Perpustakaan Digital ITB

Acid stimulation is one of the most commonly employed methods to enhance well productivity in offshore mature oil fields, particularly in the western Java Sea region, where reservoirs have undergone significant depletion and formation damage due to prolonged production. However, the success of these treatments is highly variable and influenced by complex interactions between reservoir characteristics, fluid properties, well conditions, and operational parameters. Traditional rule-based approaches often fail to capture such nonlinear dependencies, limiting their reliability in predicting treatment outcomes. This study presents the development and validation of a machine learning (ML)–based predictive model aimed at forecasting acid stimulation results in terms of job success classification, change in productivity index (?PI), and estimated operational cost. A slope-based classification method was applied using PI measurements before and after stimulation from a dataset of 46 wells. Several algorithms were tested, with Gradient Boosting identified as the most robust model, achieving an R² of 0.8925 for ?PI prediction, 0.7126 for cost prediction, and an F1-score of 0.7692 for success classification. In addition to predictive capability, the model was used to identify optimal parameter ranges for each acid stimulation method—including Foam Acid, Gelled Acid, Organic Acid, Sandstone Acid, Acid Wash, and TCVA—based on SBHP, temperature, porosity, permeability, formation type, and injection volume. These results were benchmarked against established engineering principles and literature references, validating the model's theoretical consistency. To ensure practical applicability, a user-friendly interface was developed to facilitate real-time predictions. Users can input well-specific parameters and obtain stimulation outcome forecasts along with confidence scores. This integrated approach enables more efficient planning, risk mitigation, and cost-effective stimulation design in offshore mature reservoirs, supporting data-driven decision-making for field development optimization..