Predicting shale gas production under different geological and fracturing conditions in the fractured shale gas reservoirs is the foundation of optimizing the fracturing parameters, which is crucial to effectively exploit shale gas. We present a multi-layer perceptron (MLP) network and a long short-term memory (LSTM) network to predict shale gas production, both of which can quickly and accurately forecast gas production. The prediction performances of the networks are comprehensively evaluated and compared. The results show that the MLP network can predict shale gas production by geological and fracturing reservoir parameters. The average relative error of the MLP neural network is 2.85%, and the maximum relative error is 12.9%, which can meet the demand of engineering shale gas productivity prediction. The LSTM network can predict shale gas production through historical production under the constraints of geological and fracturing reservoir parameters. The average relative error of the LSTM neural network is 0.68%, and the maximum relative error is 3.08%, which can reliably predict shale gas production. There is a slight deviation between the predicted results of the MLP model and the true values in the first 10 days. This is because the daily production decreases rapidly during the early production stage, and the production data change greatly. The largest relative errors of LSTM in this work on the 10th, 100th, and 1000th day are 0.95%, 0.73%, and 1.85%, respectively, which are far lower than the relative errors of the MLP predictions. The research results can provide a fast and effective mean for shale gas productivity prediction.

Summary: Hydraulic fracturing is one of the important stimulation methods to enhance the productivity of coalbed methane (CBM) wells. However, the commonly used water-based fracturing fluids can bring some bottlenecks such as large amount of water consumption, clay-mineral swelling, and poor fracturing performance on ductile coals. Cyclic liquid nitrogen (LN2) fracturing, as a novel nonaqueous stimulation method, has the potential to solve the above problems. In cyclic LN2 fracturing, supercooling LN2 is injected in a cyclic manner [i.e., alternating high injection rate (or pressure) and low injection rate (or pressure)]. Coals will be subjected to cyclic freeze-thaw, stress oscillation, and fatigue damage, which is expected to improve the stimulated reservoir volume. First, laboratory cyclic LN2 fracturing tests were conducted on coal samples with various coal ranks to investigate the fracture initiation/propagation behavior and fracture network patterns. Cyclic water fracturing tests were also conducted as comparisons. Then, computed tomography (CT) scanning and geomechanical/petrophysical properties tests before and after LN2 fracturing were performed to assist in understanding the cyclic LN2 fracturing mechanisms and implications. Finally, to solve the field application concerns, we investigated the possible fracture geometries at the field scale, temperature distribution of LN2 along the wellbore during injection, and the economic feasibility. The key factors affecting the temperature distribution during LN2 transportation along the wellbore were clarified for the first time. The results indicate that cyclic LN2 fracturing shows the potential to decrease the breakdown pressure and produce complex fracture networks. Different coal ranks have different responses to cyclic LN2 fracturing attributed to the variances in natural fracture development and geomechanical/petrophysical properties. Besides, increasing the cycle number is effective in enhancing the cyclic LN2 fracturing performance on coals with relatively higher geomechanical strengths and tighter rock mass. The suggested cycle numbers from low to high for different coal ranks are listed here: low-rank coal < high-rank coal < middle-rank coal. In field applications, gaseous nitrogen (N2) can be used as the annulus fluid to provide an effective insulation for heat transfer between the low-temperature LN2 and the surrounding environment. In addition, the net present value (NPV) analysis indicates that LN2 fracturing is an economically feasible stimulation method, which can exceed slickwater fracturing in some cases. The key findings are expected to provide preliminary insights into the potential field applications of cyclic LN2 fracturing in CBM or other unconventional oil/gas exploitation.