffness are chosen to be the same for all the bushings and equal to 108 N/m, however
the multibody software enables to assign clearances and stiffness-displacement laws. A reasonable
damping coefficient value was adopted for all the bushings to stabilize the simulation. A time-dependent
angular position law is assigned to the driving gear. In particular, a regime run-up is performed from 0 to
955 rpm in 5 s, with a linear increase in angular velocity, at an angular acceleration of circa 20 rad/s2. A
constant resisting torque of 100 Nm is applied to the driven gear, causing a nominal contact force of 2660
N.
The user-defined force element governs the gear meshing, applying the instantaneous contact force to the
gears based on the instantaneous static mesh stiffness (imported from GCAS), a given value of damping,
and the instantaneous TE calculated by the multibody software, namely the DTE. The contact force is
applied at the operating pitch point, located at the intersection between the instantaneous gear center line
and the common tangent to the base circles, at half the teeth facewidth. The whole system can be
represented schematically as shown in Figure 3.

Figure 3: Schematic system representation.

Therefore the driving gear follows the imposed angular position law, while experiencing a variable load
due to the DTE. The driven gear, instead, is subject to a constant applied torque, while experiencing an
oscillating angular position due to the DTE. The dynamic excitation in terms of transmission error is
shown in Figure 4.
M ULTI - BODY DYNAMICS AND CONTROL 3037

The results of the multibody simulation are calculated solving the system of equations of motion, which
can be condensed like in Equation 1.

(1)
Where a dot in accent position indicates the time derivative, x is the vector of the Lagrangian coordinates,
M, C and K are respectively the mass, damping and stiffness matrices and F is the vector of the applied
loads. Referring to this formulation and recalling the definition of TE, the dynamic formulation of gear
meshing can be considered as the scalar Equation 2 which belongs to the vector Equation 1:

(2)
In the Equation 2 the inertial contribution is taken into account implicitly into the DTE, when resolving
the system of equations of motion.
Two aspects about this equation are worthwhile to be mentioned in order to explain how the dynamic
analysis is performed. The first is that stiffness k is the static mesh stiffness which is variable along the
mesh cycle and is imported from GCAS. The second is that, since the Equation 2 is part of the Equation 1,
the DTE and the contact force are both influenced by all the multibody model parts in terms of inertia,
damping and stiffness.
The bearing forces, which can be user later for an acoustic analysis of the gear train, are part of the
solution found for the Equation 1, hence they are available in the results of the multibody simulation.

2.2 Static and Dynamic Transmission Error

Based on the general definition of the TE (Figure 1), the STE, in output from GCAS, was considered to
calculate the static mesh stiffness and it was used in the multibody simulation to obtain the DTE, which
accounts for the system’s dynamics. The differences between STE and DTE are shown in Figure 4.

Figure 4: STE and DTE comparison for the analyzed gearing.

The STE (black curve in Figure 4) is higher when only one tooth pair is in contact, and lower when two
pairs mesh simultaneously, being the contact ratio between 1 and 2 (Table 1). This is the reason why the
STE has a square wave trend. The simulated response is typical of a second order system, and dynamic
oscillations are around the STE values as expected. When the angular velocity of the gears is increased,
the excitation frequency increases, and the delay in system response becomes more evident. This simple
case of a single gear pair is shown as an example, but the response becomes more complex when a gear
train is analyzed. The adopted multibody approach is the key to capture the complex dynamic behavior of
3038 P ROCEEDINGS OF ISMA2010 INCLUDING USD2010

the system, considering the mutual interactions between several gear pairs and their interactions with the
power source and the power user.
Vibrations generated during meshing are one of the main causes of gear noise, and the TE is their main
excitation [10]-[11]. In particular, the peak to peak value of TE can be used as an indicator of the TE
variability. A constant TE, in fact, would not cause vibrations nor noise.

3 TE sensitivity to the main positioning errors

Errors affect the relative positioning of the meshing teeth. These errors affect the peak to peak TE in
different extent. They originate from assembly and tooth generation errors, and shaft deflections. The most
common errors in parallel gears are mainly located at the bearings with deviations from the ideal
positions, clearances and deflections, which affect the shafts orientation and therefore the teeth relative
positioning. Manufacturing errors can be variable from one tooth to the other, in this paper this variability
will not be addressed. Under this assumption, manufacturing errors will be considered within the
description of meshing teeth relative positioning.
Looking at the meshing tooth faces, five possible relative displacements (the allowed rotation of the gears
is of course excluded) can be identified in space to describe the relative teeth positioning. In order to
understand their effects on the TE it is useful to appropriately choose the reference system to define these
linear and rotational displacements (Figure 5). Since the TE is defined along the Line Of Action (LOA),
the first axis is chosen in this direction. The second axis is chosen along the orthogonal direction to the
line of action in a radial plane (Offline LOA). The third axis is along the rotation axis direction. In this
reference system it is possible to identify the Plane Of Action (POA), which contains the LOA and is
parallel to the rotation axis, and the Offline Plane of Action (OPOA), which contains the OLOA and is
parallel to the rotation axis.

Figure 5: Reference system and nomenclature for defining positioning errors.

An axial displacement results in a reduction of the effective facewidth. Good positioning accuracies are
also normally achieved in the axial direction. For that reason axial displacement will not be considered in
the present analysis. A uniform displacement along the LOA coincides with a constant additional TE,
which does not alter the peak to peak TE value. For that reason LOA displacement will not be considered
in the present analysis. A rotation in the POA alters the load distribution along the facewidth of the teeth,
thus affecting the mesh stiffness and so the peak to peak TE. A uniform displacement along the OLOA
can be considered as a center line variation which affects the contact ratio and so the peak to peak TE. A
rotation in the OPOA is found to be ineffective on the contact area [32] and the tooth load distribution [2],
so it is deemed to be ineffective also on the TE.
M ULTI - BODY DYNAMICS AND CONTROL 3039

On the basis of the above considerations, the effects of rotations in the POA, commonly called (angular)
misalignment, and variations in center line will be assessed hereafter by means of GCAS analyses. This is
useful to decide if both the assembly errors have to be included in the dynamic analysis or one of the two
is predominant.

3.1 Misalignment and Center Line variations

Two gear pairs without microgeometric modifications, belonging to an automotive and a wind turbine
gearboxes, have been considered (Table 2).

Parameter name Automotive Wind turbine


Driving gear tooth number 39 40
Driven gear tooth number 40 80
Helix angle 26° 0°
Pressure angle 20° 20°
Module 3 mm 4 mm
Facewidth 25 mm 80 mm
Center line 134.5 mm 240 mm
Nominal contact force 6000 N 13300 N
Table 2: Automotive and Wind turbine gear pairs specifications.

Realistic values of the center line variation are estimated in a range of ±0.2% of the whole center line, as
confirmed by the values used in [2].
Since misalignment is a rotation in the POA, it can be defined in terms of a rotation angle, the slope
associated to this angle, or the displacement caused by the rotation at a tooth face (Figure 6).

Figure 6: Schematic representation of misalignment and definitions


in terms of angle (α), displacement (δ) and slope (m).

Misalignment is affected by parameters such as shaft deflections, bearing positioning and clearances
which cannot be assumed a priori. However the 1328 ISO Standard for Gear Quality provides the allowed
ranges for slope errors along the lead direction. The slope error value has been used to estimate the range
of misalignments for the considered gear pairs. Assuming an average gear quality (ISO 7), the allowed
slope deviation is found to be for both gear pairs equal to 20 µm in terms of displacement. This value has
been doubled because the slope deviation can be the maximum and add up for both the meshing teeth. The
contribution due to shafts and bearings is then assumed to have the same magnitude of the slope deviation,
so that the final misalignment is estimated to be in a range from 0 to 60 µm in terms of displacement.
The peak to peak STE percentage variation has been evaluated for center line variations and
misalignments within the defined ranges (Figure 7).
3040 P ROCEEDINGS OF ISMA2010 INCLUDING USD2010

Figure 7: Effects on peak to peak STE value due to misalignment and center line variation.

The peak to peak STE value increases for both the gear pairs if misalignment or center line are increased.
The maximum increase of the peak to peak STE caused by the misalignment is about 100% for the wind
turbine gear pair and 600% for the automotive gearbox gear pair. Considering the center line variation, the
maximum increase is below 15% for both the gear pairs. The gear pair from the automotive gearbox,
which is helical, shows a higher sensitivity to both misalignment and center line variation. Considering the
maximum values, the misalignment can be considered as dominant with respect to the center line
variation. Although, at low misalignment values, the center line variation effects prevail on misalignment,
especially for the wind turbine gear pair. It has also to be pointed out that the sensitivity to misalignment is
increased by the absence of lead modifications, which are usually adopted when misalignments are
significant as in this case. Based on these considerations it is worthwhile to extend the technique to take
into account both the dynamic misalignment and the center line variations.

4 Variable torque modeling

Up to now, the static mesh stiffness has been calculated by the user-defined force element dividing the
nominal contact force by the STE obtained from GCAS as a function of the position along the mesh cycle.
Therefore the nominal contact force, which has essentially the same meaning of the applied torque since
the base radius is a constant, is assumed to be known and constant. The implication is that the effects of a
variable applied torque are not included in the dynamic simulation. As a consequence, the unequal load
sharing in multi-mesh gear trains (e.g. planetary stages [4, 7, 8]), which is a key point in their design,
cannot be considered. The procedure described in the next paragraph has been implemented to overcome
this current issue.

4.1 Mathematical modeling

The implemented procedure is based on the assumption that for a given position along the mesh cycle
(abbreviated with POS), which is a function of the instantaneous angular position of the driving gear, there
is a one-to-one correspondence between the instantaneous DTE and the instantaneous contact force
(higher DTE for a higher contact force, here the damping contribution is not accounted for). Namely,
given a POS and the actual DTE, the actual contact force can be calculated as a function of two variables
(Equation 3). If this assumption is verified, the procedure can be used in the multibody software to
calculate the instantaneous applied contact force.

, (3)
M ULTI - BODY DYNAMICS AND CONTROL 3041

This two-variables function is described by a database of STE values calculated through GCAS
considering different working conditions. In particular, the database is constituted by a set of STE curves
calculated for a range of contact forces (resulting from a range of applied torques). In this way, given a
POS, the STE is available as a function of the applied force (Figure 8).
The instantaneous DTE is used in place of the STE to enter the database, together with the instantaneous
POS, during the multibody simulation, so that the instantaneous contact force can be calculated. This
operation will be mentioned as “extraction”. The properties of the extracted contact force are discussed in
the next paragraph. Keeping the focus on the mathematical aspects of the procedure, the one-to-one
correspondence between applied contact force and STE is verified with the graphs in Figure 9. The one-
to-one correspondence was also verified applying a misalignment and microgeometric modifications (lead
and profile).

Figure 8: STE calculated through Figure 9: Applied force as a function of STE.


GCAS as a function of POS and (sections of the left figure surface
the applied contact force. in the Fcontact - STE plane)

It has to be noted that GCAS calculates the STE in a discrete number of POS and applied torque values,
therefore the database is discrete and requires interpolation (Figure 10).

Figure 10: Interpolation steps for contact force extraction.

Smooth interpolation (C1 continuity) is needed to avoid inducing stability problems for the multibody
solver, therefore a simple linear interpolation cannot be used. Cubic spline interpolation ensures the C1
3042 P ROCEEDINGS OF ISMA2010 INCLUDING USD2010

continuity but produces overshoots when rapid variations in STE values happen (e.g. variations in square
wave trend of Figure 4). To avoid this problem, quadratic or Akima interpolating schemes were adopted.

4.2 Multibody and dynamic considerations

An important consideration is needed about the properties of the extracted contact force. The dynamic TE
is used to enter a static TE database (Figure 11).

Figure 11: Contact force extraction procedure.

This implies that the extracted contact force is equal to the static part of the dynamic contact force. From a
mathematical point of view, the DTE time derivative which appears in the Equation 2 goes to zero. This
force is applied by the user-defined force element to the gears at the pitch point, and then the solver
includes the contribution of damping and inertia.

5 Numerical results using the proposed methodology

In this paragraph the results obtained from the simulations are shown and discussed. The simulations are
run for the gear pair described in paragraph 2.1.
The radial bearing forces for one bushing are shown in Figure 12 in a time window spanning from 4.5 s to
4.75 s during the run-up simulation.

Figure 12: Radial bearing forces in a time window from 4.5s to 4.75s during the run-up simulation.

The regimes corresponding to these values of time are respectively 860 rpm and 907 rpm, namely
14.33 Hz and 15.12 Hz in terms of angular frequencies. Multiplying the angular frequencies by the number
of gear teeth (Table 1), the meshing frequencies are 287 Hz and 303 Hz. The bearing forces spectrum
calculated for the considered time window shows the fundamental frequency, correctly located between
the two extreme meshing frequencies, and its harmonics (Figure 13).
M ULTI - BODY DYNAMICS AND CONTROL 3043

Figure 13: Radial bearing forces spectrum in a time window from 4.5s to 4.75s
during the run-up simulation.

The simulation time to execute the run-up described in paragraph 2.1 was of 307 seconds for the constant
applied torque technique and 419 seconds when taking into account the variable applied torque (CPU:
Intel Core Duo 2.0GHz, RAM: 2 GB @667MHz). This good computational efficiency is an important
feature of the proposed methodology.
The accuracy of the results has been evaluated numerically, comparing the dynamic results obtained by
the two simulation techniques and the ones statically predicted by GCAS.

5.1 Variable torque

To verify the variable torque technique, a constant torque is applied to the driven gear shaft and the
obtained results are compared with the ones obtained using the constant torque technique. The results,
shown in the next graphs, match each other while the variable torque technique relaxes the a-priori
assumption of a constant torque. This proves that the extracted static contact force is actually the static
component of the dynamic contact force and is a consistent method to excite the system.
The simulated DTE obtained using the variable torque technique (contact force extraction) perfectly traces
the one from the constant torque technique (Figure 14).

Figure 14: DTE with constant and variable torque techniques for different regimes.

Also the dynamic contact forces overlap for the two techniques (Figure 15). The dynamic contact force
oscillates around the value of the nominal applied contact force, which is equal to 2660 N. The extracted
static contact force shows impulses at tooth pair handovers due to system response delay (peaks in Figure
15). In particular, the impulse forces are higher than the nominal contact force when the second tooth pair
comes into contact, since the mesh stiffness increases suddenly while the DTE is still high (see Figure 4)
because of system response delay. The opposite happens when the second tooth pair leaves contact, since
the mesh stiffness suddenly decreases while the DTE is still low.
3044 P ROCEEDINGS OF ISMA2010 INCLUDING USD2010

Figure 15: Dynamic contact force with constant (smooth line) and variable (marked line) torque
techniques; static contact force with variable torque technique (dashed line).

6 Conclusions

In this paper a technique which enables to calculate the bearing forces aimed at the acoustic analysis of a
gearbox has been proposed. This technique integrates a multibody software (LMS Virtual.Lab Motion)
and a specialized software for gear meshing analysis (GCAS) to simulate gear dynamics (DTE and contact
force) taking into account detailed tooth contact. In particular, the multibody analysis allows the time-
domain integration of the solution, which captures the nonlinear effects of bearing stiffness and
clearances, gear backlash, large rotations and other nonlinear phenomena, while the specialized software
allows to include the three-dimensional tooth microgeometry, global and local tooth stiffness and relative
positioning of the meshing teeth. The proposed technique enables to perform run-up analysis with good
computational efficiency and the obtained results agree, in terms of TE and contact force, with the static
ones predicted by GCAS. Since the effects of a variable torque (or contact force) are part of the solution,
the proposed technique can also be used to predict the unequal load sharing in multi-mesh gear trains.
The technique is currently being improved to take into account the effects of center line variations, the
dynamic misalignment and the variable position of the contact force point of application.
Future work includes the validation of the proposed method against experimental data.

References

[1] X. Su, C. H. Menq, D. R. Houser, Estimation of Reference Misalignment of a Spur Gear and its
Application to Profile and Lead Measurement, Journal of Manufacturing Science and Engineering,
Vol. 124 (2002), p. 333.
[2] S. Li, Effects of machining errors, assembly errors and tooth modifications on loading capacity,
load-sharing ratio and transmission error of a pair of spur gears, Mechanism and Machine Theory,
Vol. 42 (2007), p. 698.
[3] G. Bonori, F. Pellicano, Non-smooth dynamics of spur gears with manufacturing errors, Journal of
Sound and Vibration, Vol. 306 (2007), p. 271.
M ULTI - BODY DYNAMICS AND CONTROL 3045

[4] F. Karpat, S. Ekwaro-Osire, K. Cavdar, F. C. Babalik, Dynamic analysis of involute spur gears with
asymmetric teeth, International Journal of Mechanical Sciences, Vol. 50 (2008), p. 1598.
[5] R. Y. Yakoub, M. Corrado, A. Forcelli, T. Pappalardo, S. Dutre, Prediction of system-level gear
rattle using multibody and vibro-acoustic techniques, SAE Transactions, Vol. 113 (2004), p. 2042.
[6] A. Singh, Load sharing behavior in epicyclic gears: Physical explanation and generalized
formulation, Mechanism and Machine Theory, Vol. 45 (2010), p. 511.
[7] A. Singh, A. Kahraman, H. Ligata, Internal Gear Strains and Load Sharing in Planetary
Transmissions: Model and Experiments, Journal of Mechanical Design, Vol. 130 (2008), 072602.
[8] A. Bodas, A. Kahraman, Influence of Carrier and Gear Manufacturing Errors on the Static Load
Sharing Behavior of Planetary Gear Sets, JSME International Journal Series C, Vol. 47 (2004), p.
908.
[9] R. G. Parker, V. Agashe, S. M. Vijayakar, Dynamic response of a planetary gear system using a
finite element/contact mechanics model, Journal of Mechanical Design, Vol. 122 (2000), p. 304.
[10] J.D. Smith, Gear noise and vibration, Marcel Dekker, Cambridge (2003).
[11] F. L. Litvin, A. Fuentes, Gear geometry and applied theory, Cambridge University Press, Cambridge
(2004).
[12] R. G. Munro, N. Yildirim, Some measurements of static and dynamic transmission errors of spur
gears, Proceedings of the International Gearing Conference, University of Newcastle upon Tyne,
(1994), p. 371.
[13] S.L. Harris, Dynamic loads on the teeth of spur gears, Proceedings of the Institution of Mechanical
Engineers, Vol. 172 (1958), p. 87.
[14] H. Walker, Gear Tooth Deflection and Profile Modification, The Engineer, Vol. 166 (1938), p. 434.
[15] D. W. Dudley, Modification of Gear Tooth Profiles, Product Engineering (1949), p. 126–131.
[16] R. G. Munro, Optimum profile relief and transmission error in spur gears, Proceedings of the First
Institution of Mechanical Engineers International Conference on Gearbox Noise and Vibration,
Cambridge (1990), p. 35.
[17] Y. Kanda, K. Hatamura, S. Kumano, K. Fujii, T. Kurisu, Y. Miyauchi, Analytical approach of
transmission gear noise, JSME International Conference on Motion and Power Transmissions,
(2001), p. 126.
[18] R.W. Gregory, S.L. Harris, R.G. Munro, Dynamic behaviour of spur gears, Proceedings of the
Institution of Mechanical Engineers, Vol. 178 (1963), p. 261.
[19] K. Umezawa, T. Suzuki, H. Houjou, T. Sato, Vibration of power transmission helical gears (The
effect of contact ratio on the vibration), Bulletin of JSME, Vol.28 (1985), No. 238, p. 694.
[20] K. Umezawa, T. Suzuki, T. Sato, Vibration of power transmission helical gears (approximate
equation of tooth stiffness), Bulletin of JSME, Vol.29 (1986), No. 251, p. 1605.
[21] K. Umezawa, J. Ishikawa, Deflection due to contact between gear teeth with finite width, Bulletin of
JSME, Vol. 16 (1973), No. 97, p.1085.
[22] K. Umezawa, The meshing test on helical gears under load transmission (3rd report: the static
behaviours of driven gear), Bulletin of JSME, Vol. 17 (1974), No. 112, p.1348.
[23] Y. Cai, Simulation on the rotational vibration of helical gears in consideration of the tooth
separation phenomenon (a new stiffness function of helical involute tooth pair), Journal of
Mechanical Design, Vol. 117 (1995), p. 460.
[24] Y. Cai, T. Hayashi, The linear approximated equation of vibration of a pair of spur gears (theory
and experiment), Journal of Mechanical Design, Vol. 116 (1994), p. 558.
[25] P. Velex, M. Ajmi, On the modelling of excitations in geared systems by transmission errors, Journal
of Sound and Vibration, Vol. 290 (2006), p. 882.
3046 P ROCEEDINGS OF ISMA2010 INCLUDING USD2010

[26] V. Abousleiman, P. Velex, A hybrid 3D finite element/lumped parameter model for quasi-static and
dynamic analyses of planetary/epicyclic gear sets, Mechanism and Machine Theory, Vol. 41 (2006),
p. 725.
[27] A. Andersson, L. Vedmar, A dynamic model to determine vibrations in involute helical gears,
Journal of Sound and Vibration, Vol. 260 (2003), p. 195.
[28] M. Kubur, A. Kahraman, D. M. Zini, K. Kienzle, Dynamic Analysis of a Multi-Shaft Helical Gear
Transmission by Finite Elements: Model and Experiment, Journal of Vibration and Acoustics, Vol.
126 (2004), p. 398
[29] T. Lin, H. Ou, R. Li, A finite element method for 3D static and dynamic contact/impact analysis of
gear drives, Computer Methods in Applied Mechanics and Engineering, Vol. 196 (2007), p. 1716.
[30] S. Vijayakar, A Combined Surface Integral and Finite Element Solution for a Three-Dimensional
Contact Problem, International Journal for Numerical Methods in Engineering, Vol. 31 (1991), pp.
525.
[31] J. A. Morgan, M. R. Dhulipudi, R. Y. Yakoub, A. D. Lewis, Gear mesh excitation models for
assessing gear rattle and gear whine of torque transmission systems with planetary gear sets, SAE
2007 Noise and Vibration Conference and Exhibition.
[32] D.R. Houser, J. Harianto, D. Talbot, Gear mesh misalignment, Gear solutions, June 2006 issue, p. 34.

View publication stats

You might also like