Prediksi Water Coning Pada Natural Fractured Carbonate Reservoir Menggunakan Metode Artificial Neural Network

Putri, Dessy Nandisa (2021) Prediksi Water Coning Pada Natural Fractured Carbonate Reservoir Menggunakan Metode Artificial Neural Network. Other thesis, Universitas Islam Riau.

Text 153210115.pdf - Submitted Version Download (1MB)

Abstract

High water production is one of the main problems in the oil and gas industry. Proven in 2017, Al - Azmi reported that daily water production worldwide reached 300 million barrels compared to oil production of only 80 million barrels. The phenomenon Water Coning is one of the causes of high water production, especially in the Natural Fractured Carbonate Reservoir (NFCR). The presence of the permeability fracture creates a pathway permeable major so that the water breakthrough time occurs earlier. Although NFCR contains almost 60% oil reserves and 40% gas reserves, due to the high amount of water production the production wells die early without optimum oil recovery. Therefore, in this study prediction of water coning through water breakthrough time on NFCR is carried out so that it is hoped that an optimal production scheme will be created so the production wells in NFCR can last longer. This research begins by constructing 225 experiments from 6 parameters, namely horizontal matrix and fracture permeability, vertical matrix and fracture permeability and matrix and fracture porosity as the most dominant parameters of the phenomenon water coning in NFCR. Making 225 experiments using the Design of Experiment (DoE) in the Computer Modeling Group (CMG - CMOST) software with Water Cut as a response parameter. Then the prediction water coning made by Artificial Neural Network based validation of the accuracy of the Coefficient of Determination (R2) using the Programming Language Python. Based on the results of the research that has been done, the Artificial Neural Network produces accurate water coning predictions in a relatively shorter time. Through the Coefficient of Determination for 12 hidden nodes in the hidden layer, an accuracy model of 0.998245 is obtained for data training and 0.991375 for data testing. Meanwhile, the water breakthrough time occurred on the 75th day after the well was first produced in 2020 with a value Water Cut of 42%. The results of this study have the potential to bring renewable information in the oil and gas industry, especially the implementation of the Artificial Neural Network which should be considered for use in making predictions with an accuracy of 0.913523.

Actions (login required)

View Item