De Man, P. and Van Schaftingen, J., "Prediction of Vehicle Fuel Tank Slosh Noise from Component-Level Test Data," SAE Technical Paper 2012-01-0215, 2012, doi:10.4271/2012-01-0215.
Fuel system suppliers are developing products that will not degrade the sound quality level of the vehicle in an effort to continuously improve vehicle interior acoustical comfort. Slosh noise originating from fuel displacement within the tank, is a typical annoyance that may emerge above the ambient background noise during low speed conditions or parking. The constraints are even more severe with stop/start and hybrid powertrains where engine masking noise is not continuous or completely absent.The integration of noise constraints into the product development cycle requires having noise specifications at the component level which are aligned with the expected vehicle sound quality. The relation between component and vehicle performances can be established with the well known source-path-receiver approach. The validation of this task is a prerequisite to defining realistic component slosh noise specifications. In addition, if some acoustical performances are obtained through simulation models, an estimation of vehicle noise may be obtained instead of relative performance ranking between different product designs.This paper presents an experimental validation of the source-path-receiver approach applied to slosh noise from a fuel tank on a commercially available passenger vehicle. First, a complete slosh noise characterization of the tank is performed in laboratory on a test bench taking into account both air-borne and structure-borne noise source terms. A set of vehicle body vibroacoustic properties are measured which represents the path-receiver data. Next, the vehicle slosh noise is predicted using the component test data and the vehicle data in the source-path-receiver approach. Finally, the real vehicle slosh noise performance is obtained in the field by both objective measurement and subjective rating for comparing with lab-based predictions. Good correlation has been obtained between predicted and real noise, confirming the validity of the approach.