Uncertainty quantification of leak noise propagation in buried fluid-filled pipes: Influence of the stochastic soil dynamic stiffnesses

The risk of water shortage has increased in many regions of the world. Since leaks represent a large amount of water wastage, water supply sectors depend heavily on leak localization technologies. Vibro-acoustic methods, such acoustic correlators, offer a non-invasive solution using time delay estimation influenced by wave propagation in buried pipelines. In this paper, a stochastic model for the wavenumber of the predominantly fluid-borne s = 1 wave is introduced. This study applies the nonintrusive Polynomial Chaos Expansion (PCE) to model uncertainties in soil and pipe properties, capturing complex wave behavior more accurately than traditional models, and is performed by transforming the analytical models for the wavenumber of s = 1 wave into computationally efficient surrogate models using the collocation point method. Experimental validation at two very different test sites, one in the UK and the other one in Brazil, confirmed the method's reliability, with measured wavenumbers aligning with predicted confidence intervals. Uncertainty related to the soil dynamic stiffnesses are examined and compared with conventional Monte Carlo method. The results highlight the potential of PCE-based models to improve leak detection and water conservation.