A critical review of coal permeability models

Abstract

The evolution of coal permeability is very complex under the influence of coupled multiple processes. This is the primary motivation why numerous coal permeability models have been developed over the last decades. Although great efforts have been made to evaluate these models, the root causes of discrepancies between lab/field observations and model predictions are still not identified. The objectives of this study are to address this problem from the perspective of model selves and explore their implications for further improvements. In this study, we collected all coal permeability models available in the literature. Through analysis, we conclude that all models can be characterized by a combination of internal structure of coal, boundary conditions and equilibrium state within the controlled volume. Different combinations lead to two major categories of coal permeability models. One category is structure-based equilibrium models including matchstick, cubic and rock bridge models. The other category is structure-based non-equilibrium models. The equilibrium models only serve as the upper and lower envelopes of experimental data while the non-equilibrium models can explain the data in-between. Further analysis concludes that if local equilibrium is achieved, gas pressure and its associated swelling strain distribute uniformly throughout the entire volume and that if not achieved, both pressure and swelling strain distribute non-uniformly. These conclusions suggest that the exclusion of equilibration process between fracture and matrix systems is the root cause of discrepancies between lab/field observations and model predictions and that future research work should integrate rock structure, boundary conditions and equilibration process in coal permeability model. This inclusion of transient process within the controlled volume represents a leap of knowledge from equilibrium to non-equilibrium theory and opens up a new realm for unconventional gas reservoir modelling.