Microscale pore-fracture system characterization of shales by digital images, gas sorption and machine learning




Tian, Xiao, Ph. D.

Journal Title

Journal ISSN

Volume Title



The challenges in sustainable and efficient development of organic shales demand a better understanding of the micro-scale pore structure of organic shales. However, the characterization of shale pore structure is challenging because the pores can be very small in size (less than several nanometers). Besides, the pore network in shales can be very different from that of conventional reservoirs. A unique feature of the pore system in organic shales is that the formation of the pore network usually depends heavily on the maturation of organic matter. In this dissertation, the properties of the pore network in the shale matrix are investigated using pore-scale network modeling constrained by low pressure nitrogen sorption isotherms. In addition to the larger, induced fracture network generated during hydraulic fracturing, the flow pathways formed by microfractures and nanopores are important for economic hydrocarbon production formations. However, due to the difficulty in detecting and characterizing microfractures and the complexity of hydrocarbon transport in shales, the role of microfractures in hydrocarbon flow during production is still poorly understood. In this dissertation, the most up-to-date machine learning and deep learning tools are utilized to characterize microfractures propagation in organic rich shales. The microfracture lengths and fracture-mineral preferential association are studied. Later a microfracture is embedded in pore-scale network models to understand the role of microfracture in permeability enhancement of shale matrix. Shale samples from three different shale formations are studied in this dissertation. These formations are Barnett shale, Eagle Ford shale (Karnes County, TX), and a siliceous shale in the northern Rocky Mountains, USA (the specific location has been withheld at the donor's request). By analyzing the nitrogen sorption isotherms for isolated organic matter clusters and the bulk shale samples, the pore size information in the organic matter and inorganic matter is explored for the Barnett shale samples. The network connectivity and pore spatial arrangement in the shale matrix is determined by comparing the modeled nitrogen sorption isotherms with laboratory experiment results. Later, the pore network model is used to predict the permeability of the shale matrix. Both Darcy permeability and apparent permeability (including Darcy flow and gas slip effect) are computed using the network model. The apparent permeability is much larger than the Darcy permeability from laboratory measurements and pore network modeling. Pore network models are constructed for four samples: two sample from Barnett shale, one from Eagle Ford and one from the siliceous samples. The pore network properties are characterized using nitrogen sorption isotherms. I concluded that on average, each pore is connected with two neighboring pores in shale matrix. The pore spatial arrangement is not totally random in the network. By analyzing the microfracture propagation in high resolution scanning electron microscopy (SEM) images for the Eagle Ford samples and siliceous samples, the fracture lengths and density in intact and deformed (in confined compressive strength testing) shale samples are explored. The preferential fracture-mineral association is scrutinized by analyzing the composed back-scatter electron (BSE) images and the energy dispersive x-ray spectroscopy (EDS) images for the Eagle Ford and Siliceous samples. The fractures and minerals are easily detectable after combining BSE and EDS images. The conclusion is that, the generation and closure of microfracture is closely related to total organic carbon (TOC), OM maturation, and minerology. A higher TOC indicate that more microfractures are generated during organic matter maturation. Clay dehydration is another reason for microfracture generation. However, the microfractures generated due to clay dehydration or organic matter maturation are also more sensitive the pressure dependent effect. Microfractures also develop at the grain boundaries of more brittle minerals such as quartz, calcite, and feldspar. The permeability of the pore network model containing one microfracture is computed for the Eagle Ford sample and the siliceous sample. The results suggest that the permeability of shale matrix increases greatly after one microfracture is added to the matrix. The relationship between permeability and fracture height-aperture ratio follows a logarithmic function for both samples. The relationship between permeability and lengths follows an exponential function for both samples. Fracture lengths have more impact on fracture permeability compared to height-aperture ratio


LCSH Subject Headings