Open access
Author
Date
2019Type
- Doctoral Thesis
ETH Bibliography
yes
Altmetrics
Abstract
Electric mobility has substantially gained in importance for the automotive industry for some time. In order to remain successful in global competition, the optimization of the NoiseVibration-Harshness-behavior and the acoustic design of electrical drives define besides a high power density and efficiency significant technical key performance indicators and unique selling propositions in product engineering. A deep physical comprehension of electromagnetic noise emissions and the ability to simulate, to analyze and to have an active influence on them represent crucial prerequisites for electrical drive engineering in order to comply with legal regulations and to generate a sustainable competitive advantage. Hence, in academic research and industrial product development large efforts are made for providing adequate vibroacoustic modeling approaches as well as simulation and analysis tools. Recent vibroacoustic simulation methods for electrical machines allow a numerical prediction in frequency domain by means of idealized, stationary operating conditions. A quite good congruence of simulation and measurement results on test-benches can already be achieved. The underlying numerical procedures are mainly applied to NVH-analyses of electrical radial flux machines for which a simplified 2D- to 3D-finite-element-coupling of electromagnetics to structural dynamics can be used and implemented. However, a highly detailed magnetomechanical 3D-FE- to 3D-FE-coupling for subsequent structural and acoustic computations is not yet possible with an acceptable degree of efficiency on an industrial scale. Required advanced numerical models and procedures such as Galerkin-projection between different physical domains and associated mathematical function spaces are currently only applied in academics. Therefore, from a technical and methodical point of view more complex electrical machine designs such as axial flux machines have not yet been investigated and characterized in industrial engineering in terms of vibroacoustics. State-of-the-art analytical approaches with reduced modeling accuracy and Fourier-based representations of magnetic air-gap forces by spatial modes and temporal orders already allow an efficient computation and root-cause-analysis of operational magnetic structure-borne and air-borne sound in electrical drives. Nevertheless, the usage of 3D-FE-based electromagnetic force density distributions over the stator and rotor as well as over electrical winding geometries has not been conducted within the above framework though this would allow to represent highresolution NVH-calculations and mechanical fatigue issues inside the electrical machine. In addition, the methodical landscape lacks of numerical approaches for system-simulations of electrical drive acoustics. This would in turn define the base for numerical studies of controlbased noise compensation and noise shaping. The same observation applies to time-dependent multibody-simulation of the entire rotor-stator system with consideration of structure-borne sound transmission across bearings. The main reason relies on numerical limitations of available magnetic force excitation models. A magnetic air-gap force model for systemsimulation has first been established in 2013 which takes into account machine-inherent spatial stator current harmonics as well as external disturbances by the inverter and the drive-control.
Based on the sketched methodic strengths and weaknesses of existing modeling, simulation and analysis approaches this work introduces a vibroacoustic modeling environment for electrical drives which merges the advantages of the respective tools. By embedding those methods consequently into a more abstract mathematical framework it is possible to apply the vibroacoustic simulation environment to a vast class of different electrical machine and drive topologies. Overall three NVH-simulation models have been developed and implemented in the course of this doctoral thesis. The determination of structural dynamic and acoustic quantities is carried out with help of modal force response superposition either in frequency domain or in time domain by solving a numerical state-space or multibody model for the entire rotor-stator system. The simulation methods have been applied to all common electrical machine and traction drive topologies in automotive engineering: an asynchronous squirrel-cage induction drive, a permanent-magnetic radial flux synchronous motor for electric vehicles, a disc-rotor axial flux electrical machine and an integrated radial flux synchronous motor for hybrid applications. Comparative operational vibration measurements on real electrical drives illustrate the high informative quality of the numerical results. Thorough investigations on magnetic structure-borne and air-borne sound for various operating conditions are preferably taken out in frequency domain. All significant acoustic orders and levels can be determined straightforwardly. The structural dynamic causes for local narrowband acoustic level exceedances are identifiable within the constructional layout of the motor. Structure-borne sound transmission paths are studied from the source of excitation to the desired structural response. Broad-band level exceedances can be traced back to structurally sensitive spatial excitation shapes and temporal orders and frequency components. The Fouriersynthesis of the magnetic force density based on underlying magnetic field quantities via the Maxwell tensor allows to quantify the influence of critical electromagnetic design parameters on specific spatial and temporal force modes and orders. This forms the starting point for electromagnetic layout improvements with respect to NVH. The magnetic flux density and electromagnetic force excitation of the rotor and stator iron, of permanent magnets and of the electrical winding as well as resulting operational structural dynamic vibrations have been synthesized by generic design-inherent harmonic force and structural mode shapes. Due to the wave character of all involved physical fields, this has been done with help of Fourier- or trigonometric approximations. Corresponding operation point dependent magnetic and vibroacoustic scalar quantities have been determined within a systemsimulation environment for the electrical drive. The numerically complex magnetomechanical force coupling to the motor structure has been performed by using the orthogonality of the generic harmonic field distributions. Based on a compact magnetomechanical model and a time dependent magnetic force model the operational NVH-behavior of electrical drives has been determined efficiently by linear modal superposition of harmonic spatial excitation and structural mode shapes. The vibroacoustic simulation in time domain or frequency domain has been done online within the systemsimulation or as a post-processing step. The implementation of algorithms is conducted in such a way that the numerical procedures are routinely applicable in daily engineering as well as independent from the underlying 2D-, 3D-finite-element or analytical modeling of electromagnetics and structural dynamics. The time-based system-simulation models and the Fourier-related 3D-finite-element magnetomechanical coupling represent considerable methodic innovations for electrical drives.
Finally, a modular, iterative and consistent workflow for an overall vibroacoustic optimization of electrical drives is derived from the CAE-based NVH-modeling approach as both share a common structure. The workflow includes all branches of electrical drive development from NVH-simulation and testing, electrical drive design, electromagnetic design to construction and prototyping. A related NVH-development process is presented according to usual samplephases. As an application of NVH-optimization on electrical drive side, a simulation study including a time-based approach is shown for a simultaneous noise and torque ripple compensation with help of the drive-control. In that sense, a promising approach for exploiting the new research area of drive-inherent acoustic design for electrical drives is pointed out. Show more
Permanent link
https://doi.org/10.3929/ethz-b-000344764Publication status
publishedExternal links
Search print copy at ETH Library
Publisher
ETH ZurichSubject
Electric Mobility; Electrical drives; NVH; simulation; Noise reductionOrganisational unit
03641 - Wegener, Konrad (emeritus) / Wegener, Konrad (emeritus)
More
Show all metadata
ETH Bibliography
yes
Altmetrics