Abstract：

Compared with the traditional gas injection method, gas-assisted gravity drainage is an emerging development of enhanced oil recovery. It has attracted more and more attention from domestic oilfields due to its unique displacement advantages. At present, there are many prediction models for the recovery of GAGD immiscible reservoirs at home and abroad,but most of the models are simple nonlinear relationships and their prediction accuracy is poor. In recent years, machine learning has been widely used in the petroleum engineering industry and artificial neural networks have become the most potential method for dealing with complex nonlinear regression problems. Based on dimensional analysis, this paper proposes an artificial neural network model that can effectively predict the recovery of GAGD immiscible development reservoirs. In view of the fact that the dip angle of the reservoir is rarely considered in other studies, the Bond number of one of the dimensional parameters is corrected by the reservoir inclination. On this basis, a genetic algorithm and a particle swarm optimization algorithm are used to optimize the model parameters, and an optimization model with the highest prediction accuracy is obtained. By comparing this with the prediction results of other nonlinear regression models proposed in the literature, it is found that the optimized neural network model has higher precision for numerical simulation, physical simulation and actual oilfield recovery.

Key words：gas assisted gravity drainage; dimensional analysis; recovery factor; neural network; genetic algorithm; particle swarm optimization

Received： 2018-12-02

Corresponding Authors： lyq89731007@163.com

Cite this article：CHEN Xiaolong, LI Yiqiang, GUAN Cuo, CHEN Cheng. An optimized neural network prediction model for gas assisted gravity drainage recovery based on dimensional analysis. Petroleum Science Bulletin, 2019, 03: 288-299.

URL：