Abstract
Semi-automation of hydraulic fracturing treatment designs often necessitates the application of simplified predictive models. Such models can only incorporate a limited subset of the relevant rock mechanical properties and an approximate representation of the stress state. This paper demonstrates the fundamental influence of three-dimensional stress states on the propagation of hydraulic fractures in coal seam gas (CSG) wells, and contrasts these results with those from two-dimensional simulations conducted in a one-dimensional stress state.
A three-dimensional, finite element-discrete element (FEM-DEM) model of a single well stage was developed as the basis for this study. This synthetic well was informed by case studies from the Surat Basin, Queensland, featuring varying complexity of key geomechnical factors. These include the existence of ∼30 coal seams within a gross rock column of more than 300 m, stress states that vary both laterally and vertically, ductile rock properties, and varying natural fracture densities and orientations. The developed model captures the full tensor description of stress, poroelastic-plastic modelling of the rock and coal, fully coupled fluid flow, and explicit modelling of fracturing.
The stress state was parametrically defined so that normal, strike-slip and reverse faulting conditions could be imposed and the magnitude of stresses varied to capture the appropriate range of varying conditions. A single perforation cluster was then used to induce a hydraulic fracture in an isotropic medium. Hydraulic fracture propagation (and propagation complexity) is influenced significantly by differential stresses, stress orientations and relative stress magnitudes. None of these are captured in two-dimensional simulations using a one-dimensional stress characterisation which is commonly derived from one-dimensional wellbore stress models.
The findings of this work clearly demonstrate the ability of fractures to turn and grow preferentially when they are not constrained to a two-dimensional plane. It also shows how the initiation of fractures (i.e. orientation to stress) impacts the propagation complexity of hydraulic fractures from the direction of maximum principal stress. In general, this paper highlights the benefit of incorporating the three-dimensionality of key geomechnical parameters when designing hydraulic fracturing stimulation treatments. Future work will incorporate greater reservoir detail (e.g. pressure-dependence, heterogeneity of stress and material properties) to further investigate fracture containment and reorientation.