Muzakky, M. Farhan (2023) Optimasi Produksi Menggunakan Extreme Gradient Boosting Algorithm Pada Intelligent Well, Sumur Z. Other thesis, Universitas Islam Riau.

Text 173210327.pdf - Submitted Version Restricted to Registered users only Download (2MB) | Request a copy

Abstract

Production optimization is one of the activities carried out to increase production from wells to drain fluid from the reservoir to the surface. The resulting imbalance in the production profile can cause water breakthrough or gas into the wellbore. To eliminate this imbalance, an inflow control device (ICD) is placed at each screen connection to balance the production inflow profile across the entire lateral length and compensate for variations in permeability. In performing production optimization there are several methods that can be done, one of which is the intelligent well method. Intelligent wells are intended to create models or simulations for prediction of production and oil recovery factors in various field development scenarios and are assisted by using an Inflow Control Valve (ICV) to control the downhole. In this study, the Computer Modeling Group (CMG)-IMEX reservoir simulation software is used for reservoir modeling. Then built 300 experimental datasets with 5 input well test parameters, namely Initial Pressure, Production Rate, Thickness, Volume Formation Factor, and Viscosity. To simplify the calculation of the intelligent well method, it will be carried out using the Extreme Gradient Boosting Algorithm (XGBoost) method. The XGBoost method is one of the machine learning methods from input data and produces output data. With a ratio of 80% of the production optimization calculation model results from CMG software for training and 20% of the model results for testing in order to get production optimization results on intelligent wells. From the results of the simulation carried out by the Extreme Gradient Boosting Algorithm, it is found that the predictive model with the value of Mean Square Error (MSE) and Mean Absolute Error (MAE) is close to 0 and for R2 training and testing, each value is 0.962 and 0.971. This study describes the application of machine learning in determining reservoir parameters and predictive models of the Extreme Gradient Boosting Algorithm that can be used as a reference and evaluate the performance of production optimization on intelligent wells.

Actions (login required)

View Item