Erawati, Erawati (2025) Prediksi Slope Pada Build Up Pressure Menggunakan Ensemble Method Machine Learning Algorithm Pada Sumur X. Other thesis, Universitas Islam Riau.

Text 203210524.pdf - Submitted Version Restricted to Registered users only Download (1MB) | Request a copy

Abstract

The Pressure Build-Up (PBU) method is a key technique in evaluating reservoir performance, as it provides critical information such as permeability, skin factor, and reservoir pressure. PBU analysis involves identifying the slope on a log�log graph by determining trendline points. Conventional approaches using software like Microsoft Excel or Saphir are time-consuming, prone to errors, and heavily reliant on manual interpretation. This underscores the need for a more efficient and accurate method to address these challenges. Machine learning offers an alternative with its ability to process data automatically and deliver consistent results without significant manual intervention. In this study, the Light Gradient Boosting Machine (LGBM) algorithm was employed to predict the slope value from PBU tests, making the analysis process faster and more efficient. The research involved building a reservoir simulation model using the Computer Modelling Group (CMG)-IMEX software. From the simulation, 1000 experimental datasets were generated, comprising six well test input parameters: Initial pressure, Production rate, Thickness, Factor Volume Formation, Permeability, and Viscosity, with slope as the output parameter. To simplify the analysis, the dataset was clustered into three groups using the K-Means algorithm, with the visualization showing the first cluster in red, the second in blue, and the third in green. The simulation results using the Light Gradient Boosting Algorithm produced a predictive model with Mean Square Error (MSE) and Mean Absolute Error (MAE) values close to zero. With an 80:20 split between training and testing data, the model achieved an R² value of 0.937 for training and 0.784 for testing, with a predicted slope (m) of 148.140.

Actions (login required)

View Item