.png){kind=icon}
.png){kind=icon}
Return
Research on Well Pattern Optimization Method for Multilayer Thin Gas Reservoir Based on DNN Algorithm
.png){kind=icon}
OA
AI
DOI：10.3390/pr13030599.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Conventional development technologies for gas reservoirs have made significant advancements, particularly in well pattern optimization, enhanced recovery, and dynamic monitoring. However, multilayer thin gas reservoirs face challenges such as strong heterogeneity, thin effective thickness, poor interlayer connectivity, and difficulties in characterizing dynamic features. These factors complicate the direct application of traditional well pattern optimization methods. The gas reservoir in the western part of Block B, Right Bank of the Amu River, serves as a typical example of a multilayered thin gas reservoir. This study aims to optimize well patterns for composite layer development. A numerical simulation model is established and verified through historical fitting of production data. Using the DNN algorithm and 150 orthogonal experimental models, the optimal horizontal well pattern is determined, with a training error of 7.6% when the training set reached 120 sets. Five optimization charts for multilayer thin gas reservoirs are developed, improving the reservoir recovery factor and supporting efficient production. This work provides a theoretical foundation for the development of similar multilayer composite gas reservoirs.
Keywords:
Deep Neural Network (DNN)
multilayer thin gas reservoir
numerical simulation
well pattern optimization