Hybrid Fuzzy-Based Modeling of Shear Strength Parameters of Rocks Using Petrographic Properties

Mohr–Coulomb failure criterion is vastly employed to understand the risk of collapse of the highly stressed boreholes in oil and gas industries. It is, thus, important to measure the shear strength parameters of Mohr–Coulomb criterion for reservoir rocks in terms of simple input data to easily mitigate the risks of collapse. For this purpose, 59 core plugs from four carbonate oil fields located in Abadan Plain were tested under triaxial stress to measure shear strength parameters (internal friction angle and cohesion). Then, these parameters were estimated based on a novel approach using petrographic properties. Fuzzy clustering (FCM) were initially imployed to cluster the input data followed by developing the adaptive neuro-fuzzy inference system (ANFIS) and interconnecting ANFIS to particle swarm optimization algorithm (PSO-ANFIS) and genetic algorithm (GA-ANFIS). The input parameters of the developed models are petrographic features, including rock texture (Tex.), sedimentary environment (SE), quartz and pyrite content (QP), micrite index (MI), and porosity (Por.), obtained from optical microscopy and XRD analyses. To compare the performance of the models, some statistical metrics were measured. A credibility assessment of the prediction performances reveals that the combination of evolutionary algorithms with ANFIS can be a practical technique to reduce the uncertainties related to shear strength parameters approximation. Therefore, PSO-ANFIS is considered as a promising method for the prediction of shear strength parameters of the studied limestones with R 2 > 0.93 and RMSE < 3.15. Finally, the estimated shear strength parameters are compared and validated with the literature results.