Research on decoupling inversion analysis of rockfill dam parameters considering dam zoning and material uncertainty: the case study of a concrete face rockfill dam

This paper proposes a parameter decoupling inversion method that integrates sparse polynomial chaos expansion (sPCE) with Bayesian inference. The method aims to accurately investigate the uncertainty in rockfill dam material parameters, improve inversion reliability, and overcome the limitations of conventional deterministic approaches, which ignore parameter randomness and suffer from multi-parameter coupling. This method first constructs the surrogate model of rockfill dam displacement response based on sPCE, employing the orthogonal matching pursuit (OMP) to select key basis functions and enhance modelling efficiency. Combined with Sobol indices, it conducts global sensitivity analysis to reveal the spatial distribution regularity of parameter influences on dam displacement. Building upon this, Bayesian inference is introduced, and the likelihood function accounting for measurement errors is constructed. Subsequently, MCMC sampling is performed using the DREAM algorithm, and the key parameters are inverted sequentially by region using a “decoupling” strategy to achieve dynamic updates of the parameter posterior distributions. Taking the Duncan-Chang EB model as an example, an engineering case study has verified the feasibility and effectiveness of the proposed method. The results demonstrate that decoupling inversion utilising the differences in sensitivity distributions of various parameters is both feasible and scientifically sound. And Bayesian inversion effectively identifies parameter posterior distributions, where the standard deviation gradually converges through iterative inversion, significantly reducing parameter uncertainty. This method systematically quantifies the material parameter uncertainty, enabling probabilistic identification and dynamic updating of parameters. It provides a theoretical basis and technical support for the operational safety assessment and full-lifecycle risk management of rockfill dams.