Geology and Mechanics of the Basin-Centered Gas Accumulation, Piceance Basin, Colorado

Stephen Cumella¹ and Jay Scheevel²
¹ Bill Barrett Corporation, Denver, CO
² Scheevel Geo Technologies, Grand Junction, CO

Introduction

A very large basin-centered gas accumulation in the Williams Fork Formation of the Mesaverde Group is currently being actively developed at 10-acre density in the southern part of the Piceance Basin. Ten-acre density is necessary to efficiently develop a reasonable amount of the gas-in-place due to the very low (microdarcy) permeability and the highly lenticular nature of the fluvial sandstone reservoirs. Figure 1 is a schematic cross section that illustrates the key features of the Piceance basin-centered gas accumulation. The Rollins Sandstone is a thick, laterally extensive marine sandstone unit at the base of the Williams Fork. Within the area of commercial Williams Fork gas production, gas is produced from a continuously gas-saturated interval of 1,500-2,400 feet. A transition zone of mixed gas- and water-saturated sandstones overlies the continuously gas-saturated interval. Pervasive natural fracturing enhances reservoir permeability in order to allow commercial production over 14-township area. This productive area of the Williams Fork is continually expanding as a result of current active exploration for this basin-centered gas resource.

Fracturing, Gas Migration and Fluid Pressure

During maximum burial a large amount of gas was generated and expelled from the Cameo coal interval in the lowermost part of the Williams Fork. The Cameo is the primary source of Williams Fork gas. Both the lateral and vertical distribution of gas and the distribution of overpressure are directly linked to physical migration of the gas phase as it moves upward and outward from the Cameo deep-basin coals.

The continuously gas-saturated reservoir intervals in the Williams Fork are complexly interconnected by fracturing, faulting and original fluvial depositional geometry. The degree to which any given reservoir sand becomes gas saturated is determined by the degree to which the depositional geometry, fracturing, and faulting connect it to the source of the gas in the Cameo coal.

Overpressure

Pressure gradients, which can be as high as 0.8 psi/ft in the lower part of the Williams Fork in the structurally deeper part of the basin, decrease upward to hydrostatic gradients (0.43 psi/ft) near the top of the continuously gas-saturated interval. Hydrostatic conditions continue upward to the top of the Williams Fork. Pressure gradients also decrease with shallower burial depths towards the flank of the basin. The rate of migration was inhibited due to the very low permeability and discontinuous nature of the Williams Fork sandstones. Consistent with this mechanism, the magnitude of overpressuring follows a trend of increasing pressure with increasing stratigraphic depth with the highest values in direct proximity to the coal intervals.

Stress State, Effective Stress and Fracturing

The overall distribution and pressure of the gas in the Williams Fork is probably the direct result of pore-pressure assisted fracturing and subsequent migration through the induced fracture systems. The orientation of the fracture systems’ orientations are determined by tectonic stresses, but the distribution and intensity of fracturing is most strongly influenced by the history of overpressuring during gas charging.

Most fractures in the Mesaverde of the Piceance Basin are extension fractures. These types of fractures are very common in all rocks, are usually vertical, open in a direction normal to their own plane, and are often referred to as joints. High density of extension fractures can occur in rocks that have achieved a net tensile stress condition, and when the strain limitations in the subsurface will also support the opening of vertical fractures.

Our analysis emphasizes the impact of the magnitude of pore-pressure on the internal stresses and strains as they relate to extension fracturing. We find that effect of high pore pressure is to compress/shrink the individual sand grains uniformly in all directions, essentially creating a rock of lesser volume. The vertical component of this shrinkage can easily be accommodated by free vertical movement at the earth’s surface, but lateral shrinkage in the subsurface is resisted by the semi-infinite horizontal extent of confining rock layers. Horizontal grain shrinkage must be accommodated by lessening of lateral compressive stress (resulting in an equivalent lateral elastic elongation) as originally suggested by Miller (1995). As pressure-induced grain shrinkage increases, lateral normal stress decreases, until under certain conditions of pore-pressure, the rocks will experience tensile horizontal stress and resultant fracturing.

Figure 2 illustrates the predictions from the elastic model illustrating the concept. In Figure 2, the abscissa depicts the Young’s modulus of a given rock. The ordinate depicts the Poisson’s ratio. The sandstones of the Mesaverde MWX test site (near Rulison Field, Piceance Basin) are plotted in blue dots on this plot. The mudstones of the MWX site are plotted as red triangles.

The curves on Figure 2 represent the boundary between compressional and tensional conditions (C – T on the curve) for the given pore pressure gradient and a burial depth of 6000 ft. The depth dependency of the C-T boundary curves is minimal. The dashed curve represents the boundary between compression and tension for a pore-pressure gradient of 0.433 at a burial depth of 18,000 ft. It is offset only slightly left from the 0.433 psi/ft gradient at 6000 ft. burial depth. One can think of the boundary curves as representing “critical pore pressure” to support fracturing in a rock with specific elastic constant values.

Figure 2 illustrates that a pore-pressure gradient between 0.5 and 0.7 psi/ft will subject most of the sandstones to a tensional stress environment. It requires between 0.6 to 0.8 psi/ft to subject the mudstones to a tensional stress environment, with most of the mudstones falling in the higher range of pore-pressures. Mudstones also have some non-elastic and time-dependent behavior, which would require even higher pore-pressures for fracturing, however we did not consider time-dependent behavior in this analysis.

The elastic analysis also indicates that subsurface strain conditions should require and additional 0.05 to 0.10 psi/ft pressure gradient greater than that required to achieve the critical pore-pressure (tensional stress), in order to ensure the that any fractures that are formed will be open and conducting flow.

Gas Migration and Pressure Gradient

Given a sufficient rate of gas generation in the Cameo interval, critical pore pressure and fracturing can be expected to be achieved for a specific combination of depth, host-rock elastic properties, and pore-pressure gradient. However, based on the elastic strain analysis, it is expected that higher pore-pressures would be required to fracture all rock types, and to open those fractures sufficiently to conduct gas from the deepest section to successively shallower horizons.

The first place one might expect fracture assisted migration to occur is within the gas-generative Cameo interval. When gas generation causes critical pore pressure to be exceeded, the rock fractures, and the rate of gas escape from the overpressured rock rapidly increases, stabilizing or reducing its pressure and allowing gas to flow into a lower-pressured adjacent sand body, more distant from the source. One might expect this process to be repeated in a daisy-chain fashion, moving outward and upward from the gas generative parts of the Cameo as each successively shallower unit achieves the critical pore-pressure.

Ultimately, the ability of a sand to sustain overpressured conditions will depend on the balance of the rate of gas entry with the rate of gas escape from that sand. The entry rate for a sand is determined by the critical pore pressure of its neighboring gas-saturated sand body and whether a fractured conduit connecting the sands is transmitting gas. The gas escape rate is initially determined by the matrix permeability, but since this rate is greatly enhanced by the creation of fractures, the upper limit for pressure in any given sand can be reasonably expected to be near the critical pore-pressure required for fracturing the sand body or its bounding facies.

The current pressure gradient within the continuously gas-saturated interval is observed to increase gradually with depth, from normal hydrostatic at the top of the gas-saturated interval, to greater than 0.8 psi/ft in the Cameo interval. This transition can be observed reliably in the MWX locality (Spencer, 1989). The unusual nature of this distribution of pressures can be illustrated by comparison with a standard pressure scenario in a conventional trap (Figure 3a). For a continuous gas phase in equilibrium, pressures are expected to increase from near hydrostatic at the base of the gas column to overpressured at the top of the column (Figure 3a). This is because a continuous gas phase demonstrates a lower pressure-versus-depth gradient than does a more dense fluid such as water (buoyancy effect). The difference in fluid pressure (degree of overpressuring) in such a conventional trap is expected to decrease with of depth as in Figure 3a. This is not the case in the continuously gas-saturated section of the Piceance Basin where the overpressuring is least at the top of the gas and increases to a maximum deep in the section (Figure 3b).

Explanations for the unusual pressure profile in the Williams Fork (Figure 3b) are tied to the rates of gas entry and exit from individual sands being controlled by the critical pore pressure required for fracturing.

Once fractured, gas pressure would stabilize and gas would flow through the sand at a stable rate, by virtue of the greater permeability afforded by fracturing. By continuously feeding gas through successively shallower sands, each with lower critical pore pressure thresholds, a dynamic equilibrium could establish the general form of the observed Piceance pressure gradient.

An alternative hypothesis involves the same concept of gas entry and escape on a sand by sand basis, but would also require that, once the pressure in a given sand fell below the critical pore pressure, the fracture network could close sufficiently to produce a statically trapped condition for that sand. A large number of stacked sands in such a statically trapped condition could conceivably “freeze” the pressures just below the critical pore-pressure. Such a vertical series of “frozen” pressure compartments could yield that the pressure gradient observed in the Piceance.

The ultimate practical implication of either of these two explanations for the observed Piceance gas-phase pressure distribution is that the reservoir sands within the vertical gas column cannot be in immediate pressure communication as they would be in a single reservoir (depicted in Figure 3a). This is despite the observation of a continuous gas phase. The validity of this statement is strongly supported by the fact that effective exploitation of Williams Fork reservoir in the Piceance is now known to require some or all of the following: 1) dense infill-drilling, 2) large vertical completion-intervals and 3) massive artificial fracturing in order to efficiently produce the resource and to make economic ultimate recoveries of gas deliverable to the a single well.

Possible Top Seals and Reservoir Compartmentalization

The top of the continuously gas-saturated interval in the Williams Fork does not appear to have an obvious top seal. In Grand Valley field (T6 and 7S-R96W) the top of continuous gas saturation is about 1,800 ft above the top of the Rollins. Moving eastward and deeper in the basin, the top of continuous gas rises to 2,000 ft above the Rollins in Parachute field (T6 and 7S-R95W), and moving farther east rises to 2,400 ft in Rulison field (T6 and 7S-R94W). No clearly definable shale bed caps the continuous gas-saturated interval in these fields. However, intervals with higher shale content can act as local top seals. These intervals are more easily visualized using color-filled smoothed gamma ray cross sections depicting a vertically averaged measurement of shale content. Figure 4 shows an example of a smoothed gamma ray and two examples of smoothed gamma ray cross sections. The log shown on the left of Figure 4a is the original spiky gamma ray trace. The trace on the right is the smoothed version gamma ray log. Figure 4b is a west-to-east cross section across Mamm Creek field. Each well on the section has the unsmoothed gamma ray curve on the left and the smoothed gamma ray on the right. The smoothed gamma ray is colored filled with low gamma ray values having red colors and high gamma ray values having blue colors. The gross stratigraphic architecture of the Williams Fork is much better visualized with the smoothed gamma ray. With regards to the preceding discussion, gas generation in the Cameo caused critical pore pressure to be exceeded, resulting in fracturing and upwards migration to lower-pressured overlying sand bodies. The top of continuous gas, which is shown in Figure 3a, coincides with the lower part of a thick interval of high shale content in the upper part of the Williams Fork. This shaly interval is less fracture prone and the upward movement of gas through the shaly interval was impeded. Locally, the top of continuous gas rises abruptly within structural compartments, as seen on the third well from the left. Within such structural compartments, increased fracturing may be caused by an increase in local differential stress. In such a circumstance, pore-pressure assisted fracturing and gas migration would rise higher in the section than in surrounding areas that experienced only the regional components of stress and strain. Smoothed gamma ray cross sections can be used to identify similar shaly intervals that are present in Grand Valley, Parachute, and Rulison fields. Figure 4c shows a cross section through numerous wells in Parachute field. The arrows point to two shaly intervals that locally coincide with the top of continuous gas saturation.

Figure 1. Schematic Cross Section Illustrating Key Features of the Piceance Basin-Centered Gas.