Evaluation of Uncertainties in Classical and Component (Blocked Force) Transfer Path Analysis (TPA) 2019-01-1544

Transfer path analysis (TPA) has become a widely used diagnostic technique in the automotive and other sectors. In classic TPA, a two-stage measurement is conducted including operational and frequency response function (FRF) phases from which the contribution of various excitations to a target quantity, typically cabin sound pressure, are determined. Blocked force TPA (also called in situ Source Path Contribution Analysis, in-situ TPA and component TPA) is a development of the classic TPA approach and has been attracting considerable recent attention. Blocked force TPA is based on very similar two stage measurements to classic TPA but has two major advantages: there is no need to dismantle the vehicle and the blocked forces obtained are an independent property of the source component and are therefore transferrable to different assemblies. However, despite the now widespread reliance on classic TPA, and the increasing use of blocked force TPA in the automotive sector, it is rare to see any evaluation of the associated uncertainties. This paper therefore aims to summarize recent work and provide a guide to the evaluation of uncertainties in both forms of TPA. The various types of uncertainty are first categorized as, ‘model’, ‘source’ and ‘experimental’ uncertainties. Model uncertainties arise due to incomplete or inconsistent representation of the physical assembly by the measurements. Criteria are provided for evaluation of completeness in terms of measured quantities. Experimental and source uncertainties are evaluated through a first order propagation approach. Expressions are provided allowing the uncertainty in the target quantity to be estimated from measured quantities. Additional data storage and analysis is required but no additional measurements are needed over and above the usual TPA measurements. An illustrative example is provided.