An improved embedded discrete fracture model with fracture growth for water-induced fracture simulation in low permeability reservoirs

Water injection is often used to supply formation energy in low permeability reservoirs. A lot of field data shows that fractures nearby the wellbores may extend hundreds of meters because of the poor injectability which is called water-induced fractures. The presence of these fractures significantly changes the flow of injected water thus affecting the development effects. However, few studies of reservoir numerical simulation methods have been presented to simulate the dynamic changes of fractures in the long-term waterflooding process. This study is based on the embedded discrete fracture model (EDFM) in which the fracture system has less dependence on the grid system. By improving the preprocessing algorithm, fracture growth can be considered in EDFM which is called the dynamic embedded discrete model (dEDFM). This new method provides an innovative idea for fracture propagation simulation by attaching new fracture elements to the original fractures once the failure criterion is satisfied. Meanwhile, there is no need to calculate the stress field once the in situ stress is provided, so the computational efficiency is greatly improved compared with the fluid-solid coupling method. Besides, dEDFM has a flexible way of handling fracture elements, so it is suitable for water-induced fracture simulation. Results show that fracture propagation speed is significantly influenced by matrix permeability, in situ stress, and injection intensity, and if the water-induced fractures propagate faster than the original waterfront movement velocity, changing law of the water content rate rising of production wells in different directions will be different. The contribution of this work lies in that it provides a suitable and efficient way for reservoir engineers to deal with water-induced fractures since the propagation of fractures can be considered when it comes to reservoir numerical simulation, and it is helpful for production data prediction and development adjustment in low permeability reservoirs.