Abstract - The hydraulic properties of faults are included in reservoir flow models using transmissibility multipliers, and their geometric properties are incorporated explicitly through the definition of the depths of each grid-block. Transmissibility multipliers for faulted grid-block to grid-block connections are a function of the permeabilities of the grid-blocks as well as the permeability and thickness of the fault zone. A constant multiplier applied to a fault in a sedimentologically heterogeneous reservoir model therefore represents a heterogeneous fault. This implicit fault heterogeneity is perhaps seldom recognised by the reservoir engineer and is of a geologically meaningless form arising entirely from the dependencies embedded in the operation of the multiplier.
A recently developed method for estimating fault transmissibility multipliers in sandstone/shale sequences nests geological fault characterisation within the framework of commercial flow simulators (Petroleum Geoscience articles of Walsh et al. (1998) and Manzocchi et al. (1999)). For each connection across a fault surface, fault rock permeability is calculated as an empirical function of the throw of the fault and the Shale Gouge Ratio (SGR), the clay fraction of the sequence that has moved past each point on the fault. Estimation of fault rock permeability is constrained by empirical data relating permeability to the phyllosilicate content of fault rocks, which is taken to correlate with SGR. Fault zone thickness is calculated as a function of throw. Application of the SGR algorithm at the resolution of the reservoir flow model is easily implemented and provides a geologically significant fault zone heterogeneity structure. The principal strengths of the method are that it is geologically-driven and plausible and that the flow response of faulted reservoirs can be interpreted in terms of meaningful fault-related geometric and hydraulic properties.
We anticipate however that application of the method will require refinements of the methodology to account for a variety of geological factors and circumstances, including the following:
(i) Fault rock permeability and thickness are known to vary considerably over short distances along fault traces, but information about the correlation lengths of fault zone structure is difficult to gather even at outcrop. The basic method assumes that the level and length-scales of fault zone permeability and thickness heterogeneity are captured at correlation lengths below that of a typical flow simulation grid block, i.e. ca 100m. If correlation lengths are greater, then stochastic simulation of both fault properties and thicknesses is required, a procedure that results in increased flow across faults.
(ii) The effects of errors in the definition of displacements of seismically imaged faults can be simulated stochastically, but will vary from reservoir to reservoir.
(iii) Intact fault relays that are irresolvable at the discretisation of a reservoir flow model, but provide continuous high permeability flow paths through the fault, can be implicitly included in flow models.
(iv) The complexity of fault zone structure at sub-seismic scales, such as the existence of sub-seismic relays, will also increase the flow across faults, but the extent of this increase will vary from reservoir to reservoir, and requires improved constraints on, for example, the development and scaling of segmented faults and on the geometries of damage zones.
(v) Diagenetic effects, particularly those which are depth-dependent, will generally lead to a decrease in fault permeability.
The significance of these, and other, factors should become clearer from future applications of the method on hydrocarbon reservoirs but will also benefit from improved constraints arising from basic research into the geometric and hydraulic properties of faults. Another outstanding issue, however, is our current inability to include fault capillary properties in flow simulators, and we therefore briefly outline a methodology for their inclusion; development and implementation of this method is at an early stage.
Abstract of talk given to:
Faults, Fractures and Reservoir Flow. FORCE Seminar, Stavanger, October 1999