Distinct from conventional reservoirs, shale formations have limited pore connectivity and unique pore spatial-distribution. Consequently, theoretical pore-network models developed for conventional formations are not representative of the porous media in unconventional rocks. This work presents a novel theoretical pore-network model, the dendroidal model, based on the analysis of pore-scale model reconstruction extracted from Scanning Electron Microscope images. The dendroidal model is a “semi-acyclic” model, which characterizes the limited connectivity of void space without sacrificing the interaction among main flow paths. The dendroidal model infers pore-body distribution based on the hysteresis effect of isothermal adsorption/desorption measurements and characterizes pore-throat distribution using mercury drainage capillary pressure experiments. The use of dual-compressibility model in the pore-network model construction eliminates the compressibility effect of void space, including connected pores and dead-end pores, in mercury drainage experiments. The total organic carbon (TOC) content and minerology are measured by experiments to determine the composition of pore bodies and pore throats in the dendroidal model. The difference in mercury intrusion and extraction caused by the trapping hysteresis and contact-angle hysteresis affects the stochastically distributed parameters, including pore-throat length, pore-throat cross-sectional geometry, coordination number and pore-body spatial distribution. I validate the dendroidal model by predicting the absolute permeability of the core samples from Marcellus and Wolfcamp shales. This newly developed pore-network model integrates the aforementioned seven distinct types of experiments to capture the realistic pore structures of shales.

Extracted pore-network modeling is an efficient and reliable way to provide a platform for mathematical simulation of fluid flow in porous media and for predicting the transport properties. However, the existing algorithms for pore-network extraction have deficiencies in characterizing the porous media of shale core samples in as much as they cannot capture the unique features of unconventional reservoirs. In nano-scale pores, the accurate characterization of the porous geometry is important, since the relative error will be significant without considering trivial information. The newly developed approach, based on the maximal-ball method, proposes a novel and enhanced algorithm for the classification of pore throats and pore bodies. It also has a better performance in characterizing the corresponding properties that include pore-throat length, pore size and geometric factors. The Marcellus shale core samples are scanned using scanning electron microscope imaging with the resolution of 4 nm. The pore-network models based on the tomographic images are constructed, and the aforementioned parameters are compared and analyzed.    
The quantification of liquid transport in liquid-rich shales is crucial for an economical exploitation of hydrocarbon. The laboratory measurement of permeability is challenging as it is time-consuming and includes large uncertainties. Direct pore-scale modeling and extracted pore-network modeling are alternatives for the prediction of transport properties. But due to its prohibitively high computational cost, its applications are limited to micro-scale. The emphasis of this work is to understand the mechanisms of nano-confined liquid transportation (nano-scale) and to quantify the liquid transport capacity in the scales of core samples (centi-scale). A modified Navier-Stokes equation is developed to integrate the variation of fluid properties with respect to the strength of liquid-wall interaction. To predict the apparent permeability in large scale, the dendroidal theoretical pore-network model is constructed by integrating mercury drainage/imbibition and isothermal adsorption/desorption experiments. The dendroidal model also integrates the data of Fourier Transform infra-red spectroscopy experiments to characterize the mineralogy distribution and total organic carbon to distinguish organic pores and inorganic pores. Results from molecular dynamics simulation indicate that the flow capacity of nano-confined liquid can be 1-3 orders different from that calculated by Navier-Stokes equation without considering the boundary-slippage effect. The geometry and composition also have considerable effect on the surface friction factor and viscosity in the near-wall fluid film, which in turn significantly influence the flow capacity in nano-pores. This work investigates the mechanisms of liquid flow in nano-confined pores with various composition and geometries. Accurate characterizations of liquid transport in shales will provide significant advantage in the field development planning of unconventional resources.