Patch-Transformer Modeling for Production Forecasting in Tight Gas Reservoirs: A Case Study of the Yan'an Field

Accurate prediction of production rates in fractured horizontal gas wells is crucial for optimizing development strategies in tight gas reservoirs. Traditional prediction methods, such as decline curve analysis, often rely on simplifying assumptions, making it challenging to capture the complex nonlinear dynamics of tight gas production. Consequently, prediction accuracy remains limited even with extensive historical data. This study introduces a novel production forecasting model based on the Patch-Transformer architecture, leveraging five readily available production parameters: the daily production time, wellhead pressure, wellhead temperature, daily water production, and daily gas production. The model captures both local and global temporal dependencies by segmenting the data into patches, thereby enhancing the predictive accuracy. The model was trained and tested using production data from 37 fractured horizontal wells in the Yan'an gas field. The model achieved an average R 2 of 0.9984 on the training set and 0.9932 on the test set, demonstrating exceptional predictive accuracy. The Patch-Transformer model effectively captures the complex temporal dependencies and nonlinear relationships in production data, making it a powerful tool for forecasting gas well performance. By providing accurate predictions, this model has the potential to significantly enhance production optimization, decision-making, and resource management in tight gas reservoir development, ultimately leading to improved productivity.