Development of computationally efficient 2D and pseudo-3D multi-fracture models with applications to fracturing and refracturing

Multi-stage hydraulic fracturing is one of the key technologies of the U.S. tight oil and shale gas revolution. Recently, fracture diagnostic methods revealed that the fracture propagation could be quite uneven when stimulating multiple fractures simultaneously. As a result, only 64% of the perforated clusters contribute to production. Promoting uniform fracture propagation, ensuring that all perforation clusters receive treatment would be a big step in improving oil recovery in shale reservoirs. The dissertation reports the development of computationally efficient, 2D and Pseudo-3D multi-fracture models. Novel methods are developed to solve the dynamic fluid and proppant partitioning among multiple perforation clusters. The Resistance Method is developed to distribute fluid among fractures. This new method could be more computationally efficient than the widely adopted Newton-Raphson Method. The Particle Transport Efficiency (PTE) correlations are implicitly incorporated into the multi-fracture models to compute proppant distribution among the fractures. It is shown that the inertial effect tends to accumulate proppant particles downstream in the wellbore while fluid leaks off from the perforations, leading to pre-mature screen out of toe-side clusters, and the heel-biased final treatment distribution. The model has been applied to two important unconventional reservoir stimulation technologies: the plug-and-perf operation and horizontal well refracturing. We investigate how parameters including the number of perforations, the size of the perforation, the injection rate and so on affect the final fluid and proppant distribution. Directional suggestions are provided regarding each parameter. An automated process to search for the optimum plug-and-perf design within the user-specified parameter range was developed. It is shown that when multiple parameters are optimized together, the propped surface area can be improved greatly. We simulated horizontal well refracturing operations employing diverting agents with the model. Two field cases were studied, and the simulation workflow of initial completion – pore pressure depletion – refracturing was carried out for both cases. Our simulation results match the field diagnostic observations well. We successfully captured the heel-biased refrac treatment distribution, and showed that both new and existing perforations can effectively break down during refrac. Strategies have been developed to improve refrac success.