- PDF LinkChapter PDF
Stephen P. Cumella, 2009. "Geology of the Piceance Mesaverde Gas Accumulation", Unconventional Energy Resources: Making the Unconventional Conventional, Tim Carr, Tony D’Agostino, William Ambrose, Jack Pashin, Norman C. Rosen
Download citation file:
Aggressive development of the Mesaverde gas accumulation in the Piceance Basin over the past decade has demonstrated that a commercial gas resource is present in much of the deeper part of the basin. Unlike tight gas resources in some other basins (e.g., the greater Green River Basin), commercial production doesn’t appear to be limited to specific fairways or sweet spots. There appears to have been a sufficient gas source within in situ coals and underlying marine shales to pervasively gas charge up to 3500 ft of the Mesaverde. An extensive vertical fracture system has resulted from over pressuring from hydrocarbon generation. Laramide tectonic fractures are also locally abundant. This fracture system has enabled vertical gas migration within an otherwise very low permeability system.
In spite of being one of the oldest areas of tight gas production in the Rocky Mountain region, innovations in drilling and completion technology continue to expand the area of commercial production. Directional drilling has allowed over 20 bottom-hole locations to be accessed from a single surface location, and laterals reach up to 5000 ft. Microseismic imaging of hydraulic fracture stimulation has helped place bottom-hole locations optimally with regard to highly elliptical drainage patterns. Large water volume hydraulic fracturing has dramatically improved estimated ultimate recoveries (EURs) of wells in some areas. Also, unconventional pay picking has added significant resources that were not previously developed.
A regionally continuous gas accumulation in the Mesaverde Group in the Piceance Basin is currently producing about 1.3 billion cubic feet per day. Most of the production is from a continuously gas-saturated interval of discontinuous fluvial sandstones in the Williams Fork Formation. The sandstones typically have permeabilities in the single digit microdarcy range and average porosities of about 9%. The thickness of the gas-saturated interval is shown in Figure 1. The gas-saturated interval is thickest in the deepest part of the basin and thins onto the basin flanks (Figs. 1 and 2). Most of the production is from fields along the Colorado River drainage, where topography and infrastructure have facilitated development. Topography, lack of roads, and lack of infrastructure present economic and environmental challenges in the northern Piceance, but development is accelerating.
Stratigraphy of the Mesaverde Group
The stratigraphy of the Mesaverde Group in the Piceance Basin is illustrated in Figure 3. The Mesaverde Group is composed of the Iles and Williams Fork formations. The Corcoran, Cozzette, and Rollins Members of the Iles Formation are marine sandstone units that were deposited during eastward regressions of the Cretaceous seaway. The Williams Fork is composed mainly of nonmarine deposits, with the exception of the upper and middle sandstone intervals that are present in the southeastern Piceance Basin. The lower Williams Fork contains coals, the aggregate thickness of which ranges from 40 to 180 ft. A detailed summary of the stratigraphy of the Mesaverde Group is presented in Johnson (1989), Hettinger and Kirschbaum (2003), Cole and Cumella (2003), and Kirschbaum and Hettinger (2005).
Correlation of stratigraphic intervals in the thick fluvial Williams Fork fluvial interval is difficult due to the discontinuous nature of the fluvial sandstones. However, smoothed gamma-ray logs can reveal gross stratigraphic architecture that is not readily apparent on conventional gamma-ray logs. Figure 4a shows a comparison of a conventional and a smoothed gamma-ray log for the entire Williams Fork interval from a well in Mamm Creek Field. There is a low frequency component to the gamma-ray log that is more obvious on the smoothed gamma-ray curve. Figure 4b is a cross section constructed using smoothed gamma-ray curves; warm colors (yellow-red) correspond to low gamma-ray values and cool colors (green-blue) correspond to high gamma-ray values. Alternating high- and low-net-to-gross intervals are more easily recognized on the smoothed gamma-ray display. In general, sand-rich intervals probably accumulate during times of low accommodation when fluvial channel sands amalgamate. Increased accommodation results in preservation of overbank and other fine-grained fluvial deposits, and fluvial channel sandstones are fewer and more discontinuous. Some of the thicker sand-rich intervals in the Williams Fork may be related to larger order sequence stratigraphic events and may overlie sequence boundaries (Patterson et al., 2003). The intervals labeled lowstand systems track and Ohio Creek on Figure 4b are two such intervals.
The cross-section shown in Figure 5 is constructed from very closely spaced wells (average 330 ft apart). The cross-section interval straddles the thick sand-rich interval labeled low stand systems tract in Figure 4b. The discontinuity of the sandstones is evident, even in the sand-rich interval.
The Piceance Basin is one of a number of structural depressions created in the Rocky Mountain region during the Laramide orogeny (Late Cretaceous-early Tertiary time). It is bounded on the southwest by the Uncompahgre uplift, on the east by the White River uplift, on the west by the Douglas Creek arch, on the southeast by the Gunnison uplift, on the northwest by the Uinta Mountains, and on the northeast by the Axial Basin anticline (Fig. 1). The Piceance Basin is strongly asymmetric and the synclinal axis is near the steeply dipping eastern flank of the basin (Figs. 1 and 2). Structural dips are very steep along the eastern part of the basin (Grand Hogback). The southwest flank of the basin is gently dipping’ dips typically are a degree or less. Outcrops of Tertiary strata (Uinta, Green River, and Wasatch formations) dominate the surface exposures within the basin, and rocks of the Mesaverde Group and Mancos Shale occur around the edges. In the structurally deepest parts of the basin, rocks of the Mesaverde Group are buried up to 4000 feet below sea level.
Model for the Piceance Mesaverde Gas Accumulation
The model for the Mesaverde gas accumulation in the Piceance Basin is illustrated in Figure 6. The primary source of gas for this accumulation is the coals in the lower Williams Fork (Johnson, 1989; Yurewicz et al., 2008), but the Mancos Shale may also be a significant source, especially where major fault zones can provide a conduit for the gas. Recent successful completions in the Mancos have demonstrated that a large gas resource is present (Antero Resources, 2009). Most of the Mancos gas has probably not migrated out of the formation due to its very low permeability. Major fault zones may provide the permeability to allow gas to migrate vertically though the Mancos.
Total coal thickness in the lower Williams Fork ranges from 40 to 180 ft (Johnson, 1989). Thermal maturities of the coal in the deeper part of the basin are high (vitrinite reflectance values of 1.0-2.0) (Johnson, 1989), especially in the southern Piceance, which has had a high thermal gradient relative to the northern Piceance (Yurewicz et al., 2008). Calculated volumes of gas generated from these coals are high, as much as 1100 bcf/mi2 in some parts of the basin (Yurewicz et al., 2008). Johnson (1989) estimates that significant gas generation from the coals began in the early Eocene, after diagenesis had greatly reduced sandstone permeabilities (Pittman et al., 1989). The very low permeability of the Williams Fork sandstones greatly restricted migration of gas generated from the lower Williams Fork coals, resulting in high overpressures (Johnson, 1989; Law, 2002). To reach the low water saturations that are typical of productive Williams Fork sandstones (≤40%), very high capillary pressures are required (Brown, 2005). The limited lateral extent (on the order of a few hundred feet, Cole and Cumella, 2003) of most lower Williams Fork sandstones preclude achieving high capillary pressure as a result of buoyancy.
Over- pressuring due to gas generation is the best explanation of how the high capillary pressures are overcome and low water saturations are achieved in microdarcy permeability sandstones. After gas displaced the water in the Williams Fork sandstones to irreducible levels, pressure builds up to the point so that the fracture gradient of the intervening shales is exceeded. At this point, gas is expelled vertically through the fractures into shallower sandstones. This process has occurred on a massive scale in the deeper part of the Piceance, resulting in a pervasive fracture system (red arrows in Fig. 6) that provides enough reservoir permeability to make the microdarcy matrix permeability sandstones commercially productive.
Evidence for this pervasive fracture system is provided by the great disparity between the sandstone matrix permeabilities (microdarcies) and reservoir permeabilities measured by long-term pressure tests of individual Williams Fork sandstone reservoirs (10’s to 100’s of microdarcies) (Lorenz, 1989). Furthermore, diagnostic fracture-injection/falloff test (DFIT) data from 810 tests showed that 61% of the tests had pressure-dependent leak-off, indicating the presence of natural fractures (Craig et al, 2005). In a study integrating well logs, core measurements, numerical modeling, borehole seismic, and surface seismic data, Lewallen et al. (2009) present preliminary results that indicate that: (1) fractures occur in sandstone intervals; (2) shear-wave anomalies have positive correlation to fracture occurrence in the borehole; and (3) borehole seismic measurements of shear-wave particle motion, travel time, and energy distribution as a function of azimuth are consistent with aligned vertical fractures.
Figure 7a shows the azimuth of the fast arrivals for the entire zero-offset VSP sourced by horizontal vibrators from their study. Throughout the reservoir section, the azimuth of the fast wave averages about 125°, an orientation that agrees closely with the natural fracture orientation interpreted from borehole image logs. Figure 7b shows that the shear-wave travel time delays increase almost linearly throughout the reservoir interval, suggesting that vertical natural fractures of similar orientation are present throughout the reservoir.
The process of over-pressuring, fracturing, and vertical migration of gas continued to progress upwards until the pressure diminished farther away from the source of the source of the overpressure, the lower Williams Fork coals. Another significant factor impeding the upwards migration of gas migration is the presence of a thick argillaceous interval in the upper part of the Williams Fork (labeled highstand and transgressive systems tract in Fig. 4b). This thick shaly interval is present throughout Mamm Creek Field and the top of continuous gas commonly occurs at the base of the shaly interval (Fig. 4b). A more detailed view of this stratigraphic interval is shown in Figure 5. The red line shows top of continuous gas saturation and it is located at the base of a thick shaly interval in most wells on the cross-section. In the four wells on the right side of the cross section, top gas has risen about 170 ft. Gas migration above the normal top gas interval may have been aided by the lower clay content of this interval in these wells. In other fields in the Piceance Basin, the top of continuous gas occurs beneath other shale intervals that likely acted as barriers to upward migration. (See Fig. 4c in Cumella and Scheevel, 2005.)
The thickness of the continuously gas saturated interval tends to be similar over large portions of different fields in the southern Piceance (e.g., 1800 ft in Grand Valley, 2000 ft in Parachute, 2400 ft in Rulison, and 2000 ft in Mamm Creek). Definitions of continuous gas saturation vary, but in this paper the top of continuous gas saturation is the top of the uppermost gas-charged sandstone below the deepest wet sandstone. Gas-charged sandstones are present above the deepest wet sandstone in what is commonly referred to the as the transition zone.
Locally, the continuously gas-saturated interval thickens significantly. A good example of this thickening is an area of Mamm Creek Field known as Gibson Gulch graben (Cumella and Scheevel, 2008). This structural feature is a seismically defined, northeast-southwest trending graben having approximately 100 ft of displacement at the top of the Williams Fork. The gas-saturated interval thickens from approximately 2000 ft in the areas around the graben to as much as 3000 ft within the graben. (See Fig. 12 in Cumella and Scheevel, 2008.) The proposed explanation for this thickening is a combination of increased natural fracturing associated with this structural feature and vertical migration along faults of high-pressured gas from the underlying Mancos (labeled GGG in Fig. 6).
Fault zones may play an important role in charging shallower sandstones in other parts of the basin, either as part of the continuous gas column or as gas-charged sandstones in the transition zone (Fig. 6). Possible examples are Rulison Field, which is located on complexly faulted northwest-trending anticlinal nose (Cumella and Ostby, 2003), and wells located along the various forks of Parachute Creek in T5N, R95-96W. The linear valleys of these forks are likely developed along fault zones and some of the highly productive wells (e.g., Piceance Gas Resources 36-24D Chevron located in sec 36-T5S-96W) are a result of gas-charged sandstones extending up to near the top of the Williams Fork.
The reservoir quality of the Williams Fork sandstones improves significantly in the shallower sandstones (Fig. 8). Therefore, the shallower sandstones can be prolific reservoirs if they are fully gas charged. Sandstones in the lower 1000-2000 ft of the Williams Fork typically have single-digit microdarcy permeabilities (Fig. 8). In the example well shown in Figure 8, permeabilities increase into the tens of micro-darcy range near the top of the gas-saturated interval. This interval is located in the upper part of the thick sand-rich interval labeled low stand systems tract in Figure 4b. This sand-rich interval is present throughout much of the southern Piceance Basin, and production logs indicate that approximately half of Mamm Creek Field’s gas production comes from this interval. The graph on the right side of the cross section shown in Figure 4b shows gas production from 25 wells in Mamm Creek Field. This graph plots the individual contribution from each productive zone in all 25 wells. In other words, each bar on the graph represents the gas production from a single zone in a single well. The graph shows that gas production is obtained from all intervals of the Williams Fork up to the top of gas as well as significant production from the Rollins. Because of their higher permeability, the most prolific zones are the shallower Williams Fork sandstones. It is important to note that these sandstones are only charged in areas where vertical gas migration is aided by faulting (Fig. 6). These areas are aerially limited, therefore the total reserves in the prolific shallower intervals are much less than the pervasive accumulation that is present below the regional top of gas.
The permeability of some of the shallower sandstones is high enough for gas to segregate from water by buoyancy. Figure 9 shows an example of a conventional accumulation that has formed in a high permeability sandstone in the upper part of the Williams Fork, about 900 ft above the top of continuous gas in this area. The thick sandstone present in the three wells on the left side of the cross section has pinched out updip in the right well. The structure map shows the pinchout of the sandstone updip onto the northwest-plunging nose of the Rulison anticline. The neutron-density logs show crossover that is shaded red in the second well from the right, indicating gas saturation that is not present in the downdip wells on the left side of the cross section.
Tight-Gas Development Technology
In spite of being one of the oldest areas of tight gas production in the Rocky Mountain region, innovations in drilling and completion technology continue to expand the area of commercial production. Directional drilling has allowed over 20 bottom-hole locations to be accessed from a single surface location, from which laterals have reaches of up to 5000 ft. Microseismic imaging of hydraulic fracture stimulation has helped place bottom-hole locations optimally with regard to highly elliptical drainage patterns. Large water volume hydraulic fracturing has dramatically improved estimated ultimate recoveries (EURs) of wells in some areas. These larger hydraulic fracture treatments doubled the water volumes of the previous fracs, but left the sand volume the same, change the sand concentrations from one pound per gallon (ppg) to 0.5 ppg. In some areas, the larger hydraulic fracture treatments have doubled EURs relative to previous wells in the same area. Another improvement of EURs has resulted from unconventional pay picking, which has added significant resources that were not previously developed.
Anisotropic drainage and bottom-hole location placement
Maximum horizontal stress can be determined by drilling-induced fracture and borehole-breakout orientation interpreted from borehole image logs. Natural fractures are interpreted from image logs and natural fracture orientation in the Piceance Basin is commonly similar to the maximum horizontal stress orientation (Cumella and Ostby, 2003). The similar orientation of natural fractures and maximum horizontal stress results in highly anisotropic drainage. Microseismic mapping of hydraulic fracture stimulation of Williams Fork sandstones documents this anisotropic drainage pattern (Weijers et al., 2009). Hydraulically stimulated rock volume is probably a good approximation of drainage area.
Figure 10 shows the microseisms observed in five fracture-stimulated wells from two observation wells. The hydraulically stimulated area is highly elliptical, approximating a drainage ellipse of about 6 to 1. In Figure 10, the 10-acre-density bottom-hole locations are located 330 ft apart in a north-south direction and 1320 ft apart in an east-west direction. Directional well paths of the hydraulically fractured wells are shown in green. Microseisms have been observed from two wells whose directional paths are shown in white. The observation well on the east side shows the unstimulated area between the two hydraulically fractured wells on the east side. This area would likely be not drained at less than 10-acre well density.
Unconventional Williams Fork pay picking
Borehole image logs can be used to identify gas seeps (Koepsell et al., 2003). Wells in the Piceance Basin are commonly drilled with mud weights that are near balance with formation pressures, allowing gas to seep into the wellbore. Gas seeps can be identified on borehole image logs as nonconductive areas that start at the seep location and continue upward for a few feet (Fig. 11). Seeps have been identified on all image logs that have been obtained by Bill Barrett Corporation and have been ranked based on quality of gas seep according to criteria established by Randy Koepsell (personal communication, 2009).
Figure 12 shows the ranked gas seeps from an interval in the lower part of the Williams Fork identified from image logs along with gas production from each perforated zone as determined from production logging. Several intervals produce gas from low porosity intervals. For example, the interval at 7025 ft has gas production at a rate of 146 MCFD, but the density porosity is only 5%. The interval at 7197 ft has gas production at a rate of 54 MCFD, but the density porosity is only 2%. Low porosity sandstones in the coal-bearing portion of the lower Williams Fork are commonly cemented with carbonate cements and are highly fracture prone. Cores and image logs show that these tightly cemented sandstones are commonly highly fractured.
Completions of Williams Fork sandstones in the Piceance Basin are commonly based on porosity cutoffs, typically 6% or 8% porosity. Image logs show that gas seeps are common in sandstones that have low porosity and would generally not be completed. Numerous low porosity intervals have been completed in many Bill Barrett Corporation wells that have had production logs run, and most of these tight zones have at least some gas contribution. Completion of these low porosity intervals has added significant reserves that were not previously being developed.
Despite being at the forefront of tight-gas development technology for several decades, recent dramatic improvements in drilling and completion technology have significantly increased reserves and improved development economics. Bottom-hole location placement has been optimized by using natural fracture and stress orientation data from image logs as well as lateral and vertical extent of hydraulic stimulation treatments from microseismic monitoring. Directional drilling technology enables greater than 20 bottom-hole locations to be reached from a single surface location and fit-for-purpose skidable rigs have reduced drill times to as little as three days per well. Hydraulic fracture stimulation improvements have increased well performance, particularly large water volume fracs, which have doubled water volumes but kept sand volumes similar to older fracture treatments. Low-porosity, unconventional pay has been completed in recent wells and production logs indicate that significant reserves are available in pay that has not typically been completed in the past.
Current activity has diminished due to the natural gas price decline of 2009, but the Piceance gas resource is continuous over a large portion of the deep part of the basin and will likely be developed as economics improve.
I started working the Piceance Basin in 2000 when I started working at Barrett Resources. My predecessors at Barrett Resources deserve most of the credit for making the tight gas production from the Williams Fork an economically viable enterprise. Key people include Bill Barrett, Kurt Reinecke, Mark Rothenberg, Robert Mueller, Ted Brown, and Terry Barrett. Ideas presented in this paper have benefited from numerous discussions with many people too numerous to all name here. A few of these many people are Jay Scheevel, Taryn Frenzel, Paul Devine, Paul Kovach, Keith Shanley, and Jack Wiener.