Certain operating conditions such as fluctuation of the external torque to planetary gear sets can cause additional sidebands. In this paper, a mathematical model is proposed to investigate the modulation mechanisms due to a fluctuated external torque (FET), and the combined influence of such an external torque and manufacturing errors (ME) on modulation sidebands. Gear mesh interface excitations, namely gear static transmission error excitations and time-varying gear mesh stiffness, are defined in Fourier series forms. Amplitude and frequency modulations are demonstrated separately. The predicted dynamic gear mesh force spectra and radial acceleration spectra at a fixed position on ring gear are both shown to exhibit well-defined modulation sidebands. Comparing with sidebands caused by ME, more complex sidebands appear when taking both FET and ME into account. An obvious intermodulation is found around the fundamental gear mesh frequency between the FET and ME in the form of frequency modulations, however, no intermodulation in the form of amplitude modulations. Additionally, the results indicate that some of the sidebands are cancelled out in radial acceleration spectra mainly due to the effect of planet mesh phasing, especially when only amplitude modulations are present.

The transfer path analysis (TPA) in terms of sound pressure has been implemented for decades in many application areas, such as car, train and construction machine. In this article, we propose a transfer path analysis where particle velocity is employed as the measure of TPA. Sound pressure is a scalar quantity, while particle velocity, which is the other fundamental quantity of sound, is a vector quantity. The phase differences among particle velocity vector components have to be generally considered. For TPA, not only the six degrees-of-freedom of each path motion, but also the three degrees-of-freedom of the particle velocity at the receiver location have to be considered together for an effective path rank ordering. We first propose the formulation of the particle velocity transfer path analysis where the same formulation of the standard sound pressure transfer path analysis is assumed to hold true for each direction of particle velocity. In order to verify the proposed particle velocity transfer path analysis, we carry out an experiment using a simple test box structure. As a result we have found that the error in the particle velocity vector synthesis is acceptably small, and is as small as the error in the standard sound pressure synthesis, which indicates that the same synthesis method can be employed. We then perform rank ordering of the particle velocity transmission paths. Here, a simple method of path rank ordering is applied. Lastly, we briefly discuss sound energy as a measure of TPA.

Due to the relatively tiny length scales involved, complex acoustic flows are not always suitable for traditional CFD codes. This paper develops a robust, semi-analytical numerical method for predicting sound fields based on the calculation of the compact Green’s function over a grid of source-observer positions. These calculations often involve singular functions, hence variations of the method are applied to several different 2D problems to investigate the impact of any singularities on the solution. The effect of grid point density and other parameters on execution time and accuracy are explored including the effects of approximating curved geometries using a number of straight lines. Comparison to known analytical solutions for the 2D problems is used to assess the accuracy of the method. For a typical application, we compute the far-field sound generated by a simple source in the vicinity of a compact, ‘2D’ fan blade in a duct. The current method demonstrates calculation of the compact Green’s function both accurately and robustly by avoiding the mapping from unbounded domains and the evaluation of potential models containing singularities. Both are seen as sources of error which have a widespread impact.

This paper addresses the construction, measurement, and analysis of a double-panel active partition (DPAP) and its accompanying analog feedback controllers. The DPAP was constructed by attaching an aluminum cone loudspeaker at each end of a short segment of a circular duct. Two analog feedback controllers were designed and built using the measured frequency response function of each panel. Two independent (decoupled) feedback controllers were then used to minimize the vibration amplitude of each panel in the presence of an acoustic disturbance. A normal-incidence transmission loss measurement system was used to assess the performance of the DPAP and of a single panel passive partition. Error signal attenuations show that it is both feasible and effective to simultaneously control both panels with decoupled feedback controllers, and that simultaneously controlling both panels of the DPAP has a distinct advantage over controlling a single panel. The reduction in vibration amplitude across the surface of the transmitting panel was confirmed with scanning laser vibrometer measurements. Transmission loss results were obtained for two passive and three active configurations. The average normal incidence transmission loss over the active measurement bandwidth (50–1,000 Hz) for the active double-panel was 60 dB. This is an average of 39 dB more transmission loss than a passive single panel partition.

In this paper we develop a data based model, design a high performance robust controller, and apply the controller in real-time to reduce narrowband acoustic noise from a cooling fan. A custom, portable enclosure houses the cooling fan. One end of the enclosure is fitted with four speakers and four microphones, inside of a short duct, connected to a data acquisition system and a personal computer. Passive materials mounted at the other end of the enclosure reduce backside noise. The frequency of the narrrowband noise is assumed unknown and therefore a control design that can be updated in realtime is needed. The control design that is presented uses a nominal model and a nominal controller. The nominal controller is enhanced by using closed loop signals and taking into account the modeling error. The end result is a data-based method for updating a nominal controller to improve performance.

Beamforming in reverberant environments is important to locate and quantify noise sources in turbofan engine nacelles, automobile interiors, factories, and architectural settings. In order to validate and explore the limits of this approach, a beamforming experiment was conducted in a reverberation chamber using a 32 channel planar phased array and a B&K sound power source. In the reference configuration, the source was located 2 m from the 1.1 m diameter array, and neither was close to the chamber walls. A less-demanding case was constructed by adding some foam absorber to the room to reduce reverberation. A difficult case resulted when the source was placed 5 m from the array, about 1 m from a corner of the chamber. Conventional frequency-domain beamforming with diagonal deletion was applied. The sound source was accurately located at the 2 m distance, with and without the added absorber. In the 5 m case, the sound source could be located at only a few frequencies and only when the processing bandwidth was increased from 48.8 Hz to 781 Hz. Processing individual eigenvectors of the CSM separated the direct and reflected source in 5 m case. The error in the deduced broadband sound power was 0.72 dB in the baseline case, 1.47 dB with the added absorber, and 5.41 dB with the speaker in the corner. Application of CLEAN-SC did not improve the accuracy of the corner results. Use of a Green’s function that attempts to account for reflections was ineffective. It is concluded that beamforming in highly reverberant environment with the free space Green’s function is practical, provided the array is designed and positioned correctly for the environment and source location.