Matrix-related pores play a significant role in controlling hydrocarbon production in organic-rich shales. Multiple matrix-related pore types of typical marine shales in the Sichuan Basin have been visually investigated and identified by transmission electron microscopy (TEM) on ultra-thin sections and by focused ion beam-scanning electron microscopy (FIBSEM) on polished sections. OM-hosted pores seem universal and range in sizes from below 1 nm to hundreds of nanometers and they are not homogeneously developed and distributed, which is mainly determined by thermal maturity and OM composition. Mineral-hosted pores are defined by mineral frameworks and occur in open spaces related to ductile or rigid grain fabric. The four porous mineral types that occur are clay intrapores, carbonate solvopores, pyrite interpores, and quartz interpores, and they range in size from less than 1 nm to more than several microns. Aggregate-hosted pores are predominantly associated with clay-organic aggregates, pyrite-organic aggregates, clay-pyrite aggregates, and clay-organic-pyrite aggregates. The most common aggregate-hosted pore networks are defined by clay-organic aggregates, and the pores are largely developed between the clay and organic layers and may be the important adsorption spaces for methane. Fracture-related pores include microchannels and microfractures of various sizes and shapes and they could play a key role in providing hydrocarbon migration pathways. FIBSEM and TEM show direct evidence that OM-hosted pores and fracture-related pores contribute more to the effective pore network and the excellent reservoir quality, whereas poor reservoir quality may come from aggregate-hosted pores.

Abnormal conditions greatly reduce the efficiency of hydraulic fracturing of unconventional gas reservoirs. Optimizing the fracturing scheme is crucial to minimize the likelihood of abnormal operational conditions, such as pressure channeling, casing deformation, and proppant plugging. This paper proposes a novel machine learning-based method for optimizing abnormal conditions during hydraulic fracturing of unconventional natural gas reservoirs. Firstly, the main controlling factors of abnormal conditions are selected through a hybrid controlling analysis, upon which a surrogate model is established for predicting the occurrence probability of abnormal conditions, rather than whether abnormal conditions happen or not. Subsequently, a machine learning-based optimization algorithm is developed to minimize the occurrence probability of abnormal conditions, acknowledging their inevitability during the fracturing process. The optimal results demonstrate the proposed method outperforms traditional methods, on average. The proposed methodology is more in line with the needs of practical operation in an environment full of uncertainty.