A Novel Machine-Assisted Technique for Extracting Multiscale Vugs and Fractures in Heterogeneous Carbonates Sequence

Abstract

Predicting petrophysical properties in carbonate reservoirs is challenging due to the deposition and diagenetic history, which creates pore-scale features and heterogeneity at multiple-length scale. Non-fractured carbonate rocks with monomodal pore distribution often provide weak transportation properties compared to carbonates with multimodal pore system. The behaviour of such formations is subject to percolation effect where the connectivity of vug clusters control the poro-perm relationship which can be explained with high-resolution microresistivity images and nuclear magnetic resonance (NMR) data. A machine-assisted processing technique, defined as "thresholding," was applied to high-resolution microresistivity images, resolving vugs and fractures with similar resistivity. Other objects of interest are removed using object-oriented filters and thresholding, resulting in a "sculptured image" containing only vugs and fractures. The image is analysed to quantify formation porosity. A Laplacian of Gaussian filter is used to avoid highlighting features of no interest. Step two analyses T1 and T2 relaxations allowing portions of signal from a pore-size group to spill across the discrete boundaries. The pore-size takes on a fuzziness near the discrete relaxation time cut-offs corresponding to pore radii breakover points. High poro-perm layers of grainstone in overall thinly bedded sequences of packstone and wackestone were successfully identified and subsequently shed light upon the ambiguities observed in mobility values obtained from formation tester across the same lithocolumn. This novel technology helps in deciphering high-resolution integrated lithofacies. The histogram from the image porosity binning demonstrates a different response within vugular zones compared to fractured zones. Where the vugs sizes are variable, they exhibit a multi-pore system nature in NMR. For the fractured interval, the images and NMR exhibit weak distribution. The resistivity independent image pixel-based filtration technique helps to define interesting features on images which can be enhanced and measurable at various scales. Machine assisted technique in NMR complement the results in aiding to characterize the heterogeneous carbonate rocks.