INCREASED OIL RECOVERY

Gas lift remains one of the most widely applied artificial lift methods for sustaining oil production, particularly in mature wells with declining reservoir pressure. However, its efficiency strongly depends on the accurate selection of operating parameters such as injection gas rate, valve depth, and tubing size. Conventional approaches often rely on trial-and-error procedures, which are time-consuming and may lead to suboptimal performance. This study presents an integrated approach that combines artificial neural networks (ANN) and sensitivity analysis within a nodal analysis framework to optimize gas lift performance in a Niger Delta field.
A dataset of well test and production parameters was pre-processed, normalized, and used to train ANN models for predicting oil production rates under varying gas injection conditions. The model achieved a high prediction accuracy (R² > 0.95), demonstrating its robustness in capturing the nonlinear relationship between gas injection and liquid production. Sensitivity analysis was performed to evaluate the relative influence of injection gas rate, tubing head pressure, and valve depth on production performance. Results indicated that injection gas rate was the most significant parameter, accounting for over 60% of the observed production variability, followed by tubing head pressure.
The integration of ANN with sensitivity analysis provided clear guidelines for selecting optimal gas injection strategies that maximize production while minimizing energy costs. The study concludes that data-driven approaches can significantly improve the efficiency of gas lift optimization, offering a reliable alternative to conventional empirical and simulation-based techniques. The findings contribute to extending the economic life of mature wells in the Niger Delta and demonstrate the potential of artificial intelligence in enhancing petroleum production systems.