Sequential Modeling for Automatic Interpretation of Pressure Transient Test

Reservoir characterization is a crucial step in developing the oil and gas fields. In this research, a new technique has been demonstrated for analyzing pressure transient tests conducted in oil wells using deep learning. The pressure data recorded at varying time periods from well sites are provided in the prediction model for characterizing the reservoir. The proposed model allows automatic interpretation of homogeneous reservoirs with circular drainage area supported by active aquifers. The constant rate pressure drawdown tests have been simulated using the available analytical model for wells with well-bore storage and formation damage. The noise was added to simulated data to replicate the real-time environment during training of the model. The model showcases significantly high-performance accuracy in automatically characterizing the reservoir.