Optimizing Plunger Lift Systems for Gas Well Deliquification: A Bayesian Approach with Comparative Algorithmic Analysis

Summary Plunger lift is an efficient and environmentally friendly deliquification gas production technique widely used in low-productivity, liquid-loaded gas wells. To enhance its operational efficiency, we propose a multiobjective parameter optimization method based on Bayesian optimization (BO). For this study, we constructed a weighted objective function to maximize cumulative gas production while minimizing the bottomhole liquid level. Comparative experiments were conducted using four optimization algorithms: particle swarm optimization (PSO), genetic algorithm (GA), simultaneous perturbation stochastic approximation (SPSA), and BO. A mathematical model of plunger movement was developed based on field data, and global optimization of the plunger cycle duration and open-well duration was performed over a 24-hour simulation period. Experimental results demonstrate that the BO algorithm achieves superior overall performance by maintaining high production improvement while effectively controlling liquid levels. It also exhibits faster convergence and stronger constraint-handling capability compared with other methods. After 50 iterations, the dimensionless weighted performance index of the BO reached 1.0441, significantly outperforming PSO (1.0294), GA (0.5049), and SPSA (0.8056). This study highlights the strong adaptability and engineering application potential of BO in complex nonlinear systems, offering a practical and computationally efficient approach for static operational parameter optimization of plunger lift wells under given operating conditions.