Unconventional completion design: Wolfcamp shale in the Permian

The play and the geology

• Setting: the Permian Basin holds two deep sub-basins, the Delaware Basin to the west and the Midland Basin to the east, separated by the Central Basin Platform, a basement-cored high (Lund Snee and Zoback 2018).
• The rock: the informally named Wolfcamp shale, of Pennsylvanian to early Permian age. In the deeper Midland Basin it is interbedded organic-rich siliciclastic mud, organic-poor clay-rich mud and fine-grained carbonate, while shallow-water carbonates grew on the platforms around it; it is mature for oil across the region (Gaswirth and others 2016).
• Benches: industry divides it, from the top, into Wolfcamp A, B, C and D, and the USGS assessed them as separate units. Where Wolfcamp B is thicker than 500 ft (about 150 m), in the southern Midland Basin, it has room for two laterals (Gaswirth and others 2016).
• Development: horizontal wells, hydraulically fractured in stages. More than 3000 had been drilled and completed in the Midland Basin Wolfcamp by 2016 (Gaswirth and others 2016).
• Stress: in the Midland Basin and on the Central Basin Platform the faulting regime is normal to strike-slip and the greatest horizontal stress $S_{Hmax}$ runs about east–west; in the Delaware Basin the regime is normal faulting and $S_{Hmax}$ turns clockwise by about 150°, from about north–south in western Eddy County, in the north, to about N155°E in western Pecos County, in the south (Lund Snee and Zoback 2018). A hydraulic fracture opens against the least principal stress, here horizontal, so it is a vertical plane along $S_{Hmax}$.
• Anisotropy: the Wolfcamp is elastically anisotropic, chiefly transversely isotropic about a vertical axis from its fine layering and the alignment of its clay and organic matter with the bedding; the steeply dipping natural fractures seen in its laterals add to the anisotropy only if their normal compliance is less than about half their shear compliance (Sayers and Dasgupta 2024).

Reading Figure 9.4

The figure is a model built from these published facts and from the book’s own kernels, not a Wolfcamp data set. Its column of nine beds uses the three lithofacies the USGS describes, played by the Part 5 rock library’s hard shale, medium shale and limestone, at the stresses of Figure 8.1.

1. Start at the default landing, 2945 m in the upper siliceous mudstone, with 3 MPa of net pressure. In (a), Rickman’s index ranks the carbonates highest (72), the siliceous mudstone in the middle (48) and the clay-rich mudstone lowest (20). In (b) the same carbonates carry the highest least horizontal stress: stiff beds take most of the tectonic strain. The fracture grows from 2900 to 2967 m, 67 m, up through the clay-rich mudstone, whose stress is only about 1.4 MPa above the landing bed’s, and stops in the carbonates above and below.
2. Land in the lower carbonate at 3036 m, the most brittle-looking bed. Its stress is the highest in the column, so the pressure that opens it exceeds the stress in every bed around it, and the fracture leaves the column. A brittleness index ranks $E$ and $\nu$; it does not decide where a fracture stops.
3. Back at 2945 m, read the dashed curves: the column as an inversion at 30 Hz would see it, with $V_P$, $V_S$ and $\rho$ averaged over a quarter wavelength, 32 m. Six of the nine beds are thinner than that. The 7 m carbonate under the landing blurs into the mudstones, and the same model predicts a fracture of 156 m instead of 67. At 60 Hz the window shrinks to 16 m, but the carbonate is still thinner and the prediction is 141 m.
4. Raise the net pressure from 3 to 4 MPa. The fracture grows from 67 to 156 m: the 7 m carbonate below the landing gives way between 3.5 and 3.6 MPa, under half its contrast of about 8 MPa, because a thin barrier holds far less than its contrast. At 4.5 MPa the fracture leaves the column.
5. In (c), a north–south lateral in the Midland Basin is normal to $S_{Hmax}$, and each stage opens its own plane, 60 m apart. Turn it east–west and the planes run along the wellbore. Move the pad to the southern Delaware Basin, where $S_{Hmax}$ is about N155°E: the same lateral is 25° from it and its planes stand 25 m apart.

Brittleness, stress and fracture height

Brittleness indices compress rock mechanics into one number. Rickman et al. (2008) rescale Young’s modulus and Poisson’s ratio between bounds chosen for the Barnett shale and average the two, with $E$ in GPa:

Mineralogical indices use the quartz fraction over quartz, carbonate and clay (Jarvie et al. 2007). Both are correlations. As Figure 8.1 showed, the elastic index reads the pore fluid as well as the frame, its bounds are the Barnett’s, and it has no physical threshold between brittle and ductile.

What contains a hydraulic fracture vertically is chiefly the contrast in least horizontal stress between beds. A fracture opens when its fluid pressure exceeds $S_{hmin}$, grows through beds whose stress is below that pressure, and stops in beds whose stress is above it. In the uniaxial-strain model of Section 8.1,

the stiffest beds take the largest share of the tectonic strain, so the carbonates an index ranks highest are often the most stressed beds in a column: barriers, not targets. Simonson, Abou-Sayed and Clifton (1978) gave the height of a fracture in a bed bounded by more stressed layers; between thick barriers it grows without bound as the net pressure approaches the stress contrast, and a thin barrier gives way far sooner. Figure 9.4 solves the same balance of stress intensity at both tips for the whole layered column.

The practical rule: land in a bed whose stress is low beside the beds around it and whose organic richness makes it worth draining, and pump so that the net pressure stays well below the barriers’ contrast: in Figure 9.4 the 7 m carbonate, about 8 MPa more stressed than the landing, gives way at about 3.6 MPa. The brittleness index is one input, rarely the deciding one.

What seismic contributes, and where it stops

• Elastic properties. Pre-stack inversion (Section 7.3) gives $V_P$, $V_S$ and $\rho$, and so $E$, $\nu$ and a brittleness index at every voxel. In organic-rich mudstone, acoustic impedance falls as organic content rises, which lets inversion map richness where wells calibrate it (Løseth et al. 2011).
• Resolution. An inversion averages over about a quarter wavelength, 32 m at 30 Hz in Figure 9.4’s column. Benches tens of metres thick are resolved; the thin carbonate and clay-rich beds that often contain fractures are not.
• Stress (Section 8.1). A stress model needs static moduli, the pore pressure, Biot’s coefficient and tectonic strains, and seismic measures none of them directly. Wells supply them: density logs for $S_v$, injection tests for $S_{hmin}$, breakouts and drilling-induced fractures for the direction of $S_{Hmax}$, the kind of data Lund Snee and Zoback (2018) compiled into their map.
• Azimuthal anisotropy (Section 8.3). Azimuthal AVO (Rüger 2002) and azimuthal velocity respond to aligned compliant fractures and to unequal horizontal stresses. The fit returns an axis pair, and which axis is the fracture strike depends on the sign of the anisotropic gradient, which changes with the fracture fill (Figure 8.3). The HFTS core holds calcite-sealed natural fractures (Gale, Elliott and Laubach 2018), and Sayers and Dasgupta (2024) showed from logs in Wolfcamp laterals that steeply dipping fractures raise the anisotropy only for some ratios of their normal to shear compliance; whether such sealed sets show at seismic scale depends on how compliant they still are, and in a layered shale the azimuthal signal is small beside the vertical anisotropy.
• Microseismic. Geophones in a nearby well or at the surface locate the small earthquakes a treatment triggers. They map where rock slipped, broadly the stimulated volume, with a location error set by the array and the velocity model; they do not show where proppant sits or which fractures will stay open. At the HFTS, intervals of intense microseismicity matched cored hydraulic fractures (Ciezobka, Courtier and Wicker 2018).
• Production closes the loop: tracers, pressure gauges and production logs show which stages and benches contribute. At the HFTS, tracers showed limited communication between the Upper and Middle Wolfcamp wells (Ciezobka, Courtier and Wicker 2018).

What the Hydraulic Fracturing Test Site showed

The Hydraulic Fracturing Test Site (HFTS), in Reagan County in the eastern Midland Basin, was a field experiment of about 25 million dollars around eleven horizontal wells, six in the Upper and five in the Middle Wolfcamp. Their laterals, about 10,000 ft (about 3 km) long, were drilled north to south, roughly normal to the predicted $S_{Hmax}$; more than 400 stages were pumped, most of them traced or monitored (Ciezobka, Courtier and Wicker 2018). After the stimulation a well was drilled at 82° between two producers and cut about 600 ft of whole core through the stimulated rock.

The core recovered hundreds of hydraulic and natural fractures (Ciezobka, Courtier and Wicker 2018). Gale, Elliott and Laubach (2018) described them, two sets of calcite-sealed natural fractures among them. Elliott and Gale (2018) found that the proppant recovered from the fractures was mostly the finest part of the sand pumped, not crushed grains, and that reactivated natural fractures sometimes held more of it than nearby hydraulic fractures.

The lesson for interpretation: a stimulated volume is a dense, complex set of fractures that no seismic or microseismic image resolves one by one. Seismic ranks benches, maps elastic properties between wells and helps orient the wells; the completion is set and checked with logs, core, pressure and production data.

Where the workflow carries over

• Other shale plays: the same chain, inversion to moduli to a calibrated stress model to a landing and completion design, applies in other organic-rich mudstones, with the rock-property transforms and the stress calibrated locally.
• Geothermal and storage: enhanced geothermal systems are stimulated by the same hydraulic fracturing, and the same stress contrasts decide how far fractures grow; injection for storage (Section 9.6) asks the opposite question, whether the seal’s stress keeps fractures out of it.
• Where it does not apply: conventional reservoirs with natural permeability, where fluid prediction (Section 9.1) and pressure support (Section 9.2) matter more than completion design.