Sembar Formation as an Unconventional Prospect: New Insights in Evaluating Shale Gas Potential Combined with Deep Learning

Since most of the petroliferous basins of Pakistan have not been fully explored, the country's energy output falls short of its energy demands. Studies conducted both internationally and domestically show that it has far higher potential for hydrocarbons than its current proven reserves. It has been estimated that there are around 105 trillion cubic feet of natural gas resources that are potentially recoverable from the shale formations of Indus Basin. The Cretaceous shales of the Sembar Formation are one of the major prospects for shale gas exploitation in Central and Southern Indus Basins. An integrated study was conducted to evaluate its conventional and unconventional resource potential in terms of reservoir characteristics, organic richness and clay type. Conventional reservoir potential within the top sand facies of the Sembar Formation exhibited consistent behavior with regards to its porosity and fluid saturation. The lower part of this formation is characterized primarily by the prevalence of shale facies. Empirical log-based techniques were utilized to assess the organic richness of these facies, which showed total organic carbon (TOC) content ranging 1-1.4%. Along with its organic richness, the category of clay minerals is a very important factor in estimating unconventional resource potential. The log-based mineralogical assessment revealed the presence of brittle clay minerals (illite), which are suitable for fracking. The diagenetic changes of clay minerals may increase brittle components, which could lead to an increase in the integral rigidity of these rocks. Moreover, deep feedforward neural network (DFNN) analysis, which is a non-linear neural network regression technique, was used to delineate the spatial and vertical variations of TOC. The impedance sections, combined with the original seismic trace and TOC curve from the well, were utilized as input for training the seismic attributes for predicting seismic-based TOC. The findings of the DFNN analysis on the Duljan Re-Entry-01 well indicate a strong correlation coefficient of 93% between Passey TOC and the predicted TOC curve. Additionally, the root mean square error was found to be a mere 0.09. The TOC prediction based on seismic-based DFNN analysis yielded an average value of approximately 1.1%, with a maximum value of 2.3%. These two-dimensional TOC sections for the Sembar shales provide an extremely positive impact in evaluating its shale gas potential.