Simulation of Unconventional Reservoirs with Complex Fracture Conditions

Abstract

The unconventional reservoirs have been a major source of the modern oil and gas industry. However, simulation of unconventional reservoirs with complex fracture conditions remains to be a difficult problem. The nano-scale porous system has increased the difficulty in modeling the multi-physics, multi-phase, multi-medium, and multi-component flow in fractured unconventional reservoirs. The fracture network dominates flow in the unconventional reservoirs but is hard to characterize due to its high complexity and uncertainty. The global Jacobian matrix is usually large in dimensions and ill-conditioned, for which specific techniques are required to enhance the simulator performance.

This dissertation has proposed a workflow to simulate unconventional reservoirs with complex fracture conditions. A multi-component multi-phase reservoir simulator is developed based on the GURU framework with a capacity of multi-porosity-permeability system model which accounts for multiple flow mechanisms including convection, diffusion, and adsorption/desorption. A fully implicit Jacobian system is set up by coupling the reservoir matrix with fracture and wellbore components. High-performance computation techniques are applied to the simulator to enhance the computational performance by CPU- and GPU-based parallelization on the linear solvers and Jacobian construction. A parallel multi-stage preconditioner with respect to the corresponding subspaces of wellbore, fractures, saturation, and pressure is developed with adaptive settings which provides a better convergence rate for iterative solvers. An LGR-EDFM model is developed to capture the large fractures in complex fracture networks. The hydraulic fractures are modeled using a novel imaging-based micro-seismic interpretation method while the natural fracture network is characterized by a fractal model with empirical risk minimization calibration through histograms. All of these govern a numerical simulator with the capability to solve fractured unconventional reservoir problems.

More specific research is also performed to further increase the knowledge and improve the accuracy of fractured unconventional simulations. A pre-stage initialization method is developed based on the fracking schedule simulation, which provides a better estimation of initial pressure and saturation profiles. Water flowback and hydrocarbon flow mechanisms are studied to get a better knowledge of unconventional reservoir physics. A non-Darcy water flux model is proposed based on the boundary layer theory to better describe the water flux in nano-pores. A domain decomposition solver is developed based on the EDFM fracture network to further increase the capability and computational performance. The fracture closure effects are studied using a dynamic EDFM method to better simulate the pressure depletion behavior during production.