Abstract:
Fractures are the main contributors in the production of hydrocarbons in tight carbonates. Characterization of these fracture networks is the essential part of the exploration cycle. Based on the analysis of the FMI data present in well-A, few fractures were identified in the sakesar formation with the fracture density of 1 fracture per meter for discontinuous conductive and layer bound conductive fractures with fracture aperture range of 0.0013-0.177 mm, in the entire logged interval. Geoscientists use combination of seismic attributes for the reservoir characterization. Attributes such as chaos, variance, dip deviation and ant track were tested for fractures prediction and the results were calibrated with the fracture trend observed in the well and the faults picked on seismic amplitude data. These attributes with the same information are merged using artificial neural networking (ANN) to generate a final cube with optimum results. Based on the ANN a linear correlation has been established between the attributes to check for the consistency between different attributes. A DFN model is defined as the collection of connected or disconnected fracture patches having certain orientation and density distribution representing the natural fractures. The Discrete Fracture Network (DFN) modelling approach refers to numerical models to describe fractured rock masses, which explicitly represents the geometry of each fractures forming the network. Both Deterministic and stochastic modelling techniques were tested to determine the prevalence of fractures in carbonate reservoirs. Deterministic model is constrained by the attribute data whereas the stochastic model needs well data to constraint the model. Due to the limited well data available in the study area, deterministic model gives more reliable information on fractures then the stochastic model.
