The main objective of this PhD research is the development and application of advanced signal-processing techniques to provide a better understanding of the impact of the design and the control settings on the vibro-acoustic behaviour of modern hybrid-vehicle powertrains. The noise of a powertrain is composed of different components generated by the IC engine (including the combustion
noise, the injection noise, the piston-slap noise, and the turbocharger noise), by the electric motor (including the high-frequency tonal noise, the battery-related noise,
the inverter noise) as well as by the gearbox and the driveline (including the mechanical noise related to the torque, the whining noise of the gearbox, etc.). Advanced vibroacoustic signal-processing techniques will be developed in
order to accurately identify, separate, localize and rank the various acoustic sources under stationary, quasi-stationary and non-stationary operating conditions. The new methodologies applied to the vibration, acoustic, electric and cylinder-pressure signals will be combined and benchmarked against conventional methods, including acoustic intensity measurements, beamforming and near-field acoustic holography.
Key innovation beyond the state of the art: The development of advanced, vibro-acoustic processing techniques for multi-source (non-) stationary operation conditions.
Model correlations and validations for linear models have been thoroughly investigated, reaching a level of maturity where they can be applied in industrial applications. Model correlations for non-linear models, however, are still done
on an ad-hoc basis, without any clear metric to quantify the success. With ESR2 the focus is on the exploitation of multi-step estimation techniques to i) correlate the model and the measurement, while considering the imperfections associated with both, to ii) combine different measurement sets and instrumentation configurations to improve performance, to iii) find generic clear metrics for quantifying the degree of correlation and the validity of the model within a prescribed working range, and to iv) find a model-based pre-test procedure to aid in the correlation process. The focus is on powertrains where the distributed flexibility is dominant and several non-linear aspects are present, i.e., velocity-dependent stiffening effects, Coriolis forces due
to bending loads, load-dependent bearing stiffnesses, etc.
The result will be a pre-test, correlation and validation approach for non-linear systems.
Key innovation beyond the state of the art: The development of a pre-test time domain correlation and a new validation method for nonlinear systems
A major problem when analysing the noise emitted by IC engines is being able to break down the measured sound into the various contributions being emitted from the different sources, such as combustion, injection, knocking, turbocharger, bearings, etc. The traditional approaches are based on signal-processing tools that assume the availability of perfect references on the source of interest (SOIs) and rooted on the simplifying assumption of the stationarity of the signals. The objective of ESR3 is to introduce advanced signal-processing techniques that push beyond these limits. Advanced microphone-array techniques will be used to separate the SOIs by jointly exploiting their spatial and statistical orthogonality. Recent latent-variable models will deal with the scenarios with noisy and possibly correlated references. A key point in achieving the separation will be to properly model the SOIs as angle-time cyclo-stationary processes in order to fully exploit their statistical specificities. ESR3 will also address the separation of effects due to excitation, which are invariant in angle, and due to transfer paths, which are invariant in time. The approach will be based on an analysis of the measurements taken in nonstationary conditions and their benchmarking against simulations from models provided by other ESRs, in particular ESR4 and ESR8. This will find applications in operational TPA. ESR3 is also linked to the theoretical foundations underpinning ESR9.
Key innovation beyond the state of the art: The development of signal processing techniques for separating and ranking sources of noise emitted by IC engines.
Emerging encoder measurements have been investigated for belt, gear and bearing monitoring in power transmissions. The instantaneous angular speed has gained a high level confidence and some original propositions in the modelling-associated phenomena emerge in the literature. Rotating components with a discrete periodic geometry (electrical engine, gears, bearings, timing belts, etc.) generate cyclic excitations that can be efficiently described in the angular domain and for non-stationary conditions. The ESR will develop a generalized model of a complete transmission line in this new framework, from the engine to the wheels. This model will be benchmarked against measurements on a transmission test rig or vehicle in order to calibrate the torsional and non-linear parameters, to define the best location for angular measurements and to investigate potential improvements by passive or active damping components. The angular approaches suggested in both the experimental and simulation research activities will also lead to a new path for drive-transmission characterization (cyclic sources and mechanical transfer paths) through angle-time analysis tools with alternative signals like angular speed, position, vibration and current. These models will provide guidelines for measurement protocols used in TPA and source separation in order to take advantage of the non-stationary conditions.
Key innovation beyond the state of the art: The angular modelling of drivelines for comprehensive analyses of the transfer path in stationary
and non-stationary conditions.
Due to stricter regulations regarding energy consumption, lightweight designs are gaining in importance in the vehicle- development process. However, the negative effects on the emitted noise have to be investigated. In this project, ESR5 will develop a test-based method for the targeted design of gearboxes with regards to lightweight specifications and acoustic behaviour, as well as influences on the complete vehicle. In this way, many limitations will be taken into account. The restrictions on the design space, for example, severely reduce the possibilities for acoustic insulation. Furthermore, an optimised lightweight design can cause impaired acoustic behaviour, due to changes in sound transmission or the shifting of resonances into the operating range. The impact at the complete vehicle level will be analysed using psycho-acoustic metrics and rating algorithms. Current methods generally only consider the overall sound-pressure level or the respective frequency spectrum. Here, psycho-acoustic metrics deliver further information about the impact on human perception. The
developed method will serve as a tool for optimizing the vehicle-development process.
Key innovation beyond the state of the art: A new test-based method for an acoustic evaluation of gearbox designs by means of psychoacoustic metrics.
ESR6 will evaluate the influence of lightweight construction approaches on gearbox acoustics by means of simulations. The downsizing of drivetrains often leads to lighter
and more flexible structures. The higher elasticity of the system’s components directly affects the existing gearmesh excitation and the transfer paths of the gearbox. For example, higher flexibilities of the housing, shafts and gear blanks cause a larger misalignment in the tooth contacts, which can lead to larger system excitations and to worse acoustic performance. The aim of this ESR project is to help identify the risks and capabilities of lightweight designs in terms of the gearbox’s acoustic performance using simulation-based sensitivity analyses and parameter studies. In addition to the investigation of lightweight construction techniques, new production technologies, for example, the additive manufacturing of gear wheels by selective laser melting, as well as acoustically promising materials (e.g., foam structures or meta-materials) will be part of the analysis. The combination of different lightweight approaches will lead to new possible design concepts for gearboxes. The potential and risks of these cutting-edge designs will be benchmarked against the best-available current systems.
Key innovation beyond the state of the art: New simulation-based gearbox-design concepts will be developed, blending lightweight technologies and low noise.
ESR7 will explore advanced signal-processing techniques for feature identification and correlate the metrics to subjective perception. This will be performed on both downsized IC engines and e-motors, since they are characterised by different acoustic signatures. ESR7 will develop and validate the technical metrics for the specific combustion and mechanical noises of IC engines, such as injector noise, combustion irregularity, piston slap, valve ticking, turbocharger noise, chain noise, belt noise, fuel pump noise, etc. Since it is well known that in electric and hybrid vehicles the NVH performance is judged predominantly in noiseperception terms rather than on noise levels, the activity on e-motors will be dedicated to the analysis of specific psycho-acoustic metrics characterising the e-motor sound. In particular, ESR7 will investigate e-motor whining noise (typical high-frequency pitch and tonal component) correlating the electrical data with noise and vibration data. This will be of paramount importance in understanding the NVH performance of the e-motor and drive and, as
a consequence, its impact in terms of human perception.
Key Innovation beyond the state of the art: Development of sound metrics targeted to the e-motor and drive.
ESR 8 will investigate component TPA approaches to predict full-vehicle noise and vibration performance from invariant load measurements on an engine test bench
combined with vehicle-transfer functions. Various load identification procedures will be tested and compared for both structural and acoustic loads (blocked-force method, free-velocity approach, pressure-inversion technique, etc.). Simulation models of different complexity and controlled TPA test-rig measurements will be utilised for the sensitivity analysis and the validation. The industrial applicability will be tested on vehicle-component TPA measurements provided by SISW.
Key innovation beyond the state of the art: Development of a TPA approach that is specifically
tailored to e-motor and downsized-IC-engine vehicles.
The ESR will apply signal processing, microphone-array techniques, in-house expertise and good practices for the vibro-acoustic troubleshooting and analysis of IC powertrains. ESR9 will focus on 3D and quantitative microphone-array techniques and sound-source-separation techniques for a sound-source-contribution assessment of the powertrain components. The approach proposed is to firstly to build a sequential response by associating sequentially blind or quasi-blind signal separation and then localize and quantify the sources on a 3D powertrain mesh for each separated signal component. Secondly, the scientific objective will consist of integrating the signalprocessing step into the microphone-array step by taking into account the signal/source properties in the Bayesian technique. This will lead in to a multidimensional Bayesian technique, taking advantage of the spatial, temporal, frequency, and statistical properties at the same time, for an efficient sound separation of the powertrain components. The ESR will conclude by validating the technique on a vehicle-powertrain test bench under a range of operating conditions.
Key innovation beyond the state of the art: Multidimensional Bayesian technique for efficient sound source separation.
In order to comply with Europe’s stringent emission regulations, combustion ECU control parameters must be optimized for fuel consumption and efficiency. This optimization often results in an increased combustion pressure to improve the engine’s power-to-weight ratio. This then has an impact on the loads exciting the engine block, the transmission and the rest of the driveline. The goal of ESR10 is to explore the best trade-off solutions with respect to ECU parameter settings for improved engine power, ecoefficiency, vibration and noise comfort. This trade-off analysis will be performed by means of multi-attribute 1D simulation models. The optimized control strategies will be implemented and tested on a HIL platform, in close comparison with baseline scenarios.
Key innovation beyond the state of the art: An ECU parameters optimization strategy for enhanced multi-attribute (eco-efficiency-power-NVH) model-based virtual engine calibration.
A knowledge of contact forces and strains in transmission gears and bearings is of paramount importance to gain an insight into the state and the performance of the system. Despite its high relevance, a correct evaluation of the contact forces is rarely achievable with available sensors, such that forward Finite Element (FE) or (flexible) multibody simulations are used to derive these quantities – but only approximately. However, due to the complex contact behaviour of gears and bearings, these quantities are often wrongly estimated and this forces engineers to oversize their designs. ESR11 will focus on the development and validation of a virtual sensing technique based on a stateestimation that will make it possible to extract the internal
forces and strain fields (e.g., at the tooth root or the bearing raceways) thanks to a combination of a few well-positioned sensors (e.g., strain gauges and accelerometer) and advanced gear-contact models. The ESR will contribute to the development of suitable simulation methodologies to be applied for virtual sensing and also to the sensor layout’s selection and final validation. The research is also of high scientific relevance given the challenges of non-smooth state estimation.
Key innovation beyond the state of the art: The development of a sensing strategy that is applicable to contact problems and will allow the development of a tool for the online/offline monitoring of drivetrain systems, including gears, bearing contact forces and stress/
strain concentration factors.
The project of ESR12 addresses the acoustic modelling and testing of electric vehicle drivetrains from noise sources such as the electromagnetics and the gear noise, and
how this all impinges on the driver’s ear. High-frequency structure-borne noise isolation at the coupling of the powertrain to the car body will be investigated. The simulation
and testing of elastomer mounts will be addressed in particular.
Key innovation beyond the state of the art: The development of methods to enable robust concept standards that consider a variety of different trade-offs.
The application context is the global design process for the next generation of electrified powertrains (including hybridized IC engine//e-motor(s), gearboxes, drivelines, etc.). ESR13 will develop advanced vibro-acoustic metamodels, typically using probabilistic, Bayesian inference and other IA-related modelling tools. The metamodel will be based on both experimental (accurate but expensive) and numerical (cheap but uncertain) results. The meta-model function is to infer a probabilistic NVH performance from a set of known design parameters, which are mostly geometrical and related to the dynamic structural properties. The NVH performances will be defined from quantities that can be easily measured and computed, such as the injected structural power, the source equivalent forces, and
the overall estimated radiated power. The end-user target is a knowledge model able to guide the early-stage engineering decision/optimization processes on a rational, scientific and objective basis. The knowledge used to build the meta-model includes an a-priori broadband energy model and both experimental and numerical results, the number of which should increase with time, thereby increasing the meta-model’s performance.
Key innovation beyond the state of the art: New meta-models for the NVH assessment of future electrified-vehicle powertrains.