Diagnosis of Multiple Fracture Stimulation in Horizontal Wells by Downhole Temperature Measurements

Diagnosis fractured well performance for a multiple-stage fractured horizontal well is critical to understand and improve fracture stimulation design. Production logging tools (PLT) can be used in this problem, and temperature distribution data (by PLT or fiber-optic sensors) is one of the valuable information for performance diagnosis. However, until today quantitative interpretation of dynamic temperature data is still challenging and requires indepth mathematical modeling of heat and mass transfer during production in a complex flow system. In this study we developed a semi-analytical model to predict temperature and pressure behavior in a multiplefractured horizontal well during production. The tri-linear model is used to simulate flow in a fracture system for horizontal wells. The model can be applied for single phase oil or gas wells. For gas well, the non-Darcy effects are considered by permeability alteration using minimum permeability plateau. Flow in the wellbore is modeled under unsteady state condition modified from a steady state horizontal wellbore model presented before. This coupled model for fracture flow and wellbore flow can predict the pressure gradient along fracture and also the pressure and flow rate distribution in the wellbore. The thermal model calculates the heat transfer in the fracture/reservoir/wellbore system considering subtle temperature changes caused by the Joule-Thomson cooling and frictional heating effects. The fluid properties are set as functions of in-situ pressure and temperature. The result shows that transient temperature behavior can be used to estimate the fracture initiation points and length, number of created fractures and distribution of fluid along the wellbore. The temperature is sensitive to the flow rate distribution along the wellbore, the fracture geometry, and also the fracture conductivity. With the developed method, we can evaluate the fracture treatment by comparing the designed fracture half-length with the generated fracture-length. When applied in history match of production data, it also can predict conductivity decline as a function of production time.