The permeability of rock fractures and its variation with effective stress is of considerable interest in broad energy and environmental applications, such as enhanced oil and gas recovery from hydrocarbon reservoirs, geothermal energy extraction, geological carbon storage, among others. The permeability of a rock fracture is a complex function of various static parameters, including fracture mechanical aperture, roughness, and surface contact area, all of which could be functions of dynamic effective stress acting on the fracture walls. The commonly used cubic law is unfit for most applications as it often overestimates the fracture permeability resulting in unreliable predictions. Several models are proposed in the literature with various levels of complexity, accuracy and general applicability. This work establishes a new comprehensive data-driven model to estimate the hydraulic properties of rock fractures as a function of the fracture static characteristics and dynamic effective stress. A dataset measuring fracture permeability in terms of confining stress, fluid pressure, and other rock parameters is compiled to identify potential correlations. We further verify the proposed model with coupled flow-geomechanics simulations. The results show that the trends observed from the dataset are consistent with the theoretical model. We show that our proposed model is superior to all models that we tested from the literature. The coupled workflow offers an efficient approach to characterize the hydraulic response of rock fractures under effective stress. The proposed model is simple, accurate, and efficient, and therefore can be implemented to capture stress-dependent permeability of fracture networks for field-scale reservoir simulations.
|Original language||English (US)|
|Title of host publication||ARMA/DGS/SEG International Geomechanics Symposium 2020, IGS 2020|
|Publisher||American Rock Mechanics Association (ARMA)|
|State||Published - Jan 1 2020|
Bibliographical noteKAUST Repository Item: Exported on 2021-04-13
Acknowledgements: We would like to thank Saudi Aramco for funding this research. We would also like to thank King Abdullah University of Science and Technology (KAUST) for the support and for providing licenses of MATLAB, COMSOL, and Ansys.