Bayesian updating for nuclear containment buildings using both mechanical and hydraulic monitoring data. (1st July 2022)

• Record Type: Journal Article
• Authors: Rossat, D.; Baroth, J.; Briffaut, M.; Dufour, F.; Masson, B.; Monteil, A.; Michel-Ponnelle, S.
• Is Part Of: Engineering structures. Volume 262(2022)
• Publication Date: 2022-07-01
• Subjects: Nuclear containment buildings -- Bayesian updating -- Transitional Markov Chain Monte Carlo -- Polynomial chaos expansions -- Thermo-hydro-mechanical modeling -- Air leakage; Structural engineering -- Periodicals; Structural analysis (Engineering) -- Periodicals; Construction, Technique de la -- Périodiques; Génie parasismique -- Périodiques; Pression du vent -- Périodiques; Earthquake engineering; Structural engineering; Wind-pressure; Periodicals; 624.105
• DOI: 10.1016/j.engstruct.2022.114294
• Languages: English
• ISSNs: 0141-0296

Abstract

Abstract: Leak tightness of concrete containment structures such as Nuclear Containment Buildings (NCB) plays a key role in their serviceability and the safety of their surrounding environment. Under the action of operating loads and aging process, leak tightness may evolve over time, which makes numerical forecasting strategies a crucial tool for improving anticipation of overpassing threshold limits, as well as decision aid in the framework of containment structures maintenance. In this paper, a Bayesian updating framework for NCB under serviceability loads is presented. Based on the Bayesian inference process, it allows to incorporate both mechanical and hydraulic NCB monitoring data to numerical predictions in a probabilistic framework, for updating a prior into a posterior state of knowledge related to model parameters. For computational purposes, the approach uses a Transitional Markov Chain Monte Carlo (TMCMC) algorithm for estimating the posterior distribution of model parameters. In order to provide a suitable approach for engineering practice in terms of computational cost, sparse Polynomial Chaos Expansions (PCE) are used for accelerating the TMCMC algorithm. The information gain resulting from Bayesian updating is quantified using a measure taken from information theory, namely the Kullback–Leibler Divergence (KLD). The proposed approach is illustrated through an application to a 1:3 scale NCB. By extracting information from various auscultation data of the structure, the knowledge on uncertain parameters of a complex finite element model is updated throughout the structure operational phase, and new predictions of the structure's strains and leakage responses are subsequently performed. Highlights: A Bayesian updating framework for nuclear containment buildings is presented. Predictions are updated using both mechanical and leakage monitoring data. A Transitional MCMC algorithm is used for Bayesian updating. Sparse Polynomial Chaos Expansions are used to accelerate the TMCMC algorithm. A case study of a 1:3 scale nuclear containment building is presented.