< Back to previous page

Project

Physically Consistent Vibroacoustic Substructuring involving Continuous Interfaces: Applied to Concept Phase Car Body Optimisation

Over the past decades, in order to keep up with both the legislations governing noise emissions and the growing customer expectations, the continuous search for improved noise and vibration performances has been playing a key-role in the development process of new vehicles. However, to meet these expectations in today’s competitive environment, it is not enough simply to bring a better product into the market. You have to bring it sooner and at a lower cost than your competitors. To keep up with the tight schedules and budgets, automotive engineers need then evermore to account for Noise, Vibration, and Harshness (NVH) aspects right from the start of the vehicle’s development process.

In this context, substructuring approaches together with Computer-Aided Engineering (CAE) methods have proven to be invaluable tools to predict NVH performance characteristics already from the first phases of the development process. On the one hand, CAE simulations allow engineers to predict the consequence of any design modification on the real-world performances of the final vehicle by using virtual models only. On the other hand, substructuring approaches allow to break the whole vehicle into its constitutive parts (frame, powertrain, tires, etc.) and separately analyse their contributions to the overall interior and pass-by noise levels. Deployed together and effectively, both tools can be used to improve the vehicle NVH design through multiple iterations, allowing to quickly assemble virtual and experimental models of single components and providing data to guide the design process from its earliest stages, while reducing the need for expensive full-system physical prototyping and labour-intensive experimental testing.

Although substructuring involving Finite Element (FE) and experimental models seems to be the winning strategy, yet this method may encounter difficulties when applied to complex systems such as vehicles. First of all, the presence of large interfaces (as for instance between the vehicle floor and the upper parts) poses significant challenges to the coupling process. A large number of interface Degrees of Freedom (DoFs) must be defined at the boundary of the two connecting structures, and their position must be coincident from both sides (condition of compatible interfaces). This implies that in principle the coupling of vehicle’s components coming from different models, useful in an early development stage to explore new design solutions, would require a post-processing step in order to make the interface nodes match, a very time-consuming operation which in most cases must be carried out manually.

Within this framework, a novel substructuring approach that tackles continuous interfaces by only considering a small set of coupling DoFs is presented. This approach employs a discretisation of the interfaces between structural parts or structural-acoustic domains in terms of pivotal points and patches. In this manner, the requirement of spatial continuity between the component models is no longer needed and subsystems with incompatible interfaces can be coupled. For an appropriate interface description by means of point and patches, an inverse approach is defined as such to guarantee that the inversely obtained subsystem model possesses physical properties of the underlying component. The potential of the presented substructuring approach is explored by considering a simplified vehicle model. It will be shown that a single component can be successfully coupled with different neighbouring parts, even when they have a different location of coupling DoFs. Given the possibility to assemble components having incompatible FE meshes, NVH engineers can quantify the contribution of design changes or new configurations to the overall cabin noise during the first modelling phases. This will efficiently support car manufacturers in cutting down the design costs and boosting the degree of diversification of the car market.

Building upon the above inverse characterisation approach, a methodology that enforces the aforementioned physical properties on a test-based model is developed. When an experimental model is indeed included in a coupling scheme, a proper data acquisition is required, which in many cases may be difficult to achieve due to environment and testing conditions. Small inaccuracies in the measured subsystem model can drastically affect the prediction of the assembled system dynamics, jeopardising the success of the whole coupling scheme. Since a detailed knowledge about the types of errors introduced by the measurement procedure is in general not available, the recovery of the ‘true’ subsystem models is not possible. What is possible, nevertheless, is to define requirements that the measured subsystems must a priori meet in order to be physically consistent: reciprocity and passivity. The solution proposed in this dissertation is therefore to enforce such physical properties on the measured subsystem models during a post-processing step. Substantial improvements in the prediction of the dynamic behaviour of the assembled system are hence achieved. 

Date:17 Sep 2018 →  13 Oct 2022
Keywords:acoustic, sub-structuring
Disciplines:Control systems, robotics and automation, Design theories and methods, Mechatronics and robotics, Computer theory, Manufacturing engineering, Other mechanical and manufacturing engineering, Product development
Project type:PhD project