The characteristics of noise, vibration and harshness (NVH) are crucial aspects in the design of vehicles and, especially, of electric ones. In this ambit, one of the primary targets is understanding the behaviour of electric motors, as well as the study of the interaction among the various vehicle components and the overall performances in terms of system. In this context, severe analyses and measurements become necessary, such as, for instance, the analysis of vibrations, the noise measurement and the modelling of the system dynamics. Such instruments are essential to identify and to solve potential NVH problems, ensuring quality, performances and comfort to passengers. In the electric vehicle, the absence of the typical noise of the internal combustion engine significantly alters the overall acoustic characteristics. Other noise sources are in fact highlighted, like the one generated by the transmission and by the wheel-road contact, needing a particular attention to the effects of these noise sources on the driver. Moreover, the different load distributions in electric vehicles determine new responses and excitation frequencies. The noise of electric motors, mainly deriving from electromagnetic forces and from the high-frequency noise generated by inverters, has a significant impact on the overall NVH features. For the design and the development of electric vehicles, a NVH approach of holistic type becomes therefore necessary, considering the complex characteristics of noise generation and propagation in electric vehicles’ powertrains, which require multidisciplinary know-how to face at best the above-mentioned challenges. The transmission is then modelled and analysed in its whole and not only at level of single components. In this context, we emphasize the analysis and the study of those complex interactions and dynamic behaviours that characterize the whole transmission system.
The primary factors that influence NVH in electric powertrains
In electric vehicles’ powertrains, the main sources of noise generation and of vibrations are the electric motor, the inverter, the transmission with gearbox and, if present, the cooling fan and pumps (figure 1).
The components highlighted in figure 1, if considered individually, would potentially have an acceptable NVH, however, criticalities emerge in case of mutual interaction. We are hereunder examining in detail the typologies of noise generated by each single part and the relative causes, as follows:
- Transmission and reducer (gearbox): mechanical noise generated by vibrations caused by the dimensional tolerances of the gearbox (TE) mechanically stressed.
- Electric motor: it generates excitations through electromagnetic forces causing the so-called “whistling”.
- Inverter and power electronics: they include components that contain higher-order harmonics, connected with the PWM (pulse width modulation) of the signal.
- Cooling fans and pumps: they generate noise connected with fluidic phenomena described with aerodynamics laws.
The combination of these elements is responsible for the vast majority of the overall noise generated by electric vehicles, which includes both the “aerial” and the structural noise.
Sources of noise and vibrations, like the electric motor, the gearbox and the power electronics, also act as sources of mechanical vibration. Vibrations and noise are propagated through various pathways: structural vibrations propagate through the vehicle structure, while the noise transmitted by aerial pathway propagates through the air. Such pathways play a crucial role in the way in which sounds and vibrations are perceived by the human ear, influencing the overall sensorial experience and the comfort inside the vehicle (figure 2). Considering both excitation sources and transfer pathways is essential to minimize NVH impacts.
In the table 1 we are analysing in-depth the primary factors that contribute in the noise generation in electric vehicles.
| The factors that influence noise and vibrations in electric vehicles’ powertrains are manifold. They include the dynamics of gears and, in particular, the transmission error. The latter is in its turn affected by a series of elements, like the macro and micro geometry of toothed wheels, the changes in their rigidity, the misalignments in assembly phase and manufacturing tolerances. Such features induce vibrations that are transmitted through shafts, bearings, casings and other components, sometimes exciting structural resonances. Therefore, a noteworthy contribution to NVH comes from the noise generated by bearings, by the unbalances of shafts, by misalignments, by eccentricity, by contact interactions and by clamping defects. To minimize the mechanical noise and to optimize the overall NVH performances of an electric vehicle, it is therefore essential to include a precise analysis of the causes and of transmission pathways of the vibrations induced by the transmission error. |
| The electromagnetic noise in electric vehicles’ powertrains is another significant factor concerning the NVH features of the vehicle and it is remarkably influenced by the interactions inside the motor itself, in particular in the case of PMSM (Permanent-magnet synchronous motor). Electromagnetic forces in the air gap, between the stator and the rotor, can cause the so-called “whistling”. These forces determine radial, axial and tangential components, causing various vibrations at the stator and at the rotor. In particular, the harmonics associated to radial forces mainly generate vibrations of the stator, whereas tangential forces may induce torsional vibrations in both the stator and in the rotor, featuring high NVH impacts. The oscillations in these frequencies, from the stator to the housing, can amplify the structural vibrations and then also determine acoustic effects. Such vibrational interactions, although complex and hardly forecast, are the essential mechanisms for the generation of the electromagnetic noise in the elements of the transmission chain of an electric vehicle. Luckily, they can be at least partly controlled. |
| A further source of electromagnetic noise, in addition to electronics, is represented by the inverter. This is mainly due to the noise of the PWM (pulse width modulation), more known as “switching” noise. The inverter, as known, contributes in the conversion of the direct current into alternating current, by using precisely the PWM strategy for a fast switching. This process generates higher-order harmonics, which influence electromagnetic forces, negatively affecting the NVH. A significant design challenge for inverters consists in finding a balance between the reduction of such harmonic excitations, which can limit the overall efficiency, taking into account that attempts to minimize NVH effects on the motor can affect its overall efficiency. This needs a thorough analysis of NVH features and of the inverter since the design phase. |
| The aerodynamic noise and the fluid flow are further key factors that influence the NVH characteristics in electric vehicles’ powertrains. In particular, the cooling fan contributes in the vast majority of the aerodynamic noise generated in the motor compartment. The primary noise is created by vent holes in unsealed motors and by external fans in sealed motors. This noise type includes broadband noise, especially at the blade pass frequency and its higher harmonics. In addition to the overall NVH profile, there is the noise generated by the fluid flow, in particular by the water pumps used for cooling. The latter may be noisy in operational conditions, further influencing the vehicle’s acoustic environment. |
| The factors that influence the noise of electromagnetic components concern the following aspects. Mechanical deformations and vibrations deriving from slot design, winding distributions, distortions of the current waveform, air gap variations, eccentricity of the rotor and phase unbalances. These aspects contribute in mechanical deformations and vibrations through complex harmonic forces and torque. Stator-frame resonance: the structure of the stator frame acts as radiator of primary noise of the machine. The resonance can occur when the frequency of the radial force aligns with the natural frequencies of the stator frame, determining a significant noise. Magnetostrictive noise: it is due to the elongation and to the periodical shrinkage of the core material, which, in high-power applications like in the case of electric and hybrid vehicles, can fundamentally share in the overall noise. Parasitic oscillation torque: in inverter-powered motors, parasitic oscillation torques arise from time harmonics in the stator currents and can be accentuated by voltage irregularities. |
Techniques for the NVH optimization in electric vehicles
During the development of an electric vehicle, satisfying both the regulations concerning the acoustic pollution and customers’ requirements is essential for manufacturers. The different measures of noise reduction must be applied to all the components of the transmission chain. Adoptable strategies for a NVH optimization include advanced design techniques, large-scale precision engineering, porous and elastic insulating materials as well as innovative approaches for specific requirements.
Simulation models for the NVH analysis
The typical powertrain of an electric vehicle is generally made up by an electric motor combined with a reduction gear. To foresee with precision the noise generated by the powertrain in its whole, it is necessary to consider the dynamic performances and the acoustic characteristics of both the electric motor and of the reduction gear as a whole, as well as the modalities with which these two primary components mutually interact. In figure 3 we report a computational model ideated to simulate the powertrain’s behaviour under various conditions and to foresee the noise generated.
Electro- magnetic simulation (EM)
This type of simulation is aimed at determining the excitations generated by electromagnetic forces that, in electric motors, represent the main source of NVH problems. Such forces are generally estimated by means of analytical or numerical calculations, implemented in specific software to solve the differential equations at the base of the model that describes the electric machine. Such equations consider the geometry and the electromagnetic non-linearities of the machine. The response is then analysed by using structural 3D FEA (finite elements analysis) or MBD (multi-body dynamics).
The vibro-acoustic simulation of an electric machine is based on a co-simulation, generally between an application that carries out the electromagnetic analysis and a solver based on the Finite Element (FE) method that defines the structural behaviour. This requires that the structural FEA is solved at each time step. To design and to analyse electric motors, actuators, sensors, transformers and other electromechanical devices, specific simulation softwares of the electromagnetic field are used such as, for instance, “ANSYS Maxwell”. The goal is simulating electromagnetic fields and quantifying the generated force and the consequent vibrations in electric motors. The software accurately captures the electromagnetic interactions that take place in a motor, taking the stator, the rotor, windings and magnetic materials into account. The combination of the results obtained with the data deriving from electromagnetic simulation tools is fundamental to understand how electromagnetic forces influence the NVH level in the powertrain group in general. Simulation data, for instance, can be imported into multi-body or numerical analysis tools to evaluate vibration and noise dynamics caused by forces in the motor and in neighbouring zones.
Furthermore, simulations allow forecasting and visualizing the entity and the distribution of electromagnetic forces in the motor. If such forces are unequal or interact with mechanical resonances, significant vibrations may occur. The data about vibrations generated by the electromagnetic analysis are then used in acoustic simulations to understand how vibrations manifest themselves in the form of audible noise, contributing in the vehicle’s NVH profile. Moreover, electromagnetic simulation tools are used to guarantee satisfactory performances already in design phase. According to modifications in the geometry of electromagnetic components, in the configuration of windings and in the properties of materials, it is possible to reduce the undesired vibrational behaviour and to improve NVH performances of electric motors. It has been demonstrated, for instance, how the optimization of sizes and of the magnet position in PMSM has implied a significant reduction of the torque oscillation from 46% to 16%. This highlights the importance of a precise electromagnetic modelling and the efficacy of advanced optimization techniques in reducing the vibratory behaviour and in improving the motor’s overall performances.
Multi-body dynamic simulation by means of flexible bodies
At the beginning of a dynamic multi-body simulation, the electric motor is a critical component of the powertrain and it is included in the system model in order to draw and to describe all its complex dynamics. Furthermore, this simulation aims at facing the impact of the electric motor on all other components of the powertrain, such as toothed wheels and shafts, analysing the rotational dynamics, the torque generation and any flexibility in the supports or in the couplings of the motor. The simulation extends beyond the normal dynamics of the stiff body and integrates bodies’ flexibility, emulating the effective dynamic behaviour of motors. The interactions among dynamic loads, the vibrations of gears and contact forces are represented with precision by a corresponding mechanical model (figure 4).
Structural analysis
In the structural analysis phase, the attention is paid to the loads and the specific stresses of the electric motor, such as electromagnetic forces, mechanical stresses deriving from the torque generation and vibrational forces. Using finite element methods, the structural integrity of the motor is analysed and its impact on the overall structure of the powertrain group, such as the envelope or assembly points. This stage is fundamental to understand how vibrations originate and propagate. It has been proven that specific harmonics of supply currents can significantly influence the vibratory behaviour of electric machines. This underlines the importance of an accurate acoustic modelling and the efficacy of advanced noise control strategies.
Calculation of acoustic radiations
The calculation of acoustic radiations is faced by means of an acoustic model that represents the transformation of these vibrations into sounds (figure 5).
The attention is paid to motor vibrations, especially to those deriving from electromagnetic forces and from the interactions of toothed wheels, which contribute in the overall noise of the powertrain. Today, active control systems of the road noise have achieved an average noise reduction of approximately 3.4 dB in the position of the driver and of the right rear seat for maximum frequencies of 400 Hz.
We are in-depth analysing below the different phases of a typical process for the NVH analysis and simulation in the development of powertrains for electric vehicles. We specify that some steps and tools can vary depending on the specific case. 1) Definition of simulation targets and requisites, focusing on aspects such as the identification of the noise source, the assessment of the noise level and strategies for its reduction. Identifying clearly the main performance parameters and indicators, including NVH metrical specifications. 2) Collection and drawing up of necessary data. Such data can come from various sources and concern the physical properties of the powertrain group, the operational characteristics and environmental factors. 3) Development of detailed models of the powertrain group’s components through FEM and by integrating the MBD modelling. It is necessary to include flexible components such as housings of the motor and of the gearbox, shafts, bearings and contacts of gear teeth. 4) Integration of electromagnetic simulations with mechanical models for a complete analysis of vibrations. 5) Development of acoustic models by means of methods such as BEM (Boundary Element Method) to obtain forecasts about the generation of soundwaves starting from the powertrain’s vibrations and to define the acoustic environment. 6) Independent validation of each model regarding the considered component versus experimental data to check its accuracy. 7) Integration of single components’ models to obtain a single model of the powertrain, for an accurate representation of interfaces and of dynamic interactions. 8) Simulation of various operational scenarios to understand NVH dynamics under different operational conditions. 9) Analysis of the results of the NVH simulation: using the equivalent radiated power (ERP) to detect the energy released under the form of sound; the sound pressure level (SPL) to quantify the acoustic energy perceived and Campbell diagrams, both 2D and 3D, to visualize the response in frequency and to identify critical speeds. Evaluating the single modalities in both the time domain and in the frequency one for a detailed analysis. 10) Perfecting of the model, on the basis of the NVH analysis, to increase its precision, adjusting the properties of materials, boundary conditions or geometries. 11) Iterative tests and optimization: iterating the simulation process, adapting the model according to NVH results and carrying out again the tests to reduce the noise or to make design improvements. Final validation and reporting: validating the final model versus known data or experimental results. 12) Interpretation of NVH results to understand the acoustic behaviour of the powertrain group. Using analyses in the time and frequency domain to identify and to assess specific vibration modalities and their NVH impact. Analysing the results of stress and displacement of the envelope to support the structural optimization in order to decrease the noise and to improve the durability. 13) According to the complete NVH analysis, making design recommendations aimed at reducing noise and vibrations, meanwhile improving the overall acoustic quality of the powertrain. Proposing eventual modifications concerning the latter’s design.
During these phases it is useful to adopt a holistic approach, essential for an efficacious noise reduction and, definitively, to improve the acoustic quality. The process described combines a detailed modelling of the powertrain’s components, flexible elements and forefront acoustic methods.
In figure 6, a process for the setup of an acoustic analysis of the powertrain of an electric vehicle is schematized.
Future scenarios
The fast diffusion of electric vehicles introduces distinct NVH challenges issued by the typical lack of noise of internal combustion engines, making the transmission noise louder. As analysed, various factors in the generation of the NVH profile in electric vehicles’ powertrains persist: from electric motors to inverters and to gear systems.
We have underlined in this article the importance of integrating simulation and modelling techniques, including the multi-body (MBD) dynamics and the finite element method (FEM). The holistic approach allows a more in-depth and complete analysis of the complex interactions and of the dynamic behaviours inside the whole transmission system, providing new intuitions about the NVH management. Furthermore, we have presented a workflow that aims at being a guide and at offering a practical and structured approach to the development and the analysis of models for the NVH management in electric vehicles’ powertrain groups.
The noise of the electric motor derives from electromagnetic forces, while inverters produce high-frequency noise owing to the pulse width modulation (PWM). NVH requisites are further complicated by mechanical problems like gear failures and shaft unbalances, which increase the undesired frequencies. Specific simulation software for the electromagnetic analysis and the multi-body dynamics aid in forecasting with precision and in minimizing NVH problems, granting a powerful and efficacious design of electric vehicles’ transmissions.
In brief, facing the NVH in electric vehicles needs to integrate advanced simulation techniques with an understanding of the dynamic interactions inside the powertrain group. A systemic approach is fundamental to make electric vehicles more silent, more comfortable and more efficient. In the future, further researches might explore the integration of advanced materials and innovative design techniques to further reduce the NVH in electric vehicles’ powertrains. Moreover, the development of more sophisticated simulation models able to capture the interactions among various NVH sources in real time could significantly improve the predictive accuracy.
Emerging technologies, like the artificial intelligence and the automatic learning, offer promising solutions to optimize NVH management strategies. Such technologies might be used for both the analysis of big data and in simulation and test phase, to identify models, to develop and to optimize NVH management solutions. Furthermore, enlarging the scope of the NVH research, including the impact on the vehicle’s overall performances and on passengers’ comfort, taking psycho-acoustic factors into account, we could obtain a more complete understanding of the NVH survey in electric vehicles.
In conclusion, although significant progresses have been made in the understanding and in the management of the NVH in electric vehicles’ powertrains, the constant research and innovation are essential to keep pace with the fast progresses in technology. The way consists in the unceasing improvement, with the target of granting the development of more silent, more efficient and more comfortable electric vehicles.
(by Giorgio De Pasquale, Department of Mechanical and Aerospace Engineering, Smart Structures and Systems Lab, Politecnico di Torino and Elena Perotti, Senior data analyst)
 are crucial aspects in Evs){kind=link}