We present a multiscale digital rock workflow to predict shale matrix permeability,that combines firstprinciples molecular dynamics and lattice Boltzmann flow simulations.Inputs are molecular models for the organic shale content,shale 3D microscopy images,fluid composition,pressure and temperature.By complementing the flow in the 3D image resolved pore regions with effective transport in the unresolved pore organic regions,the overall estimation of the shale matrix permeability can be improved.In this workflow a 3D digital rock model is created from Focused Ion Beam-Scanning Electron Microscope(FIB-SEM)images of shale rocks.In this model,solid mineral grains and pores are identified as well as the organic material,which has a nano-scale pore structure that is not fully resolved by FIB-SEM imaging,but contributes a significant fraction of the total gas shale storage,and is also known to allow for gas transport.Previous work(Fager et al.2019)has shown that including an effective organic material gas permeability can have a large impact on the overall shale sample permeability.One of the main components of shale organic material is nano-porous kerogen.Given the lack of experimental data on kerogen transport properties,alternatively molecular models can be used in combination with a multiscale lattice Boltzmann model(LBM)to predict overall shale matrix permeability.In the molecular modeling part of this workflow,a simulation box containing a number of kerogen molecules is used to construct a condensed kerogen structure using molecular dynamics(MD).Density and porosity of the condensed kerogen structure are compared with published data.Gas adsorption isotherms of methane in these kerogen structures are computed at different pressure and temperature conditions using the grand canonical Monte Carlo(GCMC)method.Our simulation results show that at a given temperature,the total methane uptake in the kerogen pore structure increases with pressure,while the excess methane adsorption first increases and then decreases.Based on the configurations obtained from the GCMC simulation,a MD simulation is used to compute the self-diffusion coefficient of methane through kerogen from the resulting trajectories.We observed that the self-diffusion coefficient does not change significantly with pressure.Finally,an effective kerogen permeability to methane is computed from the self-diffusion coefficient and used in a multiscale LBM flow simulation model to predict overall shale matrix permeability.Results are more realistic for the overall shale sample permeability when kerogen permeability is considered.
展开▼