Shale gas production prediction based on the TCN-FFT model using multi-well production data

Abstract “Background:” Accurate prediction of shale gas production is crucial to optimize shale gas well production strategies. Current research utilizing single-well historical data necessitates extensive long-term production datasets as training inputs to achieve accurate predictive outcomes. “Method of approach:” In this regard, this paper proposes a Temporal Convolutional Network-Fast Fourier Transform (TCN-FFT) model with an attention mechanism to predict well production only using short-term wellhead pressure. The model combines TCN and FFT modules to extract local and global features, respectively. This allows it to learn flow patterns from the production data of other wells and improves prediction accuracy for a well with shorter production histories. “Results:” Experiments show that for wells with short production histories, the proposed method outperforms single-well-based prediction models in terms of accuracy and trend prediction. “Conclusions:” According to the MAE metrics, the proposed model enhances prediction accuracy by approximately sixfold in comparison to single-well models, which is notably impactful for engineering applications.