Modeling aspects of reliability-based design of lined rock caverns
Time: Fri 2022-12-02 13.00
Location: D2, Lindstedtsvägen 5, Stockholm
Video link: https://kth-se.zoom.us/j/64410297205
Subject area: Civil and Architectural Engineering, Soil and Rock Mechanics
Doctoral student: Davi Rodrigues Damasceno , Jord- och bergmekanik
Opponent: Professor Rafael Jimenez, Technical University of Madrid
Supervisor: Docent Fredrik Johansson, Jord- och bergmekanik; Docent Johan Spross, Jord- och bergmekanik
The storage of large quantities of hydrogen gas in underground lined rock caverns (LRCs) could contribute to an efficient supply of fossil-free energy. The consequences of failure of such storage can be catastrophic, so representative predictive models and a small probability of failure are needed for the LRC design. However, the available predictive models are simplified. On top of that, the calculation of a small probability of failure is challenging on its own, and becomes more difficult when combined with representative numerical models, which are often computationally demanding.
The purpose of this thesis is to develop a reliability-based design tool for LRC gas storages to ensure that societal safety requirements are met. For the development of this LRC design tool, the research issues are related to the prediction of the rock cavern response to a high internal gas pressure; interaction between LRC components; suitability of reliability-based calculation methods for the LRC design; and, effect of uncertainties on the probability of failure of the LRC design.
The results show that the available analytical model to predict the rock cavern response is only applicable for idealized geological conditions and geometries, so numerical models are needed. Finite element (FE) models are therefore developed to account for the complex interaction between LRC components, including the influence of opening of discrete rock joints on the strain concentrations in the steel lining. The adaptive directional importance sampling (ADIS) method is identified to be suitable to perform reliability-based analysis with FE models, requiring only a small number of samples for sufficiently accurate estimations of small probabilities of failure. The structural reliability of the LRC design is found to be sensitive to the rock mass quality and the correlation between geological properties.