2.

3 Dynamic models of complete gearboxes

Numerous researchers have modelled complete gearboxes in order to predict gear noise. Campell et al. [33] used finite element dynamic modelling methods to predict gear noise from a rear wheel drive automatic transmission. The model was used to investigate the effects of different components inertia, stiffness and resonance. The ring gear and shaft resonances and the tailstock housing stiffness were found to be significant design factors that influenced the gearwhine. Model construction issues were discussed as well as correlation of predicted gear noise traces with operating measurements. Ariga et al. [35] described a systematic approach to reduce the overall gear noise from a four speed automatic transaxle. The vibration characteristics were identified by finite element analysis. A new gear train structure that would be effective in reducing gear noise was investigated. The effect of the modifications was verified experimentally, and the gear noise level was reduced substantially. Also changes in stiffness of the transmission case, at locations supporting the gear train bearings, were shown to affect the gear noise. Dynamic models of typical automotive gearing applications using spur, helical, bevel, hypoid and planetary gear sets were developed by Donley et al. [36]. Basic formulations used in modelling different types of gears were discussed. These models were designed for use in finite element models of gearing systems for simulating gear-whine. A procedure for calculating the dynamic mesh force generated, and gear case response, per unit transmission error was proposed. A simplified automotive transmission was analysed to demonstrate the features of the proposed gear noise reduction technique. A basic approach to gearbox noise prediction was described by Mitchell [38]. The proposed method was dynamic modelling of the gearbox from inside out. The computational strategy for the determination of the dynamic response of the internal gearbox components was described. Force coupling between gears and dynamic coupling due to for example unbalance was discussed. The transfer matrix approach was used for the analysis and a benchmark example was presented to verify the calculation method. Hellinger, Raffel and Rainer [40] used numerical methods to calculate gear noise from a transmission. They used finite element analysis to calculate natural frequencies and forced vibrations of the gearbox structure (housing). As input for the FE calculation of forced response, they used the dynamic bearing forces of the shafts in the gearbox, calculated by a MBS (multi-body system) software. Finite element analysis was used by Nurhadi [41] to investigate the influence of gear system parameters on noise generation. A direct time integrating method was used to predict sound generation, transmission and radiation from mechanical structures. Joachim, Stoffels and Troska [46] optimised the gear noise from a car gearbox. They calculated transmission error and time varying stiffness of the gear mesh by using a finite element based computer program. The results were used as input to a torsional vibration model of the power train (from the engine to the wheels). The output from the torsional vibration model was mesh forces, which were transformed to the frequency domain and used as input to a finite element model of the gearbox, which was used to predict the forced response. As a result of the simulations, a modified gearbox was tested in a car and the gear noise was substantially reduced.

17

2.4 Experimental investigations

There are also some experimental studies on gear dynamics. For example, Terauchi et al. [34] investigated the dynamics of straight bevel gears. A power circulating gear test rig equipped with contactless inductance type accelerometers were used to measure the vibrations of straight bevel gears in three directions. Displacements were shown to depend on the bending stiffness of gear shafts and bearing stiffness. The natural frequencies for torsional vibrations of the gearing system and the natural frequency for bending vibration of the gear shaft also influenced displacements. The vibrations of high frequencies (gear mesh frequency and multiples thereof) were shown to appear in axial direction. Remmers [37] studied the dynamical system of rear axle gears and predicted the existence of a vertical resonance of the pinion. The frequency, predicted with a simple single degree of freedom mass spring model, was close to the frequency of the observed noise peak. Experimental studies of a variety of rear axles confirmed the existence of a peak in vibration and corresponding observed noise. The causes and effects of amplitude modulation in rear axle gears were analysed. Pinion motion response spectrums obtained from rear axle assemblies and hypoid gear test machines were shown to relate gear quality measurements to individual gear manufacturing errors like runout and heat treatment distortions. Also Runge [45] discussed the influence of resonances on gear related noise from axles in vehicles. He showed experimentally that gears with larger errors could be quieter than gears with smaller errors and proposed the use of gear deviations to dampen resonances. Abe and Hagiwara [48] reduced rear axle gear noise in a car, by adding an inertia disk at one sideflange of the final drive, making the dynamical system non-symmetric. This had the effect of decreasing the vibration level of the hypoid gear itself and changing the resonance frequency of the driveline. They also decreased the noise level by decreasing the diameter and stiffness of the pinion shaft. The effect of tooth surface roughness on gear noise and gear noise transmitting path was investigated by Ishida and Matsuda [42]. They used a disk machine in which rolling and sliding contact were generated between two disks. The results were compared with results from a gear system in a gear noise testing machine. The influence of bearing positions on sound radiation from a single stage spur gear system was investigated by Zhou et al. [44]. They showed that the bearing offset, from the centre of the housing wall, considerably influenced the vibration and radiated noise. The radiated sound power increased with the offset.

18