Assessment of Reservoir Uncertainty for Reservoir Production Using a Deep-Learning Technique
ASSESSMENT OF RESERVOIR UNCERTAINTY FOR RESERVOIR PRODUCTION USING A DEEP-LEARNING TECHNIQUE
KIGAM
Traditional reservoir uncertainty assessments require long simulation times and considerable amounts of static and dynamic data. Recently, machine-learning algorithms have been utilized in the petroleum industry for efficient predictions of future production levels. These methods help to reduce the number of reservoir simulations and manual processing by experts significantly. Currently, we are focusing on the following research items: feature extraction algorithms such as autoencoders and time-series data analysis of, for instance, recurrent neural networks (Fig. 1).
