Hydraulic fracturing has recently been the completion of choice for most tight gas bearing formations. It has proven successful to produce these formations in a commercial manner. However, some considerations have to be taken into account to design an optimum stimulation treatment that leads to the maximum possible productivity. These considerations include, but not limited to, non-Darcy flow and multiphase flow effects inside the fracture. These effects reduce the fracture conductivity significantly. Failing to account for that results in overestimating the deliverability of the well and, consequently, to designing a fracture treatment that is not optimum. In this work a thorough investigation of non-Darcy flow and multi-phase flow effects on the productivity of hydraulically fractured wells is conducted and an optimum fracture design is proposed for a tight gas formation in south Texas using the Unified Fracture Design (UFD) Technique to compensate for the mentioned effects by calculating the effective fracture permeability in an iterative way. Incorporating non-Darcy effects results in an optimum fracture that is shorter and wider than the fracture when only Darcy calculations are considered. That leads to a loss of production of 5, 18 percent due to dry and multiphase non-Darcy flow effects respectively. A comparison between the UFD and 3D simulators is also done to point out the differences in terms of methodology and results. Since UFD incorporated the maximum dimensionless productivity index in the fracture dimensions design, unlike 3D simulators, it can be concluded that using UFD to design the fracture treatment and then use the most important fracture parameters outputs (half length and CfDopt) as inputs in the simulators is a recommended approach.

With the increase in energy consumption, new oil and gas extraction methods in unconventional resources have been explored. Hydraulic fracturing creates fractures to produce and make low permeability reservoirs economically profitable. Hydraulic fractures are also caused unintentionally by the uncontrolled injection in secondary recovery projects or CO2 geological storage. During proppant placement and CO2 injection, the permeability is reduced near the wellbore region due to pore clogging and mineral precipitation. The generated fractures act as high conductivity conduits that increase the capacity of flow in the reservoir. The fracture conductivity is strictly related to its geometry and hydraulic properties. However, these tend to degrade as pressure decreases. The current models do not consider fracture width change in the diffusivity inside the fracture. Additionally, the effect of fracture face skin in fracture closure has not been incorporated. This work focuses on the identification of fracture closure in fractured wells using Pressure Transient data. A semi-analytical model was developed for including the effects of fracture closure, fracture face skin, and complex fracture geometries. The matrix and fracture systems are coupled by pressure continuity at the interface. Fracture face skin is added, assuming a thin layer surrounding the fracture. The model is solved in Laplace space using a semi-analytical approach. The results are validated using a commercial simulator (CMG) and previous models. The pressure response in fractured wells with stress-sensitive fractures is analyzed at early, middle, and late times. In each time period, we identify pressure signals to detect fracture closure by incorporating effective fracture compressibility and fracture conductivity reduction. By incorporating the effective fracture compressibility, the model can reproduce a high storage capacity fracture signal. This signal occurs at early times and can help in post-fracture analysis. The fracture face skin creates an additional pressure drop in the fracture system, triggering conductivity reduction earlier than an undamaged fracture. We proposed a semi-log approach to identify fracture closure for slow rates of fracture closure and the pseudo-radial simplification to generate late time response curves instead of the complete solution for the model.