The source-transfer-receiver model to approach automotive NVH problems has proven its worth over the last decades. The approach allows splitting up an NVH problem into a source, for example engine vibration or road induced wheel vibration, a transfer system, for example the car body or car suspension, and a receiver such as the driver ear or steering wheel feeling. The analysis of such a system is called Transfer Path Analysis (TPA).
Whereas the determination of the transfer system for a TPA analysis through frequency transfer functions or a set of modes is fairly straightforward, the source side can pose quite some difficulties. For the sake of this paper, the sources are defined as the forces acting on the body structure of a car through the engine (for an engine noise problem) or suspension mounts (for a road noise problem).
The traditional way to determine these forces is through the use of the so-called mount stiffness method for soft mounts and the matrix inversion method for stiff mounts. Both methods work well in most cases. Both do have their limitations though. The mount stiffness method requires the availability of the (frequency dependent) mount stiffness which is not always the case and the matrix inversion method depends highly on the condition number of the FRF matrix and requires extensive test efforts.
This paper will therefore review 2 additional TPA methods that add additional load identification methods to the portfolio: operational TPA (or OPAX) which besides being a fast method requiring less instrumentation, also alleviates the need for detailed mount data for soft mounts, and Strain-based TPA which allows the determination of closely coupled forces that would otherwise result in an ill-conditioned matrix inversion problem.
OPAX simplifies the measurement effort need for TPA analysis and identifies the mount stiffness's as part of the path identification process using a parametric model. It is shown in this paper that the method is fast and works well in identifying the dominant path contributions to a noise problem.
Strain-based TPA uses strain sensors, rather than acceleration data, as response locations for matrix inversion. The advantage being that, whereas accelerations capture global deformation patterns very well, the strain sensors will be much more sensitive to local effects.
A number of industrial scale applications will illustrate the methodologies and demonstrate the added value compared to classical TPA.