Acoustics: reconciling lightweight construction and NVH
Quality can be heard. Most people associate the quality of devices, machines or vehicles with their acoustics. How does the engine sound when accelerating? Particularly when it comes to engine sounds, acoustics or acoustic feedback is quite desirable or even perceived as a quality feature. Vibroacoustic properties of vehicles have gained significantly in significance in recent years and play a central role in the customer's purchase decision.
Typically, however, noise and vibration phenomena in the vehicle interior tend to be perceived as annoying.
In this context, it is becoming increasingly important to evaluate and influence vibroacoustic behavior at an early stage of vehicle development.
At the same time, a clear trend toward high-strength steel and lightweight construction can be observed in the automotive industry. Lightweight construction concepts pursue the goal of reducing the energy consumption of vehicles and thus ensuring that the exhaust emission regulations specified by the EU are met. However, since the resulting vibrations are less reduced by the lighter structure, the use of lighter materials also has a significant impact on the vibroacoustic behavior of the vehicle. A central question is therefore how lightweight construction can be reconciled with NVH (noise, vibration, harshness).
How can vibroacoustic behavior be predicted? Possibilities and limitations.
FEM, BEM & SEA
To determine this usually requires very complex acoustic models.
Numerical methods such as the finite element method (FEM) or the boundary element method (BEM) are widely used to predict vibroacoustic behavior at low to medium frequencies or in the time domain.
At higher frequencies and for large and complex engineering systems, the wavelength is short compared to the overall system under consideration. The density of eigenmodes also increases considerably. To be used effectively, the FEM in this case requires a large number of finite elements, which entails an increase in computational cost.
Lower costs can be achieved by using BEM. Here, 3D structures are reduced to 2D surface models and thus simplified. However, a major drawback is the smaller scope of possible numerical solutions due to the strong simplification. In addition, both FEM and BEM can be extremely sensitive to parameter deviations. Therefore, statistical methods such as Statistical Energy Analysis (SEA) are mainly used for simulations in the high frequency range.
For SEA, a system is divided into several coupled subsystems and the acoustic behavior of each subsystem is described by a defined number of equations. Although the number of equations to be solved in SEA is comparatively small, the corresponding models cannot be derived directly from the CAD data and the modeling requires high and application-specific expertise. In addition, SEA does not provide information on the spatial energy distribution and thus effects such as damping or structural excitation cannot be described locally.
EFEM - Calculation based on energy density
As an alternative approach, the energy flow analysis (EFA) was developed, which describes the energy distribution in terms of the area-average energy density. The central energy balance of the EFA was transformed in the further course into a partial differential equation, in which a similarity of the propagation of acoustic energy to heat conduction is exploited. On this basis, the energy density can now be calculated using existing FEM methods - the EFA becomes the energy-based finite element method (EFEM).
While conventional FEM is based on displacements, EFEM is based on time- and location-averaged energy densities. Thus, calculations can be performed with a relatively high accuracy even in the higher frequency range. Due to the low discretization effort it is possible to simulate even large and complex structures like complete vehicles or ships and still consider local effects.
In contrast to SEA, it is not necessary to limit the damping between the subsystems and the coupling strength when defining the EFEM. This allows the execution of detailed analyses, for example for the point-exact definition of the external loads, the consideration of an arbitrarily distributed damping or the analysis of the frequency-dependent and also spatially distributed results.
The underlying energy equations are set up analogously to the FEM on an element basis. Since these elements or subsystems are much smaller than in SEA, they allow finer modeling as well as more detailed prediction of the energy fluxes and distributions within the structure under investigation. However, due to the energy-based approach, a much coarser discretization is possible compared to conventional FEM/BEM, allowing larger structures to be investigated even in the high-frequency range. Various application examples with promising results show the great potential of EFEM for the observation of large structures.
EFEM reaches its limits when calculating very small components. Basically, the shortest distance of the considered component should be at least 2.47 times the wavelength to obtain a reliable result. If smaller components are treated as part of a larger structure, a modeling deviating from the EFEM standard must be resorted to.
*** Translated with www.DeepL.com/Translator (free version) ***
Would you like to learn more about EFEM or even other ways to calculate technical systems?
In the fields of vibration engineering, fluid dynamics, electromagnetics and especially acoustics we cooperate with our partner Novicos. With the development of multiphysical concepts and models for vibration reduction, through extensive measurements and the creation of computer-aided simulations of complex processes, we support our customers in a wide range of industries.
We are also happy to arrange a non-binding consultation appointment or a telephone consultation.
Phone: +49 (0)7321 9343-0
E-Mail: info@merkle-partner.de
Of course, you are also welcome to browse our website:
www.merkle-partner.de
PS: We thank Novicos for the content contribution to this newsletter.