Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 53, 1, pp. 125-138, Warsaw 2015 DOI: 10.15632/jtam-pl.53.1.125 MODAL ANALYSIS OF BACK-TO-BACK PLANETARY GEAR: EXPERIMENTS AND CORRELATION AGAINST LUMPED-PARAMETER MODEL Ahmed Hammami Laboratory of Mechanics, Modeling and Production – National School of Engineers Sfax, Sfax, Tunisia and Department of Structural and Mechanical Engineering, University of Cantabria, Santander, Spain e-mail: ahmed hammami@voila.fr Alfonso Fernandez Del Rincon, Fernando Viadero Rueda Department of Structural and Mechanical Engineering, University of Cantabria, Santander, Spain e-mail: alfonso.fernandez@unican.es; fernando.viadero@unican.es Fakher Chaari, Mohamed Haddar Laboratory of Mechanics, Modeling and Production – National School of Engineers Sfax, Sfax, Tunisia e-mail: fakher.chaari@gmail.com; mohamed.haddar@enis.rnu.tn In order to characterize the dynamic behaviour of a back-to-back planetary gear, experi- mental and numerical modal analysis techniques are achieved. Rotational and translational modal deflections are highlighted. Natural frequencies are compared to the results from the lumped-parametermodel.Themodes arepresented in thenumerical studies in low-frequency and high-frequency bands. Distributions of modal kinetic and strain energies are studied. Keywords: back-to-back planetary gears, modal analysis, lumped-parameter, impact test 1. Introduction Gears are widely used in many power transmission applications and are characterized by their high efficiency. However, they can be subjected to severe operating conditions giving rise to dynamic problems. Bartelmus (2001) showed the influence of error mode parameters and the influence of clutch damping on the dynamic factor. Time varying stiffness and backlash, which have influence on the dynamic behaviour of the gearbox, are the main excitations of the model of Łuczko (2008) who studied the chaotic vibrations in one stage spur gear. Planetary gears are compact kind of gears. They are excellent mechanisms for transmitting significant power with large speed reductions or multiplications. These kinds of gears are used in many fields of application like wind turbines, new generation aircraft engines, hybrid car transmissions. Back-to-back planetary gear configuration is compact with mechanical power circulation for economic and energy efficiency criteria. Many researchers studiedmodal proprieties of planetary gears highlighting and the relation- ship between natural frequencies and parameters of the system. Cunliffe et al. (1974) focused on an analytical model of a planetary gear with a fixed carrier and characterised vibration modes. Botman (1976) studied the effects of planet pin loads on the eigen-frequencies using an eighteen degree of freedom system. August andKasuba (1986) used a torsional model with nine degrees of freedomto determine torsional vibrations anddynamic loads in abasic planetary gear system. Kahraman (1993) and Sondkar and Kahraman (2013) analyzed respectively three dimensional modal deflection of a helical planetary gear and a double-helical planetary gear. Kahraman 126 A. Hammami et al. (1994) provided expressions for natural frequencies by using a rotational lumped-parametermo- del. Lin and Parker categorized rotational, translational, and planet modes and explained the unique modal proprieties of planetary gear with equal planet spacing (Lin and Parker, 1999) and unequal planet spacing (Lin andParker, 2000). Tanna and Lim (2004) compared themodal frequencies of ring gears and idealized smooth rings. They quantified the frequency deviations in applying the simpler smooth ring solutions to represent the primarymodal behaviour of ring gear structures.Wojnarowski et al. (2006) used the versatile graph theory to represent the struc- ture of a planetary gear system and the gear modeling in order to achieve an eigen-vibration analysis of the gear system. WuandParker (2008) considered the elastic deformation of the ring on themodal proprieties of a planetary gearwith equal planet spacing andunequal planet spacing (Parker andWu, 2010). Eritenel andParker (2009) presented themodal properties of three-dimensional helical planetary gears. Bu et al. (2012) developed a generalized dynamicmodel for a herringbone planetary gear train in order to investigate itsmodal properties. Vibrationmodes of compound planetary gears were studied by Kiracofe and Parker (2007) and Guo and Parker (2010). Cooley and Parker (2012, 2013) modeled the gyroscopic effects and studied the vibration properties of high-speed planetary gears. There were few experimental studies to correlate the numerical modal analysis with experi- mental modal analysis. Singh et al., (2008) presented results of a comprehensive experimental and theoretical study under static condition to determine the influence of manufacturing tole- rance on gear stresses and planetary load sharing. Kahraman andVijayakar (2001) investigated the effect of internal gear flexibility under the quasi-static condition behaviour of a planetary gear set.TodaandBotman (1979) showed experimentally thatplanetary gear vibration resulting from the planet position error can beminimized by a proper indexing planet. Velex et al. (1994) compared experimental results of a double helical epicyclic gear in its natural environment to a mathematical model. But some natural frequencies do not agreed between their model and experiments. Ericson and Parker (2013) applied experimental modal analysis techniques to characterize the planar dynamic behaviour of spur planetary gears and correlated these results against lumped parameter and finite element models. Thispaperwill discuss themodalpropertiesof a special configurationof a two stageplanetary gears mounted back-to-back. To achieve this, both numerical and experimental investigations are conducted. Experimental results are correlated against results from the lumped-parameter model. Modal characteristics are obtained and classified. 2. Experimental setup 2.1. Description of the test bench A test bench composed of two identical planetary gear sets with the same gear ratio (Fig. 1) is developed at the University of Cantabria in Spain. It is composed of two identical planetary gear sets with the same gear ratio (Fig. 1). The test bench is composed of a test gear set and a reaction gear set connected back-to-back: the sungears of bothplanetary gear sets are connected through a common shaft, and the carriers of both planetary gear sets are connected to each other through a rigid hollow shaft (Fig. 1). To load the system, masses are added to an arm fixed to the ring of the reaction gear set (Fig. 2). 2.2. Excitation method In order to achieve the experimental modal analysis, an impact hammer is used. Modal analysis of back-to-back planetary gear... 127 Fig. 1. Back-to-back layout as assembled in the bench Fig. 2. Application of the external torque The hammer model is PCB 086B03 with a sensitivity of 2.15mV/N. Ametal tip is used to study the response up to 2000Hz. 2.3. Data acquisition Four tri-axial accelerometersmounted on the free ring and thefixed ring aswell as the carrier measure the rotational and translational vibration of each body (Fig. 2). The signals measured by the accelerometers will be acquired by an LMS SCADAS Mobile SCM01 system and the data will be processed with the software “LMS Test.Lab Impact test” to obtain the Frequency Response Function. 3. Numerical model The model of the back-to-back planetary gear is based on the model developed by Lin and Parker (1999). The components are the ring (r), sun (s), planets (1, 2, 3) and carrier (c) which carries the planets as shown in Fig. 3. The test ring and the test sun are respectively linked to the three planets of the test gear set via gear mesh stiffness Krt1, Krt2,Krt3 andKst1, Kst2, Kst3. The same is on the reaction gear set, the reaction ring and the reaction sun are respectively linked to the three planets via gear mesh stiffness Krr1, Krr2, Krr3 and Ksr1, Ksr2, Ksr3. The reaction ring which has the bearing stiffnessKrrx andKrry in thex and y directions is free and its torsional stiffness isKrru whereas the test ring which has the bearing stiffnessKrtx andKrty is not exactly fixed but its torsional stiffness Krtu is very high. The planets of the reaction gear set and the test gear set have the bearing stiffnessKrpix andKtpiy in thex and y directions (i=1,2 or 3). The test carrier and the reaction carrier have also the bearing stiffness respectively Kctx,Kcty and Kcrx, Kcry. The sun gears of both planetary gear sets are connected through a common shaft which has a torsional 128 A. Hammami et al. Fig. 3. Model of the planet gear stiffness Ks. The carriers of both planetary gear sets are connected to each other through a hollow shaft which has a torsional stiffness Kc. The equation of motion of the system for the back-to-back planetary gear with 3 planets is Mq̈+ΩcGq̇+[Kb+KM −Ω 2 cKΩ]q=F(t) (3.1) where M is the mass matrix, Kb is the bearing stiffness matrix, G and KΩ result from the high-speed carrier, KM is the stiffness matrix and F(t) is the external force vector applied to the system. All these matrices are defined in appendix. The planets in the test gear set and the reaction gear set are assumed identical and equally spaced. If the gyroscopic terms G, KΩ and F(t) are neglected, the equation of motion will simplify to Mq̈+[Kb+KM]q=0 (3.2) q is the degree of freedom vector defined as the following q= { qr qt } (3.3) qr is the degree of freedom vector in the reaction gear set (r) and qt is the degree of freedom vector in the test gear set defined by qr = {xcr,ycr,ucr,xrr,yrr,urr,xsr,ysr,usr,ζ1r,η1r,u1,ζ2r,η2r,u2r,ζ3r,η3r,u3r} T qt = {xct,yct,uct,xrt,yrt,urt,xst,yst,ust,ζ1t,η1t,u1t,ζ2t,η2t,u2t,ζ3t,η3t,u3t} T (3.4) The carrier, the ring and the sun translations xij and yij (i= c,r,s and j= r,t) and planet translations ζnj and ηnj (n = 1,2,3 and j = r,t) are measured with respect to the rotational frame of referenceR= {~i,~j,~k}. The rotational coordinates are urj = rrjθrj for the reaction gear set utj = rtjθtj and for the test gear set where j= c,r,s,1,2,3. θrj and θtj are the component rotation; rrj and rtj are the base radius for the sun, ring and planets and the radius of the circle passing through the planets centre for the carrier. Modal analysis of back-to-back planetary gear... 129 4. Comparative results In this part, results from themodal analysis on the back-to-back planetary gears are presented. Experimental and numerical modes are checked and compared. The modes are classified and studied in terms of modal kinetic energies andmodal strain energies for each natural frequency. Finally, the effects of loading change (by adding mass) on natural frequencies and vibration modes are studied. 4.1. Natural frequencies and vibration modes The natural frequencies and vibration modes are determined from equation of motion (3.2) and the associated eigen-value problem given by ω2iMφi = [Kb+KM]φi (4.1) where ωi are the natural frequencies andφi are the vibration modes. The rotational, translational and planetmode types were defined by Lin and Parker (1999): • The rotational and translational modes are the vibration with natural frequencies having respectively the multiplicity m=1 andm=2 • Theplanetmodesexist only if thenumberofplanetN > 3andhavemultiplicitym=N−1. In the test gear set and the reaction gear set, there are only three planets (N =3). So, only rotational and translational modes will be expected when solving the eigen-value problem. According to the energy distribution and component deflection, Cunliffe et al. (1974) and Ericson andParker (2013) classified the natural frequencies into two categories according to the frequency band. The modes of the first band are called “bearing modes” or “fixture modes”. Themodes of the second band are called “tooth modes” or “gear modes”. The distinction between the fixture and gear modes is independent of the rotational and translational modes. Table 1 shows the parameters of the model. The natural frequencies are identified numerically and experimentally. The impacts have beenmade on the arm of the free ring. 4.1.1. Teeth modes For every test, we achieved ten impacts, and the acquisition system “LMSTest.Lab Impact test” averaged the results. Figure 4 represents the Frequency Response Function (FRF) in the fix ring. Table 2 compares the natural frequencies determined from the impact test experiments and numerical model and shows the multiplicity of each mode from the analytical model. Experiments show fourteen natural frequencies whereas the numerical model shows twelve natural frequencies composed of eight rotational modes and four translational modes. The na- tural frequencies 41Hz, 182Hz and 770Hz appear only in the impulse test whereas the natural frequency 247Hz appears only in the numerical model. The remaining natural frequencies agree within 11%as themaximum rate. There is only one identified natural frequencywhere the error reaches 30%. Figure 5 shows themodal deflection for the natural frequencies 192Hz and 244Hz. It is well noticed that all planets have the samemotion andmove in phase in the rotational modes (118Hz, 136Hz, 244Hz, 376Hz, 387Hz, 547Hz and 550Hz). Besides, the carriers, the suns and the rings on the test and the reaction gear set have pure translation in the translation modes (192Hz, 288Hz, 344Hz and 536Hz). 130 A. Hammami et al. Table 1. Lumped-parameter values of the reaction and test planetary gear set and the shaft stiffness Sun Planet Ring Carrier Reaction planetary gear set Teeth number 16 24 65 – Mass [Kg] 0.485 1.225 28.1 3.643 Base diameter [mm] 61.38 92.08 249.38 57.55 Moment of inertia [Kgm2] 356 ·10−6 2045 ·10−6 697767 ·10−6 21502 ·10−6 Bearing stiffness [N/m] 8.8 ·107 3.5 ·106 2.1 ·107 4.8 ·108 Torsional stiffness [N/m] – – 0 – Mesh stiffness [N/m] 4.46 ·108 6.28 ·108 Test planetary gear set Teeth number 16 24 65 – Mass [Kg] 0.485 1.225 28.1 3.643 Base diameter [mm] 61.38 92.08 249.38 57.55 Moment of inertia [Kgm2] 356 ·10−6 2045 ·10−6 697767 ·10−6 21502 ·10−6 Bearing stiffness [N/m] 8.8 ·107 3.5 ·106 2.1 ·107 4.8 ·108 Torsional stiffness [N/m] – – 7.9 ·106 – Mesh stiffness [N/m] 4.46 ·108 6.28 ·108 Shaft stiffness Torsional [Nm/rad] 3.73 ·104 – – 8.38 ·105 Flexural 4.9 ·105 – – 1.1 ·107 Fig. 4. Frequency response function of the fix ring for low frequencies 4.1.2. Bearing modes The gear modes are identified by solving the eigen-value problem and using the impact test. Figure 6 represents the frequency response function in thefixed ringwith thehammer impact test. Table 3 compares the natural frequencies determined from the impact test experiments and the numerical model in high frequencies. Experiments showfive natural frequencies whereas the numerical model shows three natural frequencies associated with three translational modes. The natural frequencies 1019Hz and 1457Hz which can be induced by the axial vibration appear only in the impulse test. The remaining natural frequencies agree within 7% as themaximum difference rate. Modal analysis of back-to-back planetary gear... 131 Table 2.Natural frequencies from the experiments and numerical model in low frequencies Mode Type Impulse Numerical Multiplicity Error exp. [Hz] model [Hz] [%] 1 – 41 – – – 2 R 82 118 1 30 3 R 144 136 1 6 4 – 182 – – – 5 T 200 192 2 4 6 R 244 244 1 0 7 R – 247 1 – 8 T 299 288 2 4 9 T 317 344 2 11 10 R 350 376 1 7 11 R 362 387 1 7 12 T 516 536 2 4 13 R 566 547 1 4 14 R 592 550 1 8 15 – 770 – – – Fig. 5. Vibrationmodes in low frequencies Fig. 6. Frequency response function of the fixed ring in high frequencies Figure 7 shows that the carriers, the suns and the rings on the test and the reaction gear set have pure translation in the translation modes 1332Hz. 132 A. Hammami et al. Table 3.Natural frequencies from the experiments and numerical model in high frequencies Mode Type Impulse Numerical Multiplicity Error exp. [Hz] model [Hz] [%] 16 – 1019 – – – 17 T 1208 1232 2 2 18 – 1457 – – – 19 T 1732 1847 2 7 20 T 1762 1888 2 7 Fig. 7. Vibrationmodes in high frequencies 4.2. Analysis of the distribution of modal kinetic energies and modal strain energies Computation of the modal strain energy and the modal kinetic energy distributions gives information on bodies sought to critical speeds (which excite the natural frequencies) in terms of dominant motion and deformation. The total modal strain energy can be written as the sum of the strain energies of rotation and translation from each component Epφ = 1 2 φtiK̃φi = ∑ Epφij + ∑ Epφijw+ ∑ (Epφsun+Epφrin) (4.2) whereEpφij andEpφijw are the strain energy of bearing stiffness in the rotational and transla- tional movement of the suns, carriers, rings and planets (j = s,c,r,t,1,2,3) in the test gear set and the reaction gear set (i= t,r).Epφsun andEpφrin are the strain energy of sun-planets and planets-ring meshing in the test gear set and the reaction gear set. The modal kinetic energy can also be written as the sum of the kinetic energies of rotation and translation from each component of the system Ecφ = 1 2 w2iφ t iMφi = ∑ Ecφij + ∑ Ecφijw (4.3) whereEcφij and Ecφijw are the kinetic energy in the rotational and translational movement of the suns, carriers, rings and planets (j = s,c,r,t,1,2,3) in the test gear set and the reaction gear set (i= t,r). 4.2.1. Teeth modes Figure 8 shows the distribution of modal kinetic energies in low frequencies. In theX-axis the contribution of each degree of freedom in the total modal strain energy is represented. Details are given in Table 4. The dominantmovement of the translational mode 547Hz is translation of the reaction ring along theX direction. Modal analysis of back-to-back planetary gear... 133 Fig. 8. Modal kinetic energies in low-frequencies Table 4.TheX-axis location of kinetic energies 1-2 Translations of reaction carrier 19-20 Translations of test carrier 3 Rotation of reaction carrier 21 Rotation of test carrier 4-5 Translations of reaction ring 22-23 Translations of test ring 6 Rotation of reaction ring 24 Rotation of test ring 7-8 Translations of reaction sun 25-26 Translations of test sun 9 Rotation of reaction sun 27 Rotation of test sun 10-11 Translations of reaction planet 1 28-29 Translations of test planet 1 12 Rotation of reaction planet 1 30 Rotation of test planet 1 13-14 Translations of reaction planet 2 31-32 Translations of test planet 2 15 Rotation of reaction planet 2 33 Rotation of test planet 2 16-17 Translations of reaction planet 3 34-35 Translations of test planet 3 18 Rotation of reaction planet 3 36 Rotation of test planet 3 Figure 9 shows the distribution ofmodal strain energies in low frequencies, where theX-axis is defined in Table 5. Fig. 9. Strain energy in low-frequencies The dominant strain energy of the translational mode 547Hz is located in the meshing between reaction ring-planet No. 3. Table 6 presents a resume of themodal dominantmovement and the dominant strain energy in each low-frequency mode. In the teeth modes which characterize the deflection in fixture components, themodes have significant strain energy in the teeth meshing sun-planets and ring-planets of the two stages. In fact, the stiffness of themeshing is higher than the stiffness of the shafts andbearing of the gears. So, the strain energies are bigger for the components which have higher stiffness. Also, those modes are predominantly characterized bymotion of the individual planetary gear components, particularly the planet gears. 134 A. Hammami et al. Table 5.TheX-axis location of strain energies 1-3 Bearing of reaction carrier 25-27 Bearing of test sun 4-6 Bearing of reaction ring 28-30 Bearing of test planet 1 7-9 Bearing of reaction sun 31-33 Bearing of test planet 2 10-12 Bearing of reaction planet 1 34-36 Bearing of test planet 3 13-15 Bearing of reaction planet 2 37-39 Meshing reaction ring-planets 16-18 Bearing of reaction planet 3 40-42 Meshing reaction sun-planets 19-21 Bearing of test carrier 43-45 Meshing test ring-planets 22-24 Bearing of test ring 46-48 Meshing test sun-planets Table 6.Dominant motion and dominant strain energy in low-frequencies Frequencies Modal dominant movement Dominant strain energy 118Hz (R) Rotation of reaction carrier Meshing reaction sun-planet 3 136Hz (R) Rotation of test carrier Meshing test sun-planet 3 192Hz (T) Translation of reaction planet 1 (X) Meshing reaction sun-planet 1 Translation of test planet 2 (Y ) Meshing test sun-planet 1 244Hz (R) Rotation of all test planet (X) Meshing reaction sun-planet 3 288Hz (T) Translation of reaction planet 1 (X) Meshing reaction sun-planet 1 Translation of test planet 2 (X) Meshing test sun-planet 1 344Hz (T) Translation of reaction planet 3 (X) Bearing of reaction planet 3 (X) Translation of test planet 1 (X) Bearing of test planet 3 (X) 376Hz (R) Rotation of reaction carrier Meshing reaction ring-planet 3 387Hz (R) Rotation of test carrier Meshing test ring-planet 3 536Hz (R) Translation of reaction ring (X) Meshing reaction ring-planet 1 Translation of test ring (X) Meshing test sun-planet 2 547Hz (R) Rotation of reaction ring Meshing reaction ring-planet 3 550Hz (R) Rotation of test ring Meshing test ring-planet 3 (X) –X direction, (Y ) – Y direction 4.2.2. Bearing modes Figure 10 shows the distribution of modal kinetic energies in high frequencies. The X-axis is defined in Table 4. Fig. 10. Kinetic energy in high-frequencies The dominant motion in the translational mode 1867Hz is the translation of the reaction sun in the Y direction. Figure 11 shows the distribution of modal strain energies in low frequencies. The X-axis is defined in Table 5. The dominant strain energy in the translational mode 2527Hz is the bearing of the test carrier in the Y direction. Table 7 presents a resume of themodal dominantmovement and the dominant strain energy in each high-frequencymode. Modal analysis of back-to-back planetary gear... 135 Fig. 11. Strain energy in high-frequencies Table 7.Dominant motion and dominant strain energy in high-frequencies Frequencies Modal dominant movement Dominant strain energy 1232Hz (T) Translations of reaction ring (Y ) Meshing reaction ring-planet 2 Translations of test planet 1 (X) Meshing test ring-planet 2 1847Hz (T) Translations of reaction carrier (X) Bearing of reaction carrier (X) Translations of test carrier (X) Bearing of test carrier (X) 1888Hz (T) Translations of reaction carrier (X) Bearing of reaction carrier (X) Translations of test carrier (X) Bearing of test carrier (X) Table 7 shows bearing modes which are characterized by the strain and kinetic energies of the carriers (1847Hz and 1888Hz). In general, the significant strain energies in low frequencies are in gears teeth where the mesh stiffness is higher than the stiffness of the shafts and bearing, and themotion of gears has the highest kinetic energy. However, for the same condition, the dominant movement in high frequencies is the movement of carriers which have also the dominant strain energies. 5. Conclusions A modal analysis has been performed on a back-to-back planetary gear using two approaches: model basedand experimental study.Concerning themodellingapproach, amathematicalmodel of the gear system has been proposed. Solving the eigenvalue problem allowed recovering the modal characteristics of the transmission. It has been found that the natural frequencies can be divided into two main modes: rotational modes and translational modes. The experimental modal investigation done on a test rig confirmed this classification. When looking to the modal kinetic and strain energy distributions, another classification emerges according to the dominant energy in the system for each natural frequency. In fact, teeth modes and bearing modes have been highlighted. The selection of parameters in themodel has been optimized to get results as close as possible to the experimental results done by the hammer test. The adopted model in the paper considers only 3 degrees of freedom per component. Obviously, it will not be possible to get low percentage error for all the computed natural frequencies. Further modal analysis investigations will be done in future works with a model considering 6 degrees of freedom per component. Such studycanhelpdesigners to avoid critical frequencies on the systembychanging parameters of the system. It allows one also to localise critical components susceptible todamage when crossing natural frequencies. Acknowledgements This paper was financially supported by the Tunisian-Spanish Joint Project No. A1/037038/11. The authorswould like also to acknowledgeproject “Development ofmethodologies for the simulation and improvement of the dynamic behavior of planetary transmissions DPI2013-44860” funded by the SpanishMinistry of Science and Technology. Acknowledgmentto theUniversityofCantabria cooperationproject fordoctoral trainingofUniversity of Sfax’s students. 136 A. Hammami et al. A. Appendix Thematrix mass is defined M= diag(Mcr,Mrr,Msr,M1r,M2r,M3r,Mct,Mrt,Mst,M1t,M2t,M3t) (A.1) where Mij = diag(mij,mij,Iij/r 2 ij) i= c,r,s,1,2,3 j= r,t (A.2) G= diag(Gcr,Grr,Gsr,G1r,G2r,G3r,Gct,Grt,Gst,G1t,G2t,G3t) (A.3) Gij =   0 −2mij 0 2mij 0 0 0 0 0   i= c,r,s,1,2,3 j= r,t (A.4) KΩ = diag(KΩr,KΩt) (A.5) KΩr = diag(mcr,mcr,0,mrr,mrr,0,msr,msr,0,mpr,mpr,0,mpr,mpr,0,mpr,mpr,0) (A.6) KΩt = diag(mct,mct,0,mrt,mrt,0,mst,mst,0,mpt,mpt,0,mpt,mpt,0,mpt,mpt,0) (A.7) The bearing stiffness matrix Kb = diag(Kcrb,Krrb,Ksrb,0,0,0,Kctb,Krtb,Kstb,0,0,0) (A.8) Kijb = diag(Kijx,Kijy,Kiju) i= c,r,s,1,2,3 j= r,t (A.9) Km = [ Kmr 0 0 Kmt ] +Kc (A.10) Km =   ∑ Knc1 0 0 K 1 c2 K 2 c2 K 3 c2 0 ∑ Knr1 0 K 1 r2 K 2 r2 K 3 r2 0 0 ∑ Kns1 K 1 s1 K 2 s1 K 3 s1 K1c2 K 1 r2 K 1 s1 K 1 pp 0 0 K2c2 K 2 r2 K 2 s1 0 K 2 pp 0 K3c2 K 3 r2 K 3 s1 0 0 K 3 pp   (A.11) K n pp =K n c3+K n r3+K n s3 (A.12) K n c1 = kpn   1 0 −sinΨn 0 1 cosΨn −sinΨn cosΨn 1   (A.13) K n c2 = kpn   −cosΨn sinΨn 0 −sinΨn −cosΨn 0 0 −1 0   (A.14) K n c3 = diag(kpn,kpn,0) (A.15) K n r1 = krn   sin2Ψrn −cosΨrn sinΨrn −sinΨrn −cosΨrn sinΨrn cos 2Ψrn cosΨrn −sinΨrn cosΨrn 1   (A.16) K n r2 = krn   −sinΨrn sinαr sinΨrncosαr sinΨrn cosΨrn sinαr −cosΨrncosαr −cosΨrn sinαr −cosαr −1   (A.17) Modal analysis of back-to-back planetary gear... 137 K n r3 = krn   sin2αr −cosαr sinαr −sinαr −cosαr sinαr cos 2αr cosαr −sinαr cosαr 1   (A.18) K n s1 = ksn   sin2Ψsn −cosΨsn sinΨsn −sinΨsn −cosΨsn sinΨsn cos 2Ψsn cosΨsn −sinΨsn cosΨsn 1   (A.19) K n s2 = ksn   sinΨsn sinαs sinΨsncosαs −sinΨsn −cosΨsn sinαs −cosΨsncosαs cosΨsn −sinαs −cosαs 1   (A.20) K n s3 = ksn   sin2αs cosαs sinαs −sinαs cosαs sinαs cos 2αs cosαs −sinαs −cosαs 1   (A.21) Ψsn =Ψn−αs Ψrn =Ψn+αr (A.22) References 1. August R., Kasuba R., 1986, Torsional vibrations and dynamic loads in a basic planetary gear system, Journal of Vibration, Acoustics, Stress and Reliability in Design, 108, 348-353 2. BartelmusW., 2001,Gearboxdynamicmodelling,Journal of Theoretical andAppliedMechanics, 39, 4, 989-999 3. Botman M., 1976, Epicyclic gear vibrations, Journal of Engineering for Industry, 98, 811-815 4. Bu Z., Liu G., Wu L., 2012, Modal analyses of herringbone planetary gear train with journal bearings,Mechanism and Machine Theory, 54, 99-115 5. Cooley C.G., Parker R.G., 2012, Vibration properties of high-speed planetary gears with gyroscopic effects, Journal of Vibration and Acoustics, 134, 061014 6. Cooley C.G., Parker R.G., 2013,Mechanical stability of high-speed planetary gears, Interna- tional Journal of Mechanical Sciences, 69, 59-71 7. Cunliffe F., Smith J.D., Welbourn D.B., 1974, Dynamic tooth loads in epicyclic gears, Journal of Engineering for Industry, 95, 578-584 8. Ericson T.M., Parker R.G., 2013, Planetary gear modal vibration experiments and correla- tion against lumped-parameter and finite element models, Journal of Sound and Vibration, 332, 2350-2375 9. EritenelT., ParkerR.G., 2009,Modal properties of three-dimensional helical planetary gears, Journal of Sound and Vibration, 325, 397-420 10. GuoY.,ParkerR.G., 2010,Purely rotationalmodel andvibrationmodesof compoundplanetary gears,Mechanism and Machine Theory, 45, 365-377 11. KahramanA., 1993, Planetary gear train dynamics, Journal of Mechanical Design, 116, 713-720 12. Kahraman A., 1994, Natural modes of planetary gear trains, Journal of Sound and Vibration, 173, 125-130 13. Kahraman A., Vijayakar S.M., 2001, Effect of internal gear flexibility on the quasi-static behavior of a planetary gear set, Journal of Mechanical Design, 123, 408-415 14. KiracofeD.R., ParkerR.G., 2007, Structured vibrationmodes of general compoundplanetary gear systems, Journal of Vibration and Acoustics, 129, 1-16 15. Lin J., ParkerR.G., 1999,Analytical characterizationof the unique properties of planetary gear free vibration, Journal of Vibration and Acoustics, 121, 316-321 138 A. Hammami et al. 16. Lin J., ParkerR.G., 2000, Structuredvibration characteristics of planetary gearswith unequally spaced planets, Journal of Sound and Vibration, 233, 921-928 17. Łuczko J., 2008, Chaotic vibrations in gear mesh systems, Journal of Theoretical and Applied Mechanics, 46, 4, 879-896 18. Parker R.G, Wu X., 2010, Vibration modes of planetary gears with unequally spaced planets and an elastic ring gear, Journal of Sound and Vibration, 329, 2265-2275 19. Singh A., Kahraman A., Ligata H., 2008, Internal gear strains and load sharing in planetary transmissions: model and experiments, Journal of Mechanical Design, 130, 072602 20. Sondkar P., Kahraman A., 2013, A dynamic model of a double-helical planetary gear set, Mechanism and Machine Theory, 70, 157-174 21. Tanna R.P., Lim T.C., 2004, Modal frequency deviations in stimating ring gear modes using smooth ring solutions, Journal of Sound and Vibration, 269, 1099-1110 22. Toda A., Botman M., 1979, Planet indexing in planetary gears for minimum vibration,ASME Design Engineering Technical Conference,A80-15730, ASME, St. Louis 23. Velex P., Randrianarivo L., Pichon V., Wittman R., 1994, Dynamic behavior of epicyclic trains: experimental and numerical analyses,Proceedings of the International Gearing Conference, Newcastle, 265-269 24. Wojnarowski J.,Kopeć J., ZawiślakS.,Gears andgraphs,Journal of Theoretical andApplied Mechanics, 44, 1, 139-162 25. Wu X., Parker R.G., 2008,Modal properties of planetary gears with an elastic continuum ring gear, Journal of Applied Mechanics, 75, 1-10 Manuscript received April 20, 2014; accepted for print July 12, 2014