In tight reservoirs development such as shale gas it is important and yet difficult to predict the size and orientation of the drainage area around a hydraulically fractured well. Often the drainage area is determined by near-well fractures. Diagnostic fracture injection test (DFIT) is an effective way of predicting many reservoir parameters. However, it is challenging to uniquely interpret fractures’ geometry, dimension and spacing. A forward model is desired to correlate the DFIT responses with different fracture configurations and the associated drainage area. We present a 2D model that couples fluid flow with geomechanical deformations in hydraulically fractured reservoirs by solving Biot's equation. Both fluid pressure/velocity and deformations are solved on a finite element mesh. Fracture space is distinguished from the rest of the matrix by high porosity/permeability and low elastic strength. The FEM mesh is adaptively refined at the fractured area to allow the fractures to be reasonably thin and arbitrarily spaced. Pseudo time iteration is applied to seek for convergence between fracture opening/closure and fluid pressure changes. DFIT is simulated with the new numerical model with a single (bi-wing) fracture case and a complex fracture case. The complex fracture case is made by adding transverse fractures to the two wings of a single fracture. The numerical results reveal pressure changes of reservoir fluid due to matrix and fracture deformations as well as due to fluid leak-off. The model is able to generate synthetic well pressure data that show all the type curves given by analytical DFIT theory. The complex fracture case results in pressure transient such that the flow regime rapidly evolves into pseudo-radial flow. For idealized bi-wing fracture cases, this model is consistent with existing analytical tools for DFIT interpretation. The advantages of this model are the ability to implement complex fractures, and the ability to extend to 3D for non-vertical fractures (briefly mentioned in the appendices). Synthetic DFIT data from the model developed in this study has been compared to a field example from a shale gas reservoir. The discrepancy between the model result and field example suggests that some special constitutive law is needed for the modeled fracture areas to appropriately capture the real fracture closure process.