Hydraulic Fracturing of the Haynesville Shale

Representative undepleted gas-reservoir stage, East Texas and northwestern Louisiana

FASTFRAC / TECHNICAL NOTE / 20 OCTOBER 2026

This note defines a reproducible single-stage screening example for FastFrac. It combines published Haynesville context with explicitly selected engineering assumptions. It is a synthetic design, not a reconstruction of an operator treatment or a calibrated field prediction. The companion file is Haynesville Shale.json.

1. Geological and operating basis

The Haynesville is a deep, overpressured Upper Jurassic gas reservoir. In Louisiana, the Bossier overlies the Haynesville, with Cotton Valley sandstone farther above, and Smackover limestone below. Formation names alone do not establish hydraulic containment; a local stress profile is required. Chesapeake/ResFrac, Jongeling et al. (2024)

Containment: 600 ft is an imposed PKN model height. It is not claimed as the universal or measured Haynesville fracture height. A published Haynesville microseismic example shows approximately 600 ft of height growth. Microseismic extent, hydraulic aperture and proppant placement describe different properties. Warpinski (2011), Fig. 16

2. Reservoir and completion inputs

The tables identify saved model inputs and derived or contextual values. "Selected" means an assumption for this example; a supporting regional range does not make it a field measurement. Stage length and pitch provide context and are not solver inputs.

Eight clusters are consistent with published Brammer examples of 4-8 clusters. That 2017 source provides historical context, not proof of a current play-wide standard. Brammer Engineering (2017) A uniform 18,000-ft bore is assumed for hydraulics; actual tapered strings and trajectory require their own dimensions.

3. Reservoir fluid properties

The FastFrac toolkit was used because leakoff requires reservoir-fluid viscosity and total compressibility. Calculations use the gas system at 9,974.7 psia (9,960 psig plus 14.7 psi atmospheric pressure) and 320 F. Reservoir gas viscosity is separate from the treating-fluid viscosity.

Calculation and applicability

ct = cf + Sgcg + Swcw In units of 10^-6/psi: 4.000 + 0.75(51.045) + 0.25(3.403) = 43.135.

The toolkit uses Dranchuk-Purvis-Robinson for gas Z and compressibility and Lee-Gonzalez-Eakin for gas viscosity. These results are correlation estimates from an assumed gas gravity. The original gas-viscosity measurements discussed by Ling extend to approximately 8,000 psia; this example is above that pressure. Ling (2010)

The water-property and gas-solubility calculations are outside the toolkit's stated temperature and pressure ranges; gas solubility is also outside its salinity range. Their output is retained as a transparent screening estimate, not validated high-pressure brine PVT. The water term contributes about 0.85 x 10^-6/psi, approximately 2% of the total. The remaining inputs, especially gas composition and rock compressibility, also require local confirmation.

Changing only water compressibility from 2 to 5 x 10^-6/psi gives ct = 42.78-43.53 x 10^-6/psi. The wider 35-55 sensitivity in Section 6 evaluates the model response to uncertainty in the combined total. Toolkit inputs are saved in the JSON so the calculation can be reproduced.

4. Treating fluid and pump schedule

PPA means pounds of proppant added per gallon of clean fluid. All JSON schedule volumes are slurry barrels, and the constant 85-bpm rate is a slurry rate. The clean-fluid rate decreases as concentration rises. Stage labels do not change mesh size; 40/70 is the single global mesh selection.

Material balance and displacement

For clean volume Vc in barrels, concentration c in PPA and sand specific gravity SGp: Sand mass = 42 Vc c Slurry volume = Vc [1 + c / (8.34 SGp)]

The treatment before displacement contains 10,000 bbl clean fluid and 525,000 lb sand, equivalent to 10,565.59 bbl slurry. Over 150 ft, these are 66.7 clean bbl/ft and 3,500 lb/ft. These totals apply to all eight clusters together.

Displacement = ID^2 x MD / 1029.4 = 4.670^2 x 18,000 / 1029.4 = 381.35 bbl. The JSON retains additional precision. The bore starts full of clean fluid, and the final flush equals one modeled wellbore volume. The engine therefore places the final sand at the perforations by the end of displacement. Initial wellbore inventory and arrival lag are handled by the engine; the flush is not an extra sand-bearing treatment step.

Other saved inputs: drainage area 160 acres, hole diameter 8.75 in. and skin 0 are placeholders for the software's productivity section, which does not return a radial-flow improvement factor for this multi-cluster case. Inactive tubing fields are depth 0 ft, ID 1.995 in. and OD 2.375 in.; they do not affect the selected down-casing path.

5. Calculated FastFrac response

Results below were calculated from the delivered JSON using the local FastFrac engine stamped 2026-10-20, inspected and executed on 20 October 2026. They are model outputs, not measured Haynesville performance. Geometry values describe each of eight equal fractures; pumping totals describe the complete stage.

Pressure interpretation

The engine uses surface pressure = fracture-pressure baseline + pipe friction + perforation friction - hydrostatic head. It tracks fluid and sand concentrations moving through the wellbore. In this implementation, the 11,400-psi fracture-pressure baseline stays fixed; a changing net pressure from the geometry solution is not added separately.

The global peak of 13,576 psi occurs during displacement, while the displacement row ends at 12,962 psi. Row-end values and the event-based maximum describe different instants. The peak is approximately 92.6% of the engine's nominal casing-body burst estimate; that comparison is not an allowable working-pressure calculation.

Connection ratings, temperature derating, actual wall condition, combined loads, annular pressure and surface equipment are not established by this example. A 10,000-psi pressure class would be insufficient for the calculated surface pressure. The hydraulic horsepower is fluid power, not an installed horsepower requirement.

6. Sensitivity and interpretation limits

Each case below changes one input while retaining the same schedule, rate, permeability, pay thickness and other base-case settings. The tested values are scenario bounds, not statistical confidence intervals or measured regional limits.

Height materially changes predicted length. Friction reduction materially changes pressure: reducing the assumed drag reduction to 60% raises the calculated peak above 16,000 psi. The constant fracture-pressure baseline makes surface pressure insensitive to the height and compressibility changes shown here. That is a model property, not evidence that field pressure is independent of fracture growth.

Placement diagnostics

The minimum reported created-width / largest-grain ratio during sand steps is 3.46, above the model's three-grain screening threshold. Final pack fill is approximately 9.2%, and the parcel tracker does not flag volume overfill. These checks do not prove uniform sand distribution or sustained conductivity.

The pad heuristic classifies this schedule as over-padded: the input pad is 14.2% of pre-displacement slurry volume versus its estimated 6.2% guide. The single-grain settling calculation for the first sand parcel gives travel of about 0.61 times the assumed fracture height over its modeled residence time. Both results are retained rather than adjusting the inputs solely to eliminate diagnostics.

What the model does and does not establish

The complexity adjustment is a user-assumed sensitivity: effective fractures = primary fractures x (1 + adjustment / 100). This example uses 0%; it does not assume additional fracture complexity. Perforation count remains 32. The adjustment is not a calibrated branching model.

The case assumes identical, equally fed, planar PKN fractures with a prescribed height. Cluster interaction, stress shadowing, depletion, fault activation, a layered stress profile and evolving perforation erosion are not resolved. The stage length is used only to report treatment intensity; cluster spacing does not alter the computed geometry.

Treating-fluid viscosity and friction reduction are fixed effective inputs. A single 40/70 sand and simplified Newtonian settling model cannot represent a field treatment that changes mesh or rheology. The average propped width is a volume-equivalent quantity and is smaller than the model's median grain diameter; it should not be interpreted as a continuous uniform multilayer pack.

Conductivity is a screening estimate with a selected 0.50 multiplier. The software does not reproduce product-specific crushing, embedment or long-term conductivity loss under depletion. Its effective propped length is not a measured drainage length, and the multi-cluster case does not provide a validated production forecast.

7. References and reproduction

Burke, L. A. (2021). Quantitative Modeling of Secondary Migration: Understanding the Origin of Natural Gas Charge of the Haynesville Formation in the Sabine Uplift Area of Louisiana and Texas. GCAGS Journal, 10, 24-30. Table 1, p. 26, compiles reservoir properties after Wang et al. (2013). Source.

Jongeling, C., et al. (2024). A Physics-Based Approach to Characterize Productivity Loss in the Haynesville Shale. URTeC 4033310; DOI 10.15530/urtec-2024-4033310. Geological context p. 2; study population and treatment intensities p. 8; stress calibration pp. 10-11. Source.

American Oil & Gas Reporter (2024). Haynesville Operators Leverage Advantaged Position As LNG Exports Grow. Aethon operator interview provides the cited proppant loading and pump-rate benchmarks. Source.

Siriwardane, H.; Gondle, R.; Bromhal, G. (2016). Extent of Hydraulic Fractures in Shales. NETL-TRS-16-2016, 9 December. Haynesville modeling inputs: Table 9, printed p. 31. These are model assumptions, not a microseismic field dataset. Source.

Warpinski, N. (2011). Measurement and Observations of Fracture Height Growth. EPA-hosted workshop material. Figure 16 and accompanying discussion reproduce a Haynesville microseismic example attributed to Pope et al. (2009), SPE 125079. Source.

Brammer Engineering (2017). Progression of Fracing and the Haynesville Shale. Company technical newsletter; historical cluster-count and fluid-system context. Source.

Ling, K. (2010). Gas Viscosity at High Pressure and High Temperature. Texas A&M University dissertation. Reviews the experimental coverage of gas-viscosity correlations, including the Lee/Gonzalez/Eakin measurements. Source.

Software and verification record

Calculation source: the local FastFrac engine and toolkit in the project's site folder, engine version 2026-10-20. The actual import parser accepted the JSON; the engine returned a successful result with no general warning strings. Separate pad and settling diagnostics are reported in Section 6. Independent checks reconciled fluid volume, sand mass and displacement.

Source fingerprints (first 16 SHA-256 characters): engine 0dca2f99eb781cda; toolkit 135a8ff31f5ea251. These identify the local calculation code; a differently deployed engine may return different results.

Load the example

In FastFrac, open Select design, choose the option to load a design from your computer, and select Haynesville Shale.json. Confirm the well title and inputs, then select Compute. The saved toolkit settings can be restored when opening the Toolbox. Stage volumes are stored with more precision than the tables display.

This note and its companion JSON define an illustrative undepleted-reservoir case. Conversion to an operating treatment requires the actual well survey, stress and pressure measurements, fluid tests, tubular specifications and proppant-conductivity data.