Data-Driven Threshold Pressure Correlations for Subsurface Energy Applications Using Stochastic Pore Network Ensembles

Abstract

Fluid invasion into low-permeability formations requires overcoming threshold pressures governed by capillary entry and yield stress mechanisms. Current engineering practice relies on time-consuming core-flood experiments or oversimplified capillary bundle models that neglect three-dimensional pore connectivity, limiting their applicability to subsurface energy systems such as CO2 geological storage, dense nonaqueous phase liquid (NAPL) migration, and hydrocarbon accumulation. This study develops a computationally efficient framework based on a statistically representative ensemble of 6260 stochastic pore networks spanning broad geological variability. Invasion percolation algorithms determine threshold pressure (gradient) under single-phase, two-phase, and coupled mechanisms. Statistical analysis reveals robust power-law scaling between threshold pressure and the characteristic throat radius Rc50 derived from mercury injection capillary pressure curves, enabling rapid prediction without three-dimensional imaging. Furthermore, spatial wettability allocation modes, NAPL-wet preferentially assigned to large, small, or random throats, alter percolation pathways; networks with ≥70% NAPL-wet throats require no threshold pressure regardless of the spatial pattern. Linear superposition of capillary and yield stress contributions introduces acceptable engineering error (<20%) compared to fully coupled calculations. By resolving three-dimensional connectivity and local wettability states, this framework provides an image-free computational tool for evaluating critical invasion pressures in subsurface two-phase displacement processes.