Banca de DEFESA: EZEQUIAS SANTOS DE MATOS

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : EZEQUIAS SANTOS DE MATOS
DATE: 09/11/2023
TIME: 14:00
LOCAL: https://conferenciaweb.rnp.br/webconf/pse-ufba
TITLE:

Data-driven economic real-time optimization of gas-lifted oil wells

KEY WORDS:

Real Time Optimization; Gas Lift; Artificial Intelligence; Nonlinear Estimation.

PAGES: 77
BIG AREA: Engenharias
AREA: Engenharia de Produção
SUMMARY:

In the present study, one addresses a daily dynamic optimization problem in oil production through a data-driven approach. To this end, the use of an artificial neural network (ANN) architecture is proposed as a substitute for the phenomenological model that represents the gas-assisted oil well production system. Therefore, it is crucial to include information on the oil flows from each well and also from the top of the riser. However, flow rate measurements are not available in real-time, and what is available is only the total flow rate after separation, which does not allow understanding the behavior of each well individually. Due to the unavailability of individual well flow measurement, the use of a mobile horizon estimator (MHE), supported by a phenomenological model, has proven to be an appropriate solution for estimating these variables, allowing for the provision of data for training and obtaining a substitute model to be used in the dynamic optimization stage. Thus, it was possible to enable the training of the Nonlinear Autoregressive with Exogenous Input (NARX) neural network architecture employed in this research. This choice was based on the finding that the network was able to conveniently predict one-step-ahead. The results obtained from the application of the proposed approach on a single well and on a field consisting of three wells and a riser demonstrated a good performance of the artificial neural network in terms of temporal prediction, in addition to a more efficient computational time in solving the optimization problem when compared to the standard phenomenological model. The solution proposed by this approach opens up several possibilities for implementation in large-scale problems, such as optimizing the daily production of an oilfield composed of multiple wells integrated by different reservoirs and manifolds.

COMMITTEE MEMBERS:
Presidente - 2042153 - MARCIO ANDRE FERNANDES MARTINS
Externo à Instituição - GUILHERME AUGUSTO DE ALMEIDA GONÇALVEZ - PETROBRAS
Externo à Instituição - OSCAR ALBERTO ZANABRIA SOTOMAYOR - UFS