Banca de DEFESA: JÉSSICA ALVES BRASIL

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : JÉSSICA ALVES BRASIL
DATE: 30/05/2022
TIME: 08:30
LOCAL: https://meet.google.com/epj-ephy-gxp
TITLE:

DIAGNOSIS OF OPERATING CONDITIONS OF THE ELECTRICAL SUBMERSIBLE PUMP USING MACHINE LEARNING

KEY WORDS:

Artificial Lift, Electrical Submersible Pump, Machine Learning, Classification Algorithms.

PAGES: 86
BIG AREA: Engenharias
AREA: Engenharia Química
SUBÁREA: Tecnologia Química
SPECIALTY: Petróleo e Petroquímica
SUMMARY:

In artificial lift, Automation techniques are also used in order to increase the efficiency and production of oil wells. In the Electrical Submersible Pump (ESP) lift method, the use of Automation tools becomes essential in the interpretation of data available in the field, since the analysis of these data is not always sufficient to analyze, interpret, monitor and diagnose the performance. and well integrity, in addition to ESP operation and real-time efficiency. However, even though these wells operate with automated systems, some production damages can be identified decreasing the efficiency of the ESP pump and even having significant production losses. Initial diagnosis of the ESP system can lead to great cost savings and less maintenance due to technologies implemented in production fields. In oil fields, to identify the operating conditions of a ESP well, amperimetric charts are used, which are current versus time graphs. The analysis of these charts is usually performed by operators who have a large number of wells to examine, and this overload often decreases the efficiency in the process of reading the operating conditions of the ESP pump. Currently, real-time technologies based on Machine Learning algorithms (ML) have challenged and encouraged companies to create solutions for early diagnosis of abnormalities in well operation. Thus, this work aims to provide a proposal for detecting the operating conditions (normal operation, normal operation with gas, gas interference and gas blockage) of the ESP pump from the analysis of electrical current data obtained from 24 wells from Mossoró, RN, Brazil. Machine Learning classification algorithms were implemented in the Python programming language in the Google Collaboratory environment. The classification algorithms used were Decision Tree (DT), Support Vector Machine (SVM), K-Nearest Neighbor (KNN) and Multi-Layer Perceptron Neural Network (MLP). As the data sets had points ranging from 159 to 344, a standardization was performed with an interpolation technique so that all data sets had 344 points, the maximum number of points collected. The algorithms were tested without and with hyperparameter tuning, in which the hyperparameter set was specific for each ML technique. In addition, balancing tests (oversampling) of the training datasets were performed to identify the difference in relation to the unbalanced dataset. The results obtained and presented throughout the work confirm that the application of the ML algorithm is viable for the classification of the operating conditions presented, since all of them had an accuracy greater than 87%, with the best result being the application of the SVM model, which reached an accuracy of 93%.

BANKING MEMBERS:
Presidente - 1149554 - OSVALDO CHIAVONE FILHO
Externa ao Programa - 6350734 - CARLA WILZA SOUZA DE PAULA MAITELLI
Externo ao Programa - 1752014 - EDNEY RAFAEL VIANA PINHEIRO GALVAO
Externo à Instituição - ANTHONY ANDREY RAMALHO DINIZ - UNIFESP