For the design of production strategies one must identify how faults that are detected in seismic surveys influence fluid flow in hydrocarbon reservoirs. In this study we use the Entrada sandstone formation, Arches Natl. Park, Utah, as an outcrop analog of a faulted sandstone reservoir. Field measurements of the geometry, thickness, and layered permeability of the normal faults that crosscut this sandstone are incorporated in two-dimensional (2D) fluid-flow models, facilitating the simulation of fault signatures in transient well tests with a high-resolution finite-element method.
In our simulations, structure and inhomogeneous permeability lead to fault signatures in derivative plots that differ significantly from those of idealized impermeable barriers. Highly permeable slip planes facing the well create the illusion of a nonsealing or nonexisting fault because remote fluids are focused along the slip planes toward the well. Measured low, fault-normal permeability shields fault-bounded blocks of the analog sandstone reservoir from drawdown and large pressure differentials build up across faults during production.
Thus, test data from a single well are insufficient to assess the flow properties of a nearby fault with an inhomogeneous permeability like the normal faults in the Arches Natl. Park. Test responses from multiple wells need to be considered to detect fault segmentation or fault terminations even if the general fault trend is underpinned by seismic data. Flow paths in the reservoir during production are complex. Importantly, formation water is likely to flow into the reservoir along the permeable slip planes of the normal faults.
Hydrocarbon formations are often cross-cut and offset along normal faults, which can act as both flow barriers1,2 and/or conduits for fluids.3–7 Whether a particular fault impedes fault-normal fluid flow during production or whether it connects the reservoir to other permeable domains in the sedimentary pile cannot be inferred from the seismic or well-log data.
Fluid-flow properties of faults may be estimated from transient well testing.8–12 Common methods of testing examine the change of fluid pressure in a well while it is being produced at a constant rate (drawdown test) or shut in after a prolonged production period (buildup test). The well tests are interpreted by comparison with analytical "type" curves for a range of reservoir geometries and permeabilities, or nonlinear least squares estimation techniques ("automated type-curve or history matching") are used for the estimation of reservoir parameters from well-test data. Type curves are commonly plotted in derivative plots that display the dimensionless pressure in the wellbore, pD, and the dimensionless rate of pressure change with time multiplied by time, t(?p/?t) (see Fig. 1).13,14 Dimensionless time, tD, is normalized for permeability and is frequently divided by the dimensionless wellbore storage, cD, such that derivative plots obtained for different reservoirs can be compared. Wellbore storage effects are not addressed in this paper. In consistent units, pD and tD can be defined as
Equations 1 through 3
where pi, pw, k, q, B, µ, c t, rw, V, and C, are the initial reservoir pressure, pressure at the wellbore, permeability, constant rate of production, formation factor, fluid viscosity, porosity, total system compressibility, wellbore radius, fluid volume, and the storage capacity of the well, respectively.15 These definitions and notation will be used throughout this analysis.
If a reservoir is homogeneous and infinite, constant-rate production is accompanied by an initially rapid change of fluid pressure followed by a period of radial flow and constant-rate pressure decline (Fig. 1). Interaction with an inhomogeneity like a fault will bring the constant-rate pressure-decline period to an end. The rate of decline will either increase if the fault has a lower permeability than the reservoir (Fig. 1; e.g., Ref. 11) or decrease if the fault is a highly permeable fracture.16 Such rate changes are reflected in a distortion of the radial drawdown pattern, which would persist in an isotropic, infinite reservoir.
The strategy of estimating fault properties by comparing their signatures in derivative plots with existing analytical solutions for linear reservoir inhomogeneities and combinations there of is widely applied (see, for instance, Refs. 8, 11, 12, and 44). Commonly, however, faults represent composites of several semiplanar zones with different deformation structures and hydrological properties.5,6,17-19 This topology implies an inhomogeneous fault permeability for which we did not find analytical solutions in the literature.