Gerenciamento de reservatório de petróleo baseado em controle preditivo não linear por meio de filtro de partículas
Nenhuma Miniatura disponível
Data
2018-02-23
Autores
Fortunato, Társis Baia
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Federal do Espírito Santo
Resumo
Energy is one of humanity's greatest needs and this need continues highly depends on the production of oil and gas. In this sense, the production systems of oil needs advances continuously. Currently, two techniques that make up the main trends of the oil industry and contribute to the advancement of production systems are model predictive control (MPC) and state estimation techniques. The production systems have characteristics of non-linearities that too are seen in the mathematical models that reproduce their behaviors. However, MPC is a mature technique only for linear models, and its application to nonlinear processes is conditioned to simplifying hypotheses. Its Nonlinear Model Predictive Control (NMPC) variant that uses nonlinear models has been indicated for use in the control of production systems, since it does not assume simplifying hypotheses. The challenges of NMPC lie in solving the problem of model-based optimization that integrates its methodology and also in the treatment of uncertainties. Thus, it has been common to associate NMPC with state estimation. However, even though there are several estimation techniques available, there are few that deal well with non-linear models. Thus, this dissertation proposes a methodology of control of the production system considering the step of secondary waterflooding recovery with a NMPC associated with state estimation. To the challenge in the optimization stage is applied a methodology that reformulates the optimization problem as a filtering problem and the optimum is estimated with the Particle Filter (PF), which in this task is renamed to Particle Filter Optimization (PFO). In the process of state estimation, a Particle Filter is applied which makes no simplifying hypothesis in relation to non-Gaussian uncertainties. The simulations required during the application of the two techniques will be obtained with the model describing the two-phase oil-water immiscible flow and the Finite Volumes method in its Two Point Flux Approximation variant. The results showed that the PFO maintained the production at the set point and that the estimation of the PF states was satisfactory, since the monitoring results did not show degeneration or impoverishment in the PF sampling and the results regarding the uncertainty treatment showed that the PF was able to reduce the uncertainty in the estimated saturation.
Descrição
Palavras-chave
NMPC , State estimation , Filtro de partículas , TPFA , Reservatório de petróleo , PFO , Controle preditivo não linear , Particle filter , Estimação de estados