Improving the Estimation of Fracture Length Using Resistivity and Borehole Image Logs

Abstract

Abstract Understanding the geometry and distribution of subsurface fracture networks is essential for reservoir quality characterization, field production forecast, well placement optimization, and hydrocarbon recovery design. Estimating fracture properties with well data has long been a challenge, because wells only intercept a very small volume of rock, and hence have a high chance of under-sampling fracture networks. Among all the fracture properties, fracture length is one of the most difficult aspect to characterize, as the fracture planes are often truncated by wells. In practice, most fracture lengths are observed from outcrops, which may differ from the actual subsurface environment. Due to its close links to vertical and lateral connectivity within the reservoirs, fracture length could be a crucial factor in assessing the potential for hydrocarbon transport. Therefore, a reliable source for fracture length data would greatly reduce the high uncertainties associated with fractured reservoir characterization. In this study, we present a method that categorizes fracture length using laterolog-type resistivity and electrical borehole image logs, based on the different sensitivities of the two types of downhole measurements. In a well drilled with water-based mud, open fractures can get invaded and become electrically conductive, altering the responses of resistivity and imaging tools. While borehole imaging tools are capable of detecting many open fractures regardless of their length, resistivity tools are generally more sensitive to long fractures. Through numerical modeling and inversion, resistivity and borehole image can be processed to locate open fractures. By comparing the identified fractures from resistivity and borehole image in one well, one can separate the long fractures from the short ones. The method was tested on a horizontal well in a carbonate formation, where two fracture sets were originally identified. After a joint interpretation with resistivity and borehole image, one set of the fractures appeared to be consistently shorter than the other. Upon further analysis of the local stress regime, the shorter fractures were identified as drilling-induced fractures, which should not be incorporated in the modeling of natural fractures. The new results have improved the understanding of the regional fracture system, leading to a more accurate fracture network model. The presented method provides new information on the extensiveness of subsurface fractures, which is an essential input for fractured reservoir models. Geologists and engineers can leverage the results and focus on the longer fractures for modeling and simulations, as they are more likely to form effective paths for hydrocarbon flow. The case study has also shown the method could be used as a powerful tool to separate natural fractures from drilling-induced fractures, which may be extremely difficult to differentiate in high-angle or horizontal wells.