Abstract
ABSTRACTIdeally, if there is no limitation in the available resources of gas in the gas lift operation, the adequate amount of gas can be injected into the reservoir to have the highest production rate. However, the amount of available gas and the capacity of the compressors are limited and it is required to optimize the gas injection into each well to have the highest recovery. Most of the former studies have been done by complex optimization algorithms and neural networks which require programming languages and a huge amount of time. In this study, we tried to optimize the amount of the injected gas in one of the heavy oil reservoirs in the west of Iran by the use of excel solver software. Then, the results are compared with the famous methods of optimization by genetic algorithm (GA) and particle swarm optimization (PSO) algorithm. Additionally, to improve the accuracy of calculation of the amount of oil production versus the amount of gas injection to the well, the empirical correlations are modified to fit the gas lift performance curves and a new correlation is developed. The results show that the model developed by the authors have the highest accuracy in comparison to other methods. Moreover, the results show that the accuracy of excel solver software is high enough in comparison to GA and PSO algorithm.Abbreviations: ANN - Artificial Neural Network; ARE - Absolute Relative Error; BR - Bayesian Regularization; D - Day; GA - Genetic Algorithm; GLPC - Gas Lift Performance Curve; GOR - Gas Oil Ratio; IPR - Inflow Performance Relationship; LM - Levenberg–Marquardt; MD - Measured Depth; MINLP - Mixed Integer Nonlinear Programming; MMSCF - Million standard Cubic Feet; MSCF - Thousand Standard Cubic Feet; P.I - Productivity Index; ppm - parts per million; PSO - Particle Swarm Optimization; PVT - Pressure-Volume-Temperature; SQP - Sequential Quadratic Programming; SSE - Sum of Squared Errors; SST - Total Sum of Squares; SSSV - Sub Surface Safety Valve; STB - Stock Tank Barrel; TBG - Tubing; TLBO - Teaching-Learning Based Optimization; TVD - True Vertical Depth; WPC - Well Production Curve.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have