Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

Journal of Sound and Vibration 331 (2012) 45544574

Contents lists available at SciVerse ScienceDirect

Journal of Sound and Vibration


journal homepage: www.elsevier.com/locate/jsvi

A dynamic model to study the inuence of planet position errors in planetary gears
X. Gu, P. Velex n
Universite de Lyon, INSA de Lyon, LaMCoS, UMR CNRS 5259, France

a r t i c l e in f o
Article history: Received 7 September 2011 Received in revised form 5 March 2012 Accepted 6 May 2012 Handling Editor: L.N. Virgin Available online 5 June 2012

abstract
An original lumped parameter model of planetary gears is presented which accounts for planet position errors and simulates their contribution to the quasi-static and dynamic load sharing amongst the planets. A unique feature of the model is that instantaneous gear geometry is used which depends on the combination of deections and errors. The numerical results compare well with experimental evidence from the literature thus validating the modelling methodology. Simulations are then extended to dynamic tooth loads and trajectories for both xed and rotating carriers. Finally, the advantages and drawbacks of oating members in high-speed planetary gears are illustrated and commented upon. & 2012 Elsevier Ltd. All rights reserved.

1. Introduction Planetary gears offer several advantages such as co-axial shafts, high power density and efciency which make them widely used in automotive and aeronautical applications (turboprop engines, helicopter transmissions, etc.). Gear vibrations are primary concerns in these applications both in terms of dynamic forces and noise since dynamic loads increase the risk of tooth or bearing failures whereas noise can become a sensitive issue in helicopters for instance. A number of papers have been published on planetary gear dynamics which comprise lumped-parameter models [110] and deformable or hybrid models of varying complexity [1116]. Modal analyses were performed by Saada and Velex [8], Parker et al. [1720] who emphasized the structured modal properties of single-stage drives and showed, for equally and unequally-spaced planets, that only planet, rotational and translational modes could exist. The studies by Cunliffe et al. [1], August and Kasuba [2], Hidaka et al. [37], Kahraman [9], Velex et al. [8], [10] involve planetary gear models to determine critical frequencies and dynamic responses. Hidaka et al. [6] found that the phases between the planets could modify sungear trajectories and several studies (Seager [21], Parker [22], Parker et al. [23,24]) conrmed the contribution of mesh phasing as well as the possibility of cancelling some frequency components of the global translational or torsional excitations on sun-gears and ring-gears. In planetary gears under ideal conditions, each path will carry an equal amount of load but it has been shown that, due to the presence of manufacturing errors, such an equal sharing of the input torque between the parallel paths is hardly realised in practical applications. The particular contributions of geometry or mounting errors on gear dynamics have been tackled by Botman and Toda [25] who studied the inuence of planet run-out errors and found signicant differences depending on planet indexing. In the continuation of this work, Ma and Botman [26] introduced prole and run-out deviations in a 2D model to investigate planet load sharing. Hybrid nite element analyses were performed by Cheon and

Corresponding author. E-mail address: Philippe.Velex@insa-lyon.fr (P. Velex).

0022-460X/$ - see front matter & 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jsv.2012.05.007

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

4555

Nomenclature damping matrix radial and tangential position errors of planet j F gj t,X gj elastic contributions to mesh j force caused by initial separations FKE total elastic contribution to mesh forces caused by initial separations F0 constant force vector (torque) IGi polar moment of inertia of solid i Kb total bearing stiffness matrix K G t,X global time-varying possibly nonlinear mesh stiffness matrix kgj t,X gj time-varying possibly nonlinear mesh stiffness function associated with mesh j K gj t,X gj time-varying possibly nonlinear mesh stiffness matrix associated with mesh j kkx,kky,kky bearing stiffness elements (two directions of translation, torsion) K averaged global stiffness matrix M global mass matrix mi mass of solid i n nj ,nn =n0j ,nj 0 constant/ variable outward unit normal j vector to sun gear and planet j tooth ank RaS,Raj addendum circle of sun gear and planet j RtR addendum circle of ring gear RbS,RbR,Rbj base radius of sun gear, ring gear and planet j RC ,Rgj ,R0gj radius of carrier, constant / variable center distance of sun/ring gear-planet j X total DOF vector (3N 9) components, N: number of planets X(Mgj),Xt Dt(Mgj) X-coordinate of any potential point of contact M gj X kj elastic displacement vector of sun / ring / carrier-planet j Un see Eq. (6) j uR Ok ,uC Oj translational displacement vector of j k solids relative to rigid-body motions uY1 Ok ,uYj Oj displacement vector of solid k and j k planet j relative to the rotating frames of fSY1 g,fSYj g j k n R uk M=M ,uC M=M n translational displacement vecj tor at point M/M* relative to rigid-body motions Wgj structural vector (dependent on gear geometry) C exj,eyj

V j M=M n relative velocities of sun/ring gear-planet j S=R at the point of contact Mor M* X gj elastic displacement vector associated with sun/ring gear-planet j xi,yi translational degrees of freedom in xi and yi directions agj , a0gj pressure angle with constant and variable gear geometry bb base helix angle dM0gj Deection at any potential point of contact M0gj n n dj M general expression of the separation at any point of contact M* n dej , dej initial separations caused by position errors of sun/planet j, of ring/planet j DM0gj contact deection at M0gj eS 1 for sun gear counter-clockwise rotation, eS 1otherwise z 1 if ring-gear is xed, z 0 if the carrier is xed yi torsional degree of freedom of solid i li initial position angle of eccentricity (i C, S, R, j ) Fj , F0gj planet spacing angle for constant and variable geometries jj,fk additional rigid-body angle induced by errors for planet j and for central members ci no-error rotational angle with respect to the carrier rotary frame ci Oi0t li, i C, S, R, j 2OC D, O2 L gyroscopic matrix and centrifugal stiffness C matrix induced by carrier rotation. 0 Oi angular velocity vector with respect to the inertial frame xR , xC angular displacement vector of solids with j k respect to rigid-body motions xY1 , xYj angular displacement vector of solids with j k respect to reference frame AB vector rooted at point A and head at point B 99,:: vector amplitude, remainder of natural division Indices: Subscripts and superscripts h i g s, r, c : sun-gear, ring-gear and carrier s, r, c, j : sun-gear, ring-gear, carrier and planet s, r: sun-gear, ring-gear

Parker [27,28] to characterise the effects of some manufacturing errors on dynamic bearing and tooth loads. Inalpolat and Kahraman [29,30] described the expected patterns of modulation sidebands using a simplied mathematical model which qualitatively accounts for errors and compared with the measurements from several experimental planetary gears. Using an instrumented planetary gear with tightly controlled intentional manufacturing errors operated under realistic load and speed conditions, Ligata et al. [31,32] presented a number of experimental results at low speed on the inuence of carrier pinhole position errors on planet load sharing showing the prominent inuence of tangential planet position errors. The experimental data compared well with the predictions of a multi-body contact analysis model [33,34] but also with the dimensionless closed form expressions obtained by Singh [35,36] valid for any error and torque as long as quasi-static conditions prevail.

4556

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

In this paper, an original lumped parameter model is set up which makes it possible to capture the quasi-static and dynamic load sharing characteristics in basic types of planetary gears with xed and rotating carriers. For the sake of simplicity, the sun-gear is supposed to be the input member and only planet position errors are considered. The inuence of deections on instantaneous gear mesh properties is incorporated leading to amplitude, frequency and phase modulated mesh stiffness functions depending on the errors. The resulting state equations point to a non-linear parametrically excited differential system which is solved iteratively by combining a time-step integration scheme, a xed-point method and a normal contact algorithm. A number of results are presented which prove that the model is representative of actual planetary gear behaviour and illustrate the contributions of position errors and the interest of oating members in balancing planet loads. 2. State of referenceRigid-body motions Following [14,15], rigid-body motions dene the state of reference in the vicinity of which the degrees-of-freedom are dened. For multi-contact closed loop systems, the presence of errors induces particularities since contact can be temporarily lost at some points normally connected for errorless conditions. Considering the situation at one given time t, the deviations with respect to errorless conditions can be simulated via screws of innitesimal generalised displacements attributed to every solid k. For the sun-gear (k S), carrier (kC) and ring-gear (kR), their coordinates are: ( R uk Ok 0 fSR g (1) k xR fk z0,k k where superscript R indicates that the perturbations are measured from the errorless conguration of solid k and fk is the additional angular rotation of solid k possibly induced by errors. The corresponding geometry and the associated frames of reference are represented in Fig. 1. For planets (subscript j, j 1, y , N), errors are dened with respect to the carrier and radial and tangential position errors are characterised by the following screw coordinates: 8 < uC Oj exj xj eyj yj j C fSj g (2) : xC jj z0,j j where exj and eyj are the radial and tangential position errors and jj is the possible additional rigid-body angle induced by errors. The total perturbations from errorless conditions result from the combinations of the planet and carrier perturbations as: fSR g fSC g fSR g j j C (3)

Planet #j
yj
Oj
Oj 0

j
xj

eyj exj

j
O0

Z0

x1

Carrier
Fig. 1. Frame denition and position errors of planet j.

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

4557

In what follows, it will be supposed that the sun-gear is the driving member and that its rigid-body angular speed is constant (the additional angle fS is therefore nil). Similarly, the angle fk associated with the reaction member, either the ring-gear or the carrier, will be set to zero (xed member). In the analysis of rigid-body motion transfer, it is further assumed that there is always contact at all the sun-gear / planet interfaces regardless of geometry errors implying that there is always a combination of additional small rigid-body angles on planets (jj) , carrier (fC) or ring-gear (fR) such that contacts are re-established on the sun-gear/planet base planes. The corresponding contact conditions can be formulated as: uR MuR M : nj uR OS uC MuR M:nj 0 S S C j j which, using the shifting properties of screw moments, leads to: uR OS uC Oj zuR OC z fC z0,j OC Oj jj z fC z0,j Oj M :nj 0 S C j
-

(4-1)

(4-2)

with nj, a unit normal vector at the point of contact M on the sun-gear/ jth planet base plane. z 1 if the ring-gear is xed and z 0 if the carrier is xed. Focusing on planet / ring gear interfaces, the situation is different to that on sun-gear/planet base planes since, because of deviations and the additional angles jj required to ensure contacts at the sun-gear / planet meshes, contacts between planets and ring gear may be lost. The solution technique consists in nding the base plane(s) with the minimum initial separation and determining the additional angle on either the ring-gear (jR) or carrier (jC) depending on what the reaction member is, to re-establish contact at one planet / ring-gear mesh at least. The general expression of the separation at any point of contact M* is expressed as:

dn Mn 1zuR Mn uR Mn :nn R j j j
- 1 1zuR OR 1z fR z0,j OR M n uC Oj jj z fC z0,j Oj Mn C R j B @ A:nn j zuR OC z fC z0,j OC Oj C

(5)

where * stands for points and vectors associated with planet / ring-gear meshes. Solving for jj zfC from (4-2), keeping rst order terms only and injecting in (5) nally gives:

dn Mn dn U n 1zes RbR cosbb fR zes cosbb RC cosaSj cosaRj fC j j j


With;es 1 for counter clockwise sun-gear rotation; es 1, otherwise. U n 1zuR OR :nn uR OS :nj uC Oj zuR OC :nj -nn R S C j j j j es cosbb exj sinaSj sinaRj es eyj cosaSj cosaRj

(6)

The various angles in the expression above are dened in the nomenclature. n Note that dj is necessarily positive or nil if nn is the outer unit normal vector with respect to the planet tooth anks (the j n n condition dj o 0 would correspond to interpenetration of the parts) and, for one given base plane or mesh, dj is independent of the position of the point of contact M*. The angle on the carrier or the ring-gear to ensure contact at one ring-gear/planet mesh (at least) is the smallest possible value of either fR or fC whose expression is:

fR minj U n =es RbR cosbb if z O j


or

(7-1)

fC minj U n =es cosbb RC cosaSj cosaRj if z 1 j

(7-2)

Finally, the initial normal separations between the teeth of planet j and those of the sun-gear and ring-gear are given by:

dej 0

for sungear=planetmeshes for planet=ringgearmeshes

(8-1) (8-2)

den dn minj dn j j j

At every non-singular point of contact, the fundamental property of rigid-body kinematics imposes: Vj M:nj 0 S Vj M n :nn 0 j R (9-1) (9-2)

The positions of the contacting points M and M* correspond to those where the initial separations dej and den in (8-1) j and (8-2) are nil.

4558

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

3. Planetary gear dynamic model 3.1. Degree-of-freedom (DOF) denition A number of sophisticated 3D models have recently been presented which enable accurate predictions of load sharing [1315], [33], [37,38] or dynamic response for errorless systems [16]. In this paper, a simplied approach has been preferred in order to focus on the physical behaviour and generate extensive dynamic results. Based on the literature results, it appears that a lumped-parameter model combining torsion and bending translations for all gears is an interesting compromise as it can cover a wide range of practical situations (oating members in particular) while remaining sufciently light in terms of modelling effort and computational times. The corresponding dynamic model is shown in Fig. 2; the dynamic displacements of all the parts are characterised by 3 innitesimal angles and translations (DOFs) superimposed on rigid-body motions (as dened in Section 2) which can be represented by displacement screws. Two different screw congurations are used: i) for the central members (sun-gear, carrier and ring-gear), the translations DOFs are expressed in a unique frame attached to the sun-gear/planet #1 centre line (conventional) whereas ii), for planet j, the translational DOFs are in the direction of the sun-gear / planet j centre-line and perpendicular to it. The vectorial coordinates of the screws read therefore: a) for central member k:

8 < uY1 O O O x x y y 0 k k k 1 k 1 Y1 k fSk g Y1 : yk z0 k

k S,R,C

(10)

b) for planet j: fSYj g j

8 > uYj O O - x x y y < j j0 Oj j j j j


j

> :

xYj yj z0 j

(11)

It is to be noted that, for a rotating carrier, the DOFs are dened in a rotating coordinate system thus leading to gyroscopic and centrifugal effects.

Fig. 2. Degree of freedom denition-Lumped parameter model.

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

4559

3.2. Mesh stiffness and mass matrices Using the developments in [8], [14,15], the deection at any potential point of contact can be expressed as the normal approach with respect to rigid-body positions minus the initial separation deM 0gj under the form:

dM0gj WT XSj deM0gj gj

(12)

where index g refers to either sun-gear or ring-gear (g S or R) and index j refers to planet #j and using the following expressions: 9 8 8 9 0 > > sinrS > xS > > > > > > > > > > e cosr 0 > >y > > > S > S> > > > > S > > > > > > > > > > e Rb >y > = < S S < S= XSj , WSj cosbb 0 > > sinaSj > xj > > > > > > > > > > > > > > eS cosaSj 0 > > yj > > > > > > > > > > > > > > > > ; ; : eS Rbj : yj >

rS 0 aSj 0 eS FSj 0 sS 0 aSj 0 eS FSj 0 eS cS


9 8 9 > > xR > > > > > > > > >y > > > R> > > > > > > > > > > >y > = < R= 0 0 XRj sj aSj eS cj WRj cosbb sina 0 > > > xj > Rj > > > > > > > > > > > > > eS cosaRj 0 > > yj > > > > > > > > > > > > > > > > ; : eS Rbj > : yj ; 8 0 > sinrR > > > e cosr 0 > S > R > > > e Rb < S R

rR 0 aRj 0 eS FR

sR 0 aRj 0 eS FR j 0 eS cR sn aRj 0 eS cj j
(13)

The symbols are dened in the nomenclature. deM0gj represents the equivalent normal deviations in rigid-body conditions for sun-gear / planet j (g S) and planet j / ring-gear meshes (gR) respectively. It is to be noticed that, when considering radial/tangential position errors, deM 0gj is constant for one given mesh which greatly simplies the analytical expressions of the forcing terms. The corresponding mesh stiffness matrices and mesh forcing terms are then deduced under the general form: Kgj t,Xgj kgj t,Xgj HDM gj 0 Wgj WT gj Fgj t,Xgj kgj t,Xgj HDMgj 0 deM0gj Wgj where the normal contact condition is introduced via the unit Heaviside function such that: ( 1, if DM0gj 40 HDM 0gj 0, if DM 0gj r 0 For the sake of simplicity, all the bearing-shaft assemblies connections, the following stiffness matrix is obtained: 2 1 0 RC sinFj 6 1 RC cosFj 6 6 6 h i R2 6 j C K C kbp 6 6 6 6 sym 4 (14) (15)

(16)

are supposed to be isotropic and for pin planet / carrier cosFj sinFj 0 1 sinFj cosFj RC 0 1 0 3

7 07 7 07 7 7 07 7 7 05 0

(17)

when the DOFs are arranged as XCj {xC,yC,yC,xj,yj,yj}. For the bearing-shaft systems supporting the central members, the stiffness matrices reduce to: Kkb diag kkx ,kky ,kky ,k C, S, R (18)

4560

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

Neglecting second-order terms, the dynamic sums and moments for a sun-gear / planet j or planet j / ring-gear pair lead to the following mass, gyroscopic, centrifugal matrices: 9 8 2 3> x k > 2 38 x k 9 > > > _ > mk 0 mk > > > > > > > _ > > > > > 6 7> y k > 6m 7> y k > > > mk 0 > > 6 7> 6 k 7> > > > > _ > > > > > 6 7> y > 6 7> y > = 6 7< k = 6 7< k > IGk 0 6 7 6 7 2O0 6 6 7 x 7 x C mj 0 mj 6 7> j > 6 7> _ j > > > > > > > 6 7> 6 7> > > > _ > 6 7> y j > 6 7> y j > mj mj 0 > > > > > > 4 5> 4 5> > > > > > > > _ > IGj > y j > > ; > : 0 > yj > ; : (19-1) 9 8 2 3 x > k > mk > > > > 6 7> yk > > > > mk > > 6 7> > > 6 7> > > yk > = 6 7< 0 7 0 26 OC 6 7 x mj 6 7> j > > > > 6 7> > > 6 7> yj > mj > > > 4 5> > > > > ; : 0 > yj > along with the inertial forcing terms: 8 >0 > > >0 > > > > > <0 9 > > > > > > > > > =

0 > mj OC 2 RC exj > > > > > > > > > > > mj O0 2 eyj > > C > > > > ; : 0

(19-2)

Using Newtons second law and assembling all the elemental matrices nally lead to a differential system of the form: _ MX 2OC D CX Kb KG t,X O2 LX F0 FX t FKE t _ C (20)

where M,Kb,KG(t,X) are the mass, bearing-shaft and mesh stiffness matrices, respectively. C a M b K is the damping matrix (Rayleighs damping determined by using the averaged global stiffness matrix K). 2OC D, O2 L represent the gyroscopic and centrifugal stiffness matrices generated by the carrier rotation. C X {xC,yC,yC,xR,yR,yR,xS,yS,yS,x1,y1,y1,...,xN,yN,yN} is the total DOF vector (3N 9 components, N is the number of planets). F0 ,FX t are the static load vector (torques) and the inertial forcing terms. _ FKE(t) represents the forcing term vector which stems from the initial separations between the teeth on all the base planes and results of the assembly of the elemental vectors dened in (15). 4. Mesh stiffness functions: Because of the inuence of position errors and deections, the non-linear mesh stiffness functions kgj(t,Xgj) cannot be determined prior to solving the equations of motion (20). The situation is illustrated in Fig. 3 for sun-gear/planet and planet/ring-gear meshes. At every time step, the generic problem can be formulated as follows: a) knowing the positions of the base circle centres (depending on errors and deections) along with the base and outer n radii, nd the orientations of the base planes (by dening unit vectors I0j and Ij 0 ) and deduce the limits of the meshing areas b) calculate the instantaneous positions of all the contact lines in all the base planes c) deduce the various mesh stiffness functions (amplitudes and phases). These various operations are detailed below: Point (a): Because of errors and elastic displacements, the centres of solid k move to O0 k and the actual apparent pressure angles a0 gj for solid g-solid j mesh are given by: cosa0gj Rbg 7 Rbj O0g O0j (21)

where is for sun-gear/planet whereas is for planet/ring-gear and the instant limits of the active meshing zones can be derived by seeking the intersections between the base planes and the outer radii or root and outer radii for

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

4561

I I*j

O O

Planet j

Ring gear

sj '
Os O0 Or

rj '

Sun gear

Fig. 3. Planetary gear geometry with varying parameters.

I j
addendum circle ring-gear

S E

I*

Ra
addendum circle sun-gear

Oj Esj
Ra

addendum circle planet #j

Ra

Ssj

Os

Or

Fig. 4. Limits of engagement on base planes.

I j Lrj L*j Mrj I*j

Lj

Oj

Msj Os Or Lsj

Fig. 5. Potential contact points on base planes.

planet / ring-gear meshes (Fig. 4). Similarly, the angular spacing between the various planets is evaluated based on the instant positions of the sun-gear, planet, carrier and ring-gear centres. Point (b): Once the planes of action are known, the positions of the contact lines are calculated by assuming that, conventionally, there is contact at the initial time between the sun-gear and planet 1 at one point at the limit of engagement (Fig. 5). From this

4562

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

position, all the other points in contact at the initial time can be determined following the technique described in [14]. A time-step scheme is then used which, by knowing the contact positions at t, makes it possible to determine the contact positions at t Dt. Considering a given base plane s, the X-coordinate of any potential point of contact Msj is given by: XM sj L0S M sj :I0j where L0S is a point of tangency on one base circle (see Fig. 5)
x 108 Mesh stiffness of sun-gear/planets (N/m)
sp1 sp2 sp1
-

(22)

6 5.8 5.6 5.4 5.2 5 4.8 0 0.5


sp-0

sp4

1.5

t/Tm (Tm: mesh period) x 108 Mesh stiffness of ring-gear/planets (N/m) 6 5.8 5.6 5.4 5.2 5 0 0.5 1 1.5 2
rp1 rp-0 rp3 rp2 rp4

t/Tm (Tm: mesh period)


Fig. 6. Examples of mesh stiffness functions in the presence of planet position errors with tangential position error ey1 10 4 m on planet #1, ey2 7.10 5 m on planet #2, and a radial position error ex1 5.10 5 m, sun gear rotational speed is 500 rpm, rotating carrier ( Planetary gear data: Table 1 - case 2). Keys: Errorless: s/rp-0 sun(ring)-gear/planets ; With error: s/rp1 sun(ring)-gear/planet #1, s/rp2 sun(ring)-gear/planet #2, s/rp3 sun(ring)-gear/planet #3, s/rp4 sun(ring)-gear/planet #4.

Table 1 Planetary gear data. Sun-gear Tooth number Case 1 (experiment) 73 Case 2 (planets are in phase) 72 1.81 Planet Ring-gear

20 23.04 25

Normal module [mm] Helix angle [1] Pressure angle [1] Addendum diameter [mm] Dedendum diameter [mm] Centre distance [mm] Active face width [mm]

139.7

26 26 1.81 51.6 92.12

20 23.04 25

125 124 1.81 229.4

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

4563

Table 2 Simulation ow chart.

0.6
N=2 N=4

0.5 Planet load ratios 0.4 0.3 0.2 0.1 0 100 200 300 400 500 600 Torque (Nm) 700 800 900 1000
P2,P4

average

P1,P3

Fig. 7. Static planet load sharing ratios versus sun gear torque Tangential position error ey1 70 mm on planet #1, sun gear rotational speed is 500 rpm, (N: planet number), Pj refers to planet #j, rotating carrier (Planetary gear data: Table 1 - case 1). Keys: Lumped parameter model: sun-gear/planet #1 sun-gear/planet, #2 sun-gear/planet, #3- - -sun-gear/planet #4 Experimental results: sun-gear/planet #1 , sun-gear/planet #2, sun-gear/planet, #3, n sun-gear/planet #4.

4564

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

Introducing innitesimal time-variations, the following rst order approximation is obtained:  -   -  d d L0S M sj :I0jt Dt L0S M sj : I0jt Dt X t Dt M sj X n M sj t dt t t dt Noticing that  -  L0S M sj and
t d 0 I dt jt

(23)

are always perpendicular, (23) is rewritten as: X t Dt Msj X t M sj


d d OI M sj t :I0jt Dt OI Ls :I0jt Dt dt dt

(24)

which, after some manipulations, leads to: _0 _ X t Dt M sj X t M sj eS OS RbS Dt eS RbS eS a 0Sj F Sj t Dt By using the same expansion, the positions of point on the planet / ring-gear base planes are given by: _0 _ X t Dt M rj X t M rj eS Oj Rbj Dt eS Rbj eS a 0Rj F Rj t Dt (26) (25)

From (25) and (26), it can be noticed that the motions of the contact lines in the base planes are modulated by the timevariations in the pressure angles and angular positions of the solid centres. At each time-step, the instant contact lengths are determined by calculating the total length of contact over a rectangular mesh area in the base planes whose edges are the (constant) face width and the time-varying active line of action. Physically

(a)
1.6 Max and min load ratios of sun-gear/planets

1.4

1.2

0.8

0.6

0.4

0.5

1.5

2 x 104

Rotational speed of sun gear (rpm)

(b)
Max and min load ratios of ring-gear/planets 1.6

1.4 1.2 1 0.8 0.6 0.4

0.5

1.5

2 x 104

Rotational speed of sun gear (rpm)

Fig. 8. Maximum load ratios versus sun gear speed, carrier is allowed to rotate, (Planetary gear data: Table 1 - case 2). Keys: Constant geometry: independent on initial condition Varying geometry: , initial condition1 fork line: static solution with averages stiffness , initial condition2 dotted line: steady-state response at previous speed.

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

4565

1.7

Mean contact ratios

1.65

1.6

1.55

1.5

0.5

1.5

2.5

Rotational speed of sun gear (rpm) x 104


Fig. 9. Mean contact ratio of sun(ring)-gear/planets versus rotational speeds of sun-gear. Keys: Sun-gear/planets Ring-gear/planets.

x 108 6.4 Mesh stiffness of sun-gear/planets 6.2 6 5.8 5.6 5.4 5.2 5 4.8 4.6 0 0.5 1 1.5 2 sp2 sp-0 sp1 sp3 sp4

t/Tm (Tm: mesh period)


Fig. 10. Mesh stiffness function with a tangential position error ey1 0.0002 m on planet 1, (sun-gear speed Os 11800 rpm, rotating ring-gear). Keys: Constant geometry: sp-0 sun-gear/planets without errors, Varying geometry: sp1 sun-gear/planet #1, sp2 sun-gear/planet #2,, sp3 sun-gear/planet #3, sp4 sun-gear/planet#4.

2.4 2.2 2 Max load ratio 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 Rotational speed of sun gear (rpm) x 104

2.4 2.2 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 Rotational speed of sun gear (rpm) x 104

Fig. 11. Maximum dynamic load ratio of sun gear/planets versus sun-gear speed (rotating carrier, tangential position error ey1 0.0002 m on planet #1, Planetary gear data: Table 1 - case 2). Keys: sun-gear/planet #1, sun-gear/planet #2, sun-gear/planet #3,  sun-gear/planet #4. (a) Constant parameters and (b) Varying parameters.

4566

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

3 2.5 Max load ratios 2 1.5 1 0.5 0

3 2.5 2 1.5 1 0.5 0

0.5

1.5 x

2 104

0.5

1.5 x

2 104

Rotational speed of sun gear (rpm)

Rotational speed of sun gear (rpm)

Fig. 12. Maximum dynamic load ratio of sun gear/planets versus sun-gear speed (rotating carrier, tangential position error ey1 0.0005 m on planet #1, Planetary gear data: Table 1 - case 2). Keys: sun-gear/planet #1 sun-gear/planet #2 sun-gear/planet #3  sun-gear/planet #4. (a) Constant geometry and (b) Varying geometry.

(a)
2.5 Max load ratio

1.5

1 2 1.5 x 104 1 0.5 0 -2 -1 0 ey1(m) 2 1 x 10-4

Rotational speed of sun gear (rpm)

(b)
2.5 Max load ratio

1.5

1 2 1.5 x 104 1 0.5 Rotational speed of sun gear (rpm) 0 -2 -1 0 ey1(m) 2 1 x 10-4

Fig. 13. Maximum dynamic load ratios of sun-gear/planet meshes versus sun-gear rotational speed and tangential error amplitude (rotating carrier, planetary gear data: Table 1 case 2). (a) Constant geometry and (b) Varying geometry.

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

4567

speaking, the contact lengths appear as functions of time modulated by the time-varying changes in geometry. Point (c): Finally, following [39], the mesh stiffness functions are derived by assuming that they are proportional to the instant contact lengths. Fig. 6 shows examples of mesh stiffness functions calculated by considering a variety of errors on the planetary gear set dened in Table 1 (case 1).

5. Numerical solution The numerical solution procedure is detailed in Table 2. It combines a time-step integration scheme (Newmark), a xed-point method with relaxation aimed at updating mesh geometry based on the DOFs and a unilateral normal condition which sets mesh stiffness to zero when contact is lost on the associated mesh.

(a)

x 10-4 C

x 10-4

Sun displacement in y direction (m)

Sun displacement in y direction (m)

-1.74

-1.752 -1.754 -1.756 -1.758 -1.76 -1.762 -6.73

-1.745 V -1.75

-1.755

-1.76 -6.7 -6.69 -6.68 -6.67 -6.66 -6.65 -6.64 Sun displacement in x direction (m) S = 200rpm x 10 Sun displacement in y direction (m) -1.796 -1.798 -1.8 -1.802 -1.804 -1.806 -1.808 -6.915 -6.91 -6.905 -6.9 -6.895 x 10-5 Sun displacement in x direction (m) S = 12000rpm V C
-4

V -6.72 -6.71 S = 5000rpm x 10


-4

-6.7

-6.69 x 10-5

x 10-5

Sun displacement in x direction (m)

Sun displacement in y direction (m)

-1.869 -1.87 -1.871 -1.872 -1.873 -1.874 -1.875 V -7.26 -7.24 -7.22 -7.2 -7.18 x 10-5 Sun displacement in x direction (m) S = 18000rpm C

Fig. 14. (a) Trajectory of sun-gear at various speeds (rotating carrier, tangential position error ey1 0.0002 m on planet #1). Keys: varying geometry: V constant geometry: C (b) Trajectory of planet #1 at various speeds (rotating carrier, tangential position error ey1 0.0002 m on planet #1). Keys: varying geometry: V constant geometry: C . (c) Trajectory of sun-gear at various speeds (xed carrier, tangential position error ey1 0.0002 m on planet #1). Keys: varying geometry: V constant geometry: C . (d) Trajectory of planet #1 at various speeds (xed carrier, tangential position error ey1 0.0002 m on planet #1). Keys: varying geometry: V constant geometry: C .

4568

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

4.146 Trajectory of planet #1 in y direciton 4.144 4.142 4.14 4.138 4.136 4.134

x 10-4 Trajectory of planet #1 in y direciton


C

4.155
V

x 10-4

4.15 4.145 4.14 4.135 4.13 4.125 1.06

C V

0.5 1 1.5 2 Trajectory of planet #1 in x direciton x 10-7 S = 200 rpm


-4

1.08 1.1 1.12 1.14 Trajectory of planet #1 in x direciton x 10-5 S = 5000 rpm

4.2 Trajectory of planet #1 in y direciton 4.18 4.16 4.14 4.12 4.1 4.08

x 10

4.11 Trajectory of planet #1 in y direciton 4.105

10-4

4.1 4.095 4.09 4.085 4.08 1.4


V

4.06 6.25

6.26

6.27 S = 12000 rpm

6.28

6.29 10-5

1.405

1.41 S = 18000 rpm

1.415

1.42

Trajectory of planet #1 in x direciton x

Trajectory of planet #1 in x direciton x 10-4

Fig. 14. (continued)

6. Results 6.1. Elements of validation [31,32] for quasi-static behaviour Based on the gear data for the 4-planet gear set in Table 1 (case 1), the results in Fig. 7 show the evolutions of the planet load sharing versus sun torque for a oating sun arrangement with a tangential error of 70 mm on one planet (i) as measured by Ligata et al. [31] and (ii) as calculated by using the proposed dynamic model at very low speed. It can be observed that, at low torque, only two planets are loaded and that, as the torque increases, all the planets come progressively into mesh. For this example, slight differences are noticed at the lowest torque which might be explained by the simplied mesh stiffness model but, generally speaking, an excellent agreement is obtained which validates the error model presented in Sections 2 and 3. A number of comparisons have been made with various planet numbers and error amplitudes in [41] which all conrm that the proposed modelling is sound. Finally, it can be observed that, even if it is assumed that all planets are in contact with the sun-gear in the state of reference, the model actually captures the contact losses caused by errors (for example, in Fig. 7, only two planets withstand the load at the lower torque amplitudes).

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

4569

(c) -1.738
Trajectory of sun gear in y direction

x 10-4

-1.741
Trajectory of sun gear in y direction

x 10-4

-1.74 -1.742 -1.744 -1.746 -1.748 -1.75 -6.74 -6.72 S = 200 rpm x 10-4 -6.7 x 10-5 V C

-1.742 C -1.743

-1.744

-1.745 V -6.74 -6.73 -6.72 -6.71 -6.7 -6.69 x 10-5

Trajectory of sun gear in x direction

Trajectory of sun gear in x direction S = 5000 rpm x 10-4 -1.742

Trajectory of sun gear in y direction

Trajectory of sun gear in y direction

-1.742 C -1.743

-1.7425 -1.743 -1.7435 -1.744 -1.7445 -1.745 V -6.73 -6.72 -6.71 -6.7 Trajectory of sun gear in x direction S = 18000 rpm

-1.744

-1.745 V -6.73 -6.72 -6.71 -6.7 -6.69 x 10-5

-1.746 -6.74

Trajectory of sun gear in x direction S = 12000 rpm

x 10-5

Fig. 14. (continued)

6.2. Dynamic results 6.2.1. Rotating carrier with no position error: In this section, the classic model relying on a constant geometry and mesh characteristics is confronted with the proposed approach where updated error and DOF-dependent geometry and tooth contact conditions are considered. The planetary gear set is dened in Table 1 (case 2). For a xed carrier, the two models generally give very similar dynamic tooth loads and no obvious differences have been reported. For a rotating carrier (epicyclic gear set), the results in Fig. 8 show that different dynamic tooth load curves are obtained depending on the modelling hypotheses. This effect is particularly marked in the high-speed supercritical zone where the dynamic forces on the meshes are modied compared with the classic ndings. These deviations can be related to the evolutions of mesh geometry caused by the centrifugal effects on the planets. Fig. 9 shows the variations of the average contact ratios for the sun-gear/planet and planet/ring-gear meshes and it can be observed that starting from approximately the same value of 1.6 at low-speeds, the external and internal mesh geometries progressively diverge as the carrier speed increases. The area where the models give different dynamic results is also characterised by some sensitivity to the initial conditions used in the numerical process as illustrated by the set of response curves in Fig. 8. It is to be noted that this slightly chaotic behaviour is not caused by contact losses and shocks since the minimum dynamic tooth loads are always positive over the speed range (Fig. 8).

4570

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

1.192 Trajectory of planet #1 in y direciton

x 10-4 Trajectory of planet #1 in y direciton

1.2

x 10-4

1.19

1.195 C 1.19

1.188

1.186

1.185

1.184

1.18

1.182 -5

5 S = 200 rpm

10

15

1.175 -4

-2

0 S = 5000 rpm

4 x 10-7

Trajectory of planet #1 in x direciton x 10-8

Trajectory of planet #1 in x direciton

1.194 Trajectory of planet #1 in y direciton 1.192 1.19 1.188 1.186

x 10

-4

1.19 V Trajectory of planet #1 in y direciton

x 10

-4

V 1.189 C

1.188

C 1.184 1.182 1.18 -1

1.187

1.186

-0.5

0 S = 12000 rpm

0.5

1.185 -1

Trajectory of planet #1 in x direciton x 10-6

Trajectory of planet #1 in x direciton x 10-7 S = 18000 rpm

Fig. 14. (continued)

6.2.2. Inuence of planet position errors: A negative tangential error of 0.2 mm on planet #1 is simulated and Fig. 10 shows the consequences in terms of mesh stiffness functions for every sun-gear/planet contact at high-speed (11 800 rpm on the sun-gear). Every mesh is found to possess a particular level of average stiffness along with different time evolutions or parametric excitations suggesting that, as opposed to the errorless case, each mesh may experience a different dynamic behaviour. However, the corresponding deviations in terms of dynamic forces are actually limited as shown in Fig. 11 where maximum tooth loads are found to be only marginally sensitive to the couplings between mesh geometry and deections. Signicant deviations emerge at high-speeds with a larger error of 0.5 mm as illustrated in Fig. 12 but, in view of the usual tolerances in planetary gears, this case certainly corresponds to a rather extreme situation whose interest is mostly academic. Compared with the errorless dynamic responses, the average level of the maximum tooth loads are identical on diametrically opposed planets; higher on planets #1 & 3 and lower on planets #2 & 4. Two critical speeds are observed which are slightly shifted with respect to those for ideal planet positioning. These results can be generalised over a range of tangential errors on one planet as illustrated in Fig. 13 where the maximum load ratio for all planets is plotted versus error amplitude and speed. The results at low speeds compare very well with those obtained analytically by Singh [35,36] using the so-called neutralising ratio and effective sun-planet-ring-bearing tangential stiffness. At higher speeds, tooth load peaks are observed whose positions and amplitudes differ depending on the modelling hypotheses (constant or

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

4571

DOF-dependent solution). It is observed that, even for helical gears, signicant tooth load amplications can be found in the presence of planet position errors. These dynamic effects exacerbate the consequences of the quasi-static uneven loading on the planets potentially leading to reliability problems. Finally, considering trajectories, the two models lead to somewhat different sun-gear and planet trajectories depending if the carrier is xed or rotating as illustrated in Fig. 14.
0.4 0.35 Planet load ratios 0.3 0.25 0.2 0.15 100 200 300 400 500 600 700 800 900 1000 Input torque (Nm) 0.27 0.265 Planet load ratios 0.26 0.255 0.25 0.245 0.24 0.235 0.23 -500 -400 -300 -200 -100 0 100 200 300 400 500 Tangential position errors on planet #1 (um)
Fig. 15. (a) Static planet load sharing ratio versus sun gear torque (tangential position error ey1 0.0002 m on planet #1, sun-gear speed is 200 rpm, rotating ring-gear, Planetary gear data: Table 1 - case 2). (b) Static planet load sharing ratio versus tangential position error on planet #1 (sun-gear torque is 1000 Nm; sun-gear speed is 200 rpm, rotating ring-gear). Keys: proposed model: sun-gear/planet #1, sun-gear/planet #2,3,4. Formulae in [35]: sun-gear/planet #1, sun-gear/planet #2,3,4.

P1 average

P2,P3,P4

P2,P3,P4 average

P1

2.5

Max load ratios of sun-gear/planets

1.5

0.5 0 0.5 1 1.5 2 x 104

Rotational speed of sun gear (rpm)

Fig. 16. Maximum dynamic load ratios for sun-gear/planets versus sun-gear speedComparisons between oating sun-gear and oating planet arrangements (tangential position error ey1 0.0002 m on planet #1; rotating ring-gear; planetary gear data: Table 1 - case 2). Keys: Floating sun-gear: sun-gear/planet #1, sun-gear/planet #2, sun-gear/planet #3, sun-gear/planet #4, Floating planets: sungear/planet #1 sun-gear/planet #2,3,4.

4572

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

6.2.3. Inuence of oating members: It is well-known that, by allowing some members to oat using very exible mounts, it is possible to compensate for some geometrical errors such as run-out, non uniform diameters, tooth spacing, etc. and improve the load sharing amongst the planets. Floating sun-gears are quite common but it has been demonstrated that the elasticity of the planet shafts can be very inuential [37], [40] leading to improved load sharing. The versatility of the proposed model is rst illustrated for quasi-static conditions by comparing the results obtained at low speed and those given by Singhs formulae [35,36]. Two sets of results for the 4-planet gear set dened in Table 1 (case 2) with isotropic exible planet supports (1E6 N/m in two perpendicular directions) are presented. In Fig. 15a, a constant tangential error of 200 mm on one planet is considered and the sun torque is varied between 100 and 1000 Nm whereas, in Fig. 15b, the torque is kept constant (1000 Nm) and the tangential error varies between 0.5 and 0.5 mm. It is to be noted that the practical considerations associated with exible pin design are not considered here and that oating mounts are simply simulated by low stiffness elements at the planet centres. For every conguration, the load factor calculated by the dynamic model and the results from [35,36] have been superimposed. The main conclusions are: (i) the static planet loads calculated by Singhs formulae and by the dynamic model at low speeds are similar and (ii) oating planets are interesting for load sharing superior to the conguration with a oating sun-gear only (compare with the results in Fig. 7 which are worse even if the position error is smaller). These observations are partly conrmed for dynamic regimes in Figs. 16 and 17 which, for a tangential error of 0.2 mm on one planet, show the evolutions of the dynamic load factors versus speed. For a xed carrier (Fig. 16), oating planets are more interesting over the entire speed range since every planet mesh experiences nearly the same dynamic load. However, the curves in Fig. 17 reveal that oating planets are not as favourable in the case of a rotating carrier since, above a certain speed threshold, signicant dynamic overloads are observed. This effect is caused by the combination of the centrifugal forces on the planets and the exibility of the planet pins leading to a reduction in the sun-gear/planet contact ratios (while ring-gear/planet contact ratios are increased) below the admissible values for continuous motion

Max load ratios of sun-gear/planets

2.5

1.5

0.5 0 0.5 1 1.5 2 x 104

Rotational speed of sun gear (rpm)

Fig. 17. Maximum dynamic load ratios for sun-gear/planets versus sun-gear speed Comparisons between oating sun-gear and oating planet arrangements (tangential position error ey1 0.0002 m on planet #1; rotating carrier; planetary gear data: Table 1 - case 2). Keys: Floating sun-gear: sun-gear/planet #1, sun-gear/planet #2, sun-gear/planet #3, sun-gear/planet #4. Floating planets: sun-gear/planet #1, sun-gear/planet #2,3,4.

3.5 3 Contact ratios 2.5 2 1.5 1 0.5 0 2000 4000 6000 8000 Rotational speed of sun gear (rpm)
Fig. 18. Variations of contact ratios with speed for a rotating carrier and oating members (tangential position error ey1 0.0002 m on planet #1). Keys: Floating sun-gear: sun-gear/planets, ring-gear/planets. Floating planets: sun-gear/planets, ring-gear/planets.

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

4573

transfer. Finally, it is to be noted that the positions of the tooth critical speeds are largely independent of the sun-gear and/ or planet support stiffness (Fig. 18).

7. Conclusion A classic lumped parameter model of planetary gear with two translational and one torsional DOF has been extended to account for planet position errors and determine quasi-static and dynamic load sharing amongst the various meshes. Rigid-body rotations dene the reference from which the degrees-of-freedom are dened and a specic procedure has been presented which makes it possible to solve for the motion transfer in the presence of errors and gives the initial separations between the teeth. A unique feature of the proposed approach is that mesh properties, errors and deections are coupled. At each time step, the actual pressure angles, contact ratios, positions of base planes, meshing areas and consequently mesh phases are re-evaluated accordingly thus leading to a set of non-linear parametrically differential equations. However, the individual mesh stiffness model has been kept simple with the hypothesis of mesh stiffness functions directly proportional to the instantaneous contact lengths and, in this paper, tooth prole and lead modications have not been considered. Based on a number of comparisons with the experimental evidence of Ligata et al.[31,32], it has been demonstrated that, beyond dynamic simulations, the model is also representative of quasi-static load sharing. Similarly, the calculated planet load ratio compares very well with the closed-form formulae proposed by Singh [35,36] for a range of planetary gear geometries and planet tangential errors. The dynamic results in this paper show that the inuence of the couplings between the errors, DOFs, mesh geometry and excitations is not uniform, contrasts can be noticed between dynamic tooth loads (largely insensitive) and displacements (trajectories) which are more likely to be affected by the above-mentioned factors. Systems with rotating carriers can also exhibit rather different dynamic behaviour depending on the modelling hypotheses mostly because of the centrifugal effects on the planets at high speeds. It can therefore be posited that the classic models with constant mesh geometry are adapted to dynamic force predictions whereas more sophisticated approaches might be necessary to capture vibration signals. Regardless of the technological limitations and/or constraints, it has been conrmed that, for a xed carrier, oating planets arrangements (combined or not with a oating sun-gear) can be interesting with regard to planet loading in static and dynamic conditions. Similar conclusions can be drawn for rotating carriers as long as the centrifugal forces on the planets are not too large. Further studies are under way in order to account for gear eccentricities and tooth shape modications. References
[1] F. Cunliffe, J.D. Smith, D.B. Welbourn, Dynamic tooth loads in epicyclic gears, ASME Journal of Engineering for Industry 95 (1974) 578584. [2] R. August, R. Kasuba, Torsional vibrations and dynamic loads in a basic planetary gear system, ASME Journal of Vibration Acoustics Stress and Reliability Design 108 (1986) 348353. [3] T. Hidaka, Y. Terauchi, Dynamic behavior of planetary gear (1st Report : Load distribution in planetary gear ), Bulletin of JSME 19 (132) (1976) 690698. [4] T. Hidaka, Y. Terauchi, K. Ishioka, Dynamic behavior of planetary gear (2nd Report: Displacement of sun gear and ring gear), Bulletin of JSME 19 (138) (1976) 15631570. [5] T. Hidaka, Y. Terauchi, M. Nohara, J. Oshita, Dynamic behavior of planetary gear (3rd Report: Displacement of ring gear in direction of line of action), Bulletin of JSME 120 (150) (1977) 16631672. [6] T. Hidaka, Y. Terauchi, K. Nagamura, Dynamic behavior of planetary gear (6th Report: Inuence of meshing phase), Bulletin of JSME 22 (169) (1979) 10261033. [7] T. Hidaka, Y. Terauchi, K. Nagamura, Dynamic behavior of planetary gear (7th Report: Inuence of the thickness of the ring gear), Bulletin of JSM 22 (170) (1979) 11421149. [8] A. Saada, P. Velex, An Extended Model for the Analysis of the Dynamic Behavior of Planetary Trains, Proceedings of the 6th ASME International Power Transmission and Gearing Conference, Phoenix, 1992, pp.513520. [9] A. Kahraman, Planetary gear train dynamics, ASME Journal of Mechanical Design 116 (3) (1994) 713721. [10] P. Velex, L. Flamand, Dynamic Response of Planetary Trains to Mesh Parametric Excitations, ASME Journal of Mechanical Design 118 (1996) 714. [11] R.J. Parker, V. Agashe, S.M. Vijayakar, Dynamic response of a planetary gear system using a nite element/contact mechanics model, ASME Journal of Mechanical Design 122 (3) (2000) 304310. [12] A. Kahraman, S.M. Vijayakar, Effect of internal gear exibility on the quasi-static behavior of a planetary gear set, ASME Journal of Mechanical Design 123 (2001) 408415. [13] A. Kahraman, A.A. Kharazi, M. Umrani, A deformable body dynamic analysis of planetary gears with thin rims, Journal of Sound Vibrations 262 (3) (2003) 752768. [14] V. Abousleiman, P. Velex, A hybrid 3D nite element/lumped parameter model for quasi-static and dynamic analyses of planetary/epicyclic gear sets, Mechanism and Machine Theory 41 (6) (2006) 725748. [15] V. Abousleiman, P. Velex, S. Becquerelle, Modeling of spur and helical gear planetary drives with exible ring gears and planet carriers, ASME Journal of Mechanical Design 129 (1) (2007)95-106 129 (1) . [16] J. Helsen, F. Vanhollebeke, B. Marrant, D. Vandepitte, W. Desmet, Multi-body modelling of varying complexity for modal behaviour analysis of wind turbine gearboxes, Renewal Energy 36 (11) (2011) 30983113. [17] J. Lin, R.G. Parker, Analytical characterization of the unique properties of planetary gear free vibration, ASME Journal of Vibration And Acoustics 121 (3) (1999) 316321. [18] J. Lin, R.G. Parker, Structured vibration characteristics of planetary gears with unequally spaced planets, Journal of Sound Vibrations 228 (1) (2000) 109128.

4574

X. Gu, P. Velex / Journal of Sound and Vibration 331 (2012) 45544574

[19] D. Kiracofe, R.G. Parker, Structured vibration modes of general compound planetary gear systems, ASME Journal of Vibrations and Acoustics 129 (1) (2007) 116. [20] T. Eritenel, R.G. Parker, Modal properties of three-dimensional helical planetary gears, Journal of Sound Vibrations 325 (1-2) (2009) 397420. [21] D.L. Seager, Conditions for the neutralization of excitation by the teeth in epicyclic gearing, IMechE Journal of Mechanical Engineering Science 17 (5) (1975) 293298. [22] R.G. Parker, A physical explanation for the effectiveness of planet phasing to suppress planetary gear vibration, Journal of Sound Vibrations 236 (4) (2000) 561573. [23] R.G. Parker, J. Lin, Mesh phasing relationship in planetary and epicyclic gears, ASME Journal of Mechanical Design 126 (2) (2004) 365370. [24] V.K. Ambarisha, R.G. Parker, Supression of planet mode response in planetary gear dynamics through mesh phasing, Journal of Vibrations and Acoustics 128 (2) (2006) 133142. [25] M. Botman, A. Toda Planet Indexing in Planetary Gears for Minimum Vibrations, ASME Paper No 79-DET-73, 1979. [26] P. Ma, M. Botman, Load sharing in a planetary gear stage in the presence of gear errors and misalignment, ASME Journal of Mechanical Transmission and Automation Design (1984) 7. 84-DET-54. [27] G.-J. Cheon, R.G. Parker, Inuence of manufacturing errors on the dynamic characteristics of planetary gear systems, Journal of Mechanical Science and Technology 18 (4) (2004) 606621. [28] G.-J. Cheon, R.G. Parker, Inuence of bearing stiffness on the static properties of a planetary gear system with manufacturing errors, Journal of Mechanical Science and Technology 18 (11) (2004) 19781988. [29] M. Inalpolat, A. Kahraman, A theoretical and experimental investigation of modulation sidebands of planetary gear sets, Journal of Sound Vibrations 323 (3-5) (2009) 677696. [30] M. Inalpolat, A. Kahraman, A dynamic model to predict modulation sidebands of a planetary gear set having manufacturing errors, Journal of Sound Vibrations 329 (4) (2010) 371393. [31] H. Ligata, A. Kahraman, A. Singh, An experimental study of the inuence of manufacturing errors on the planetary gear stresses and planet load sharing, ASME Journal of Mechanical Design 130 (4) http://dx.doi.org/10.1115/1.2885194. [32] H. Ligata, A. Kahraman, A. Singh, A closed-form planet load sharing formulation for planetary gear sets using a translational analogy, ASME Journal of Mechanical Design 131 (2) http://dx.doi.org/10.1115/1.3042160. [33] A. Singh, Application of a System Level Model to Study the Planetary Load Sharing Behavior, ASME, Journal of Mechanical Design 127 (3) (2005) 469476. [34] A. Singh, Inuence of planetary needle bearings on the performance of single and double pinion planetary systems, ASME Journal of Mechanical Design 129 (1) (2007) 8594. [35] A.Singh Load, Sharing behavior in epicyclic gearsPhysical explanation and generalized formulation, Mechanism and Machine Theory 45 (3) (2010) 511530. [36] A. Singh, Epicyclic load sharing mapDevelopment and validation, Mechanism and Machine Theory 46 (5) (2011) 632646. [37] A. Kahraman, Load sharing characteristics of planetary transmissions, Mechanism and Machine Theory 29 (8) (1994) 11511165. [38] A. Bodas, A. Kahraman, Inuence of carrier and gear manufacturing errors on the static load sharing behavior of planetary gear sets, JSME International Journal, Series C 47 (2004) 908915. [39] P. Velex, J. Bruyere, D.R. Houser, Some analytical results on transmission errors in narrow-faced spur and helical gearsInuence of prole modications, ASME Journal of Mechanical Design 133 (3) http://dx.doi.org/10.1115/1.4003578. 031010 (11 pages). [40] A.N. Montestruc, Inuence of planet pin stiffness on load sharing in planetary gear drives, ASME Journal of Mechanical Design 133 (1) http://dx.doi.or g/10.1115/1.4002971. 014501 (7 pages). [41] X. Gu, Inuence of Planet Position Errors and Eccentricities on Planetary Gear Dynamics, PhD Thesis INSA Lyon, April 2012, 169 p.

You might also like