The exploitation of hydrocarbon reserves in naturally fractured reservoirs composed of different types of rocks has drawn considerable attention from the fracture characterization research community because of the importance of fractures to the prediction of fluid flow. One of the most common methods for rapidly analyzing fracture features is the scanline technique, which provides an estimate of fracture density and frequency. Despite the confidence provided by the systematic use of this method, errors and uncertainties caused by sampling biases exist. The problems caused by these uncertainties can detrimentally affect the construction of a computational model due to misleading trends.
This study evaluated the uncertainty caused by sampling biases in the scanline data of opening-mode fractures in outcrops of naturally fractured Aptian laminated limestone from the Crato Formation, Araripe Basin, northeastern Brazil. The Monte Carlo method was chosen to introduce random values into the sampled values, which enabled us to verify the importance of errors in the accuracy of the method of representing the fracture network.
In this study, errors and uncertainties were grouped into one parameter, termed the coefficient of uncertainty, which was defined as the ratio between the uncertainties, created by the errors and artifacts introduced artificially, and the original scanline data. The propagation of errors and uncertainties in the scanline data to the coefficients of the corresponding power law were determined. This evaluation can be applied in the construction of more reliable geomechanical models using analog geological models for naturally fractured reservoirs.