Research

Left: Pore network model of a Bandera Gray sandstone sample.  Right: Relative permeability (Kr) curves for hydrogen and water in a depleted oil and gas reservoir simulated by the pore network model.  “PD” denotes “primary drainage” and “IM” denotes “imbibition”. It is clear that during the IM process water’s mobility is enhanced whereas hydrogen’s mobility is reduced. Picture credit: Qingqi Zhao.

a) Reservoir model for hydrogen injection and production in a water-filled depleted gas reservoir, and dissolved hydrogen concentration in pore water with b) CO2, and c) CH4 as the cushion gas.  We developed this two-phase (gaseous and liquid), three-component (hydrogen, cushion gas, and water) reservoir simulator to investigate the effect of different cushion gases in underground hydrogen storage. Picture credit: Qingqi Zhao.

Lab-measured fracture hydraulic conductivity as a function of increasing proppant concentration (i.e., proppant amount in fracture). Non-monotronic development of fracture hydraulic conductivity can be observed. Picture from Li et al. 2022.  

We use a customized fracture conductivity cell, fabricated according to the API RP-19D standard, to measure fracture's hydraulic conductivity and study how it is influenced under various confining stresses, proppant areal concentrations, proppant particle sizes, proppant materials, rock types, rock hardness, rock surface roughness, and testing fluids. Images from Fan et al., 2020.

We use UNet++ and entropy-based-masking indicator kriging (IK-EBM) for boundary and small target segmentation in digital rock images. Picture from Wang et al. 2022.  

Unsupervised deep learning model associated with the over-segmentation strategy and agglomerative hierarchical clustering (AHC) post-processing. Picture from Wang et al. 2023.

Top: Particle crushing simulated by PFC3D: a big proppant particle is replaced by several small particles after crushing. 

Bottom: Generation of fine-sized particles in a monolayer structure due to proppant crushing under increasing stress and temperature, leading to reduction of fracture width and proppant pack porosity and permeability. Images from Fan et al., 2020.

Left: we used micro-CT imaging to show that the surface contact angles in a sandstone reservoir follow a log-normal distribution, and there is a spatial correlation length associated with contact angle’s spatial heterogeneity. 

Right: influence of the variation of surface contact angle on pore-scale distribution of trapped CO2 (black). “SD” and “L” indicate standard deviation and spatial correlation length of contact angle distribution. Spatial distribution of CO2 (black) and water (white) was simulated using LB multiphase flow modeling in 3D space. Images from Guo et al., 2020.    

Various reservoir rocks scanned by nano-CT (Chen, 2016).

Left: Glass beads (blue) and deposited colloid particles (red) (Gaillard et al., 2007).

Right: Kerogen (red), intra-kerogen pores (blue), and pyrite (white) in a shale sample (Chen et al., 2013).

LB-simulated single-phase flow field in a sandstone. The LB simulation is done in 3D space and these 2D flow field pictures are cross sections cut from the 3D domain. 

Comparison between micro-CT-scanned CO2/water distribution and LB-simulated CO2/water distribution in a Berea sandstone (left), and using LB multiphase flow simulation to study the role of contact angle, Capillary number, and viscosity ratio on the relative permeability (Kr) curves (right) (Fan et al., 2019). The LB multiphase flow simulation is done in 3D space and the 2D CO2/water distributions are cross sections cut from the 3D domain. 

Nanoindentation experiments are combined with DEM/LB-integrated modeling to account for proppant compaction and embedment in a hydraulic fracture and to study their role on the evolution of fracture conductivity (Fan et al., 2019).

Mitigation of proppant embedment in a vertical fracture with increasing proppant amount (left), and fracture conductivity as a function of proppant amount (right). The hybrid, experiment/simulation-integrated workflow is used to account for the effects of proppant compaction and embedment on fracture conductivity (Fan et al., 2019).

Single-/multiphase flows in proppant assemblies sandwiched in a rock fracture and subject to continuous compaction (Fan et al., 2018), and application to larger-scale optimization of hydraulic fracturing (Gu et al., 2017).

We developed a novel multi-physics shale transport (MPST) model to account for the coupled multi-physics processes of geomechanics, fluid dynamics, and Klinkenberg effect for gas transport in shales (left), and the MPST model successfully predicts the change of measured permeability as a function of pore pressure in shale cores subjected to various effective stress (right) (Li et al., 2020).