Multi-fidelity Techniques and Uncertainty Quantification for Mechanical and Transport Processes in Fractured Rock

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68145535%3A_____%2F24%3A00603686" target="_blank" >RIV/68145535:_____/24:00603686 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.ejp-eurad.eu/publications/eurad-d47-report-describing-numerical-improvement-and-developments-and-their" target="_blank" >https://www.ejp-eurad.eu/publications/eurad-d47-report-describing-numerical-improvement-and-developments-and-their</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Multi-fidelity Techniques and Uncertainty Quantification for Mechanical and Transport Processes in Fractured Rock

Popis výsledku v původním jazyce

The report summarizes the results achieved by teams from the Technical University of Liberec and the Institute of Geonics of the Czech Academy of Sciences, as part of Task 4 within the DONUTnworkpackage. We introduce a parallel Bayesian inversion library based on the delayed acceptance Metropolis-Hastings method for accelerated sampling using suitable surrogate models. Then we applynthe library to an inverse problem, specifically the identification of the parameters to the hydro-mechanical model describing formation of the excavation disturbed zone. Additionally, we introduce a multilevel Monte Carlo method (MLMC) library, which has been effectively applied to test transport problems. We present a neural network developed for fast numerical homogenization of small-scale discrete fractures in the 2D case. This represents a crucial advancement for applying MLMC to models with discrete fracture networks. Finally, we integrate developed methods into a benchmark safety study for near-field transport.

Název v anglickém jazyce

Multi-fidelity Techniques and Uncertainty Quantification for Mechanical and Transport Processes in Fractured Rock

Popis výsledku anglicky

The report summarizes the results achieved by teams from the Technical University of Liberec and the Institute of Geonics of the Czech Academy of Sciences, as part of Task 4 within the DONUTnworkpackage. We introduce a parallel Bayesian inversion library based on the delayed acceptance Metropolis-Hastings method for accelerated sampling using suitable surrogate models. Then we applynthe library to an inverse problem, specifically the identification of the parameters to the hydro-mechanical model describing formation of the excavation disturbed zone. Additionally, we introduce a multilevel Monte Carlo method (MLMC) library, which has been effectively applied to test transport problems. We present a neural network developed for fast numerical homogenization of small-scale discrete fractures in the 2D case. This represents a crucial advancement for applying MLMC to models with discrete fracture networks. Finally, we integrate developed methods into a benchmark safety study for near-field transport.

Druh

V_souhrn - Souhrnná výzkumná zpráva

CEP obor

—

OECD FORD obor

20701 - Environmental and geological engineering, geotechnics

Projekt

—

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Počet stran výsledku

19

Místo vydání

EU

Název nakladatele resp. objednatele

EU

Verze

—