Haswinda Putri, Rizka (2022) Optimasi Volume Injeksi Pada Waterflooding Menggunakan Metode Artificial Neural Network (ann). Other thesis, Fakultas Teknik Perminyakan.

Text 183210093.pdf - Submitted Version Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Waterflooding is one of the second stage methods in which water is injected with the aim of maintaining reservoir pressure (pressure maintenance). The water injected must be adjusted so that formation damage does not occur, and a decrease in oil recovery occurs. The purpose of this research is to optimize the value of injection volume and recovery factor (RF) in waterflooding using the ANN-BP method. The parameters used in this study are porosity, water saturation, oil saturation, rock compressibility, vertical permeability, and horizontal permeability. Reservoir modeling was built using CMG-IMEX software. Then CMG-CMOST is used to build data iterations that will be brought to machine learning (ML). Then to optimize the value of the recovery factor and injection volume, an approach is made using an artificial neural network with a back-propagation algorithm. By using a ratio of 70% for training and 30% for testing. In order to get optimal prediction results from RF and injection volume, trial and error is carried out on the number of hidden layer nodes. The hidden layer nodes at nodes 10 obtain a coefficient of determination R2 training 0.989824 and R2 testing 0.988112 have high accuracy results because they are close to 1. Then they are supported by the mean absolute percentage error (MAPE). The MAPE value in training data is 2.4745 and data testing is 2.3250 which means means Highly accurate prediction. The optimal result for the recovery factor value is 26.17 % increasing 5.85% from the basecase and the injection volume is 15387684 bbl or 15.4 MMbbl.

Actions (login required)

View Item