Fracture detection from water saturation log data using a Fourier–wavelet approach

Fracture detection as applied to reservoir characterization is a key step towards modeling of fractured reservoirs. While different methods have been proposed for detection and characterization of fractures and fractured zones, each is associated with certain shortcomings that prevent from their full use in different related engineering application environments. In this paper a new method is proposed for detection of fractured zones and fracture density in which water saturation log data is utilized. For detection of fractures, we have used wavelet transform and properties of wavelets that are highly suitable for detection of changes and local features of data. To choose the optimum mother wavelet, we have used energy matching strategy in which a wavelet with the highest energy match between spectral energy of the signal at the dominant frequency band and the coefficient energy at the same band of wavelet decomposition of the signal is selected. We have used wavelet packet for a more narrow frequency band selection and enhanced results. Decomposing the water saturation data using wavelets showed that the majority of information of the original log is hidden at low frequency bands. As a result, approximated section of wavelet transform of data was used for fracture detection, while shale volume (or gamma ray) log data was used to filter part of the errors in prediction and identification of the uncertain zones. This increased the accuracy of the results by 70%. Finally, a linear relation was derived between energy of approximated section of water saturation log and fracture density, allowing us to estimate the number of fractures in each fractured zone. The method was applied to four wells belonging to one of the Iranian oilfields located in the southwest region of the country and the results are promising. The use of large volume of data and the subsequent analysis increased the generalization ability of the proposed method.