Abstract - Determination of input parameters for dual porosity/permeability models, used to predict multi-phase flow in fractured hydrocarbon reservoirs, is hindered by relatively poor quantitative constraints on the geometrical systematics of natural fracture systems compounded by the inherent difficulties in capturing the effective properties of fracture systems in flow models. We review available quantitative geometrical data derived from outcrop studies of fracture systems, and briefly consider how these data are best incorporated either explicitly or implicitly in flow simulations of fractured reservoirs.
Flow within so-called 'fractured reservoirs' is strongly controlled by the geometry and porosity-permeability characteristics of fractures contained within otherwise low permeability reservoir rocks. Fractures that are conductive to fluid flow are conveniently described under two basic headings: faults and fractures. In a typical fractured reservoir, faults will occupy the larger-scale geometry of the combined fault/fracture system, whereas fractures (such as joints, veins and microfractures) will be more numerous and occupy the small-scale geometry of the system. A combined fault and fracture system is therefore analogous to a dual porosity/permeability system, with faults controlling the large-scale flow connectivity and with fractures representing the notional matrix permeability. Although seismic data provide some constraints on the geometry of large-scale faults within a reservoir and well data provide constraints on a range of fracture attributes for both faults and fractures, neither provide direct geometrical measures of the network properties and connectivity of the system. The incorporation of fracture system geometry in fracture models, therefore requires constraints either from theoretical concepts or from quantitative descriptions of well-exposed natural fracture systems.
Using well exposed outcrop analogues from Denmark, Ireland, Norway, Saudi Arabia and Italy, we describe the principal quantitatitive geometrical and poro-perm characteristics of faults and fractures. Fault systems often display scaling properties which are characteristic of fractal geometries, in which faults have a broad size range characterised by power-law distributions and a positive correlation between fault dimensions and maximum displacement. Relations supporting a weak correlation between fault displacement and both fault rock thickness and damage zone width, combined with an assessment of upscaling issues provide useful constraints on the conditioning of seismically imaged faults explicitly incorporated in flow simulations. Issues such as the channelised nature of fluid flow within faults are very poorly constrained and the absolute values of the permeabilities of seismically imaged faults are best constrained from well test data. Implicit incorporation, as equivalent properties, of the effects of sub-seismic faults in flow models is possible but is complicated by the inherent scale independence of fractal systems (a meaningful REV, i.e. representative elementary volume, may not exist) and the often relatively poor constraints on the spatial distributions of sub-seismic faults.
The geometry and scaling of fracture systems (joints, veins and microfractures), is characterised by two end-members. Where layers are mechanically decoupled, the fracture systems are 'stratabound', i.e. fracture networks in adjacent layers are independent, fracture sizes span a narrow scale range and fracture spacing is regular. In more massive rocks, fracture systems are 'non-stratabound', with fractures spanning a large scale range (often with power-law length distributions), and spatially clustered. These differences influence the fluid flow properties of the fractured rock mass, by their effects on the size of fractures that dominate flow, on the rock mass volume for which an equivalent porous media approach is valid, and on the bulk rock permeability anisotropy. Even for non-stratabound fracture systems, if an REV exists below the scale of a grid block the effective properties of fracture systems can be derived and incorporated in reservoir flow models from a combination of discrete fracture flow modelling and a model relating fracture system permeability with fracture density and aperture distributions.
Abstract of talk given to:
EAGE Workshop on Modelling and monitoring fractured reservoirs, Helsinki, June 1999