Uncertainty quantification in porous media with multi-level Monte Carlo
Document typeConference report
Rights accessOpen Access
Geological storage of CO2 is an attempt at controlling future climate changes. Modelling and simulation of underground CO2 storage are however subjected to significant sources of geological uncertainties, which requires the use of stochastic approaches. Sources of uncertainties in the CO2 storage problem can be classified as either geological, physical, or operational uncertainties. Ranking the importance of the model parameters based on their influence can provide a better understanding of the system. Computationally efficient methods for sensitivity analysis, uncertainty quantification, and probability risk assessment are therefore needed. Furthermore, due to the computational complexity of such problems, as even a single deterministic simulation may require parallel high-performance computing, stochastic simulation techniques based on standard Monte Carlo are currently inefficient for these problems. To overcome the prohibitive computational cost of standard Monte Carlo, we propose a Multi-Level Monte Carlo technique to estimate statistical quantities of interest within some prescribed accuracy constraint. We illustrate and verify our proposed approach by a comparison with a Monte Carlo simulation using a common benchmark problem for CO2 injection. We obtain a significant computational speed-up compared with Monte Carlo.
CitationSaad, B. [et al.]. Uncertainty quantification in porous media with multi-level Monte Carlo. A: ADMOS 2015. CIMNE, 2015, p. 80.
All rights reserved. This work is protected by the corresponding intellectual and industrial property rights. Without prejudice to any existing legal exemptions, reproduction, distribution, public communication or transformation of this work are prohibited without permission of the copyright holder