• Informações do Trabalho

Titulo An Improved Approach for Uncertainty Quantification in Enhanced Oil Recovery Subtítulo Autor GABRIEL BRANDÃO DE MIRANDA Orientador BERNARDO MARTINS ROCHA Resumo Knowledge about the uncertainties associated with a computational model is essential to understand the level of reliability of a prediction, especially in the context of decision making. The model's reliability is a central object in the scope of uncertainty quantification (UQ) techniques, which focus on discussing the uncertainty of predictions and guaranteeing more reliability to the simulations. In the petroleum engineering field it is no different, the need for reliable simulations is mandatory, especially as the drilling and extraction of new wells becomes more expensive. In the context of enhanced oil recovery (EOR) processes, the foam injection technique is used to reduce gas mobility and increase the apparent viscosity, which in turn increases recovery efficiency. Uncertainty from input parameters may affect the simulators' reliability, and therefore uncertainty quantification is a key procedure while studying these physical systems. This work presents an improved workflow for UQ in mathematical models used for the simulation of EOR processes. A study case based on experimental data from core flooding experiment is investigated with the aim of estimating the input parameters and their uncertainty. In particular, the CMG-STARS implicit texture foam model is adopted considering the term that describes the presence of water saturation in the foam. Different strategies for the Bayesian modeling and estimation are studied in order to increase the level of information available propagated to the whole process of UQ. Palavras-Chave Uncertainty quantification, Computational modeling, Bayesian modeling, Enhanced oil recovery.