“Besides black art, there is only automation and mechanization.” Federico Garcia Lorca, Spanish poet and playwright (1898–1936)

On this basis, fracture stimulation sits firmly in the realm of mysticism, crystal balls, and tarot.

The reason for this is that unlike automation (e.g., car manufacturing) we still are uncertain of what makes fracture stimulation “work” (i.e., the critical components) and of how we evaluate and improve performance (i.e., of what constitutes a “vehicle service”).

The car industry understands the components that make a car operate because each is built meticulously and repetitively. The key components that drive fracture stimulation are more difficult to determine because the process is subject to a high degree of variability in subsurface conditions. And what adds more “smoke and mirrors” to stimulation design is that much of the research that forms our basic understanding was developed more than 30 years ago on tight gas reservoirs, which bear no resemblance to the shale, coal, or unconsolidated sands stimulated today. In other words, while we have paid for a Rolls Royce we unwittingly may be driving a DeLorean.

The automotive industry assesses the performance of its cars through periodic, thorough checkups. How often do we conduct a multicomponent service on our stimulation treatments? Microseismics, continuous reservoir monitoring (by use of distributed-temperature surveys and permanent downhole gauges), and radioactive-fluid tracers are some of the technologies used to determine the “health” of stimula-tion treatments. Their use, however, remains the exception not the rule. Usually, expensive stimulation-treatment checkups are avoided; instead, only the oil dipstick continues to be used (i.e., pressure and rate during injection and production).

Is it not time that you better understood the key components “under the hood” and gave a complete service to your fracture-stimulation process? Only through better understanding will the “black art” be replaced by science, the performance be mea-sured, and step changes truly be made.

Hydraulic Fracturing additional reading available at the SPE eLibrary: www.spe.org

SPE 116124 • “Case History of Sequential and Simultaneous Fracturing of the Barnett Shale In Parker County” by P.N. Mutalik, Williams Companies, et al.

SPE 118831 • “Optical-Fiber Distributed Temperature for Fracture-Stimulation Diagnostics and Well-Performance Evaluation” by Paul Huckabee, SPE, Shell E&P

SPE 119350 • “Stress Amplification and Arch Dimensions in Proppant Beds Deposited by Waterfracs” by N.R. Warpinski, SPE, Pinnacle

SPE 111431 • “New Viscoelastic Surfactant Fracturing Fluids Now Compatible With CO2 Drastically Improve Gas Production in Rockies” by K. Hughes, Chevron, et al.

Hydraulic Fracturing

TECHNOLOGY FOCUS

50 JPT • MARCH 2009

JPT

Simon Chipperfield, SPE, is Team Leader—Central Gas Exploitation at Santos. During the last 14 years, he has held positions in petroleum engineering (drilling, completions, and stimulation) and reservoir engineering. Chipperfield previously worked for Shell International E&P. He was awarded the 2007 SPE Cedric K. Ferguson Medal. Chipperfield has authored more than 18 technical publications in the areas of hydraulic fracturing, reservoir engineering, com-pletion technology, and sand control. He holds a petroleum engineering degree (Honors) from the University of New South Wales. Chipperfield serves on the JPT Editorial Committee and has served as a reviewer for SPEPO.

JPT • MARCH 2009 51

Hydraulic fracturing is the most popular and successful stimulation treatment in the petroleum industry. Fracturing technology has opened numerous unconventional hydrocarbon frontiers that, were it not for this process, would be uneconomic to develop. However, designing the optimal hydraulic frac-ture is anything but an exact science. The industry relies on many simplify-ing assumptions that assist in making expedient decisions but can prevent reaching a truly optimized treatment. The full-length paper outlines the approach taken by one Canyon sand operator to apply sound science and best practices to meet their financial.

IntroductionA field trial comparing the performance of 20/40 economy-lightweight-ceram-ic (ELWC) proppant to 20/40 Brady sand was completed in 2006 and 2007 in the Canyon sand, Sutton County, Texas. The trial was initiated after modeling suggested that the current hydraulic fractures were conductiv-ity limited. Twenty-one wells in close proximity to each other were included in the study. Production analyses were

performed after all wells had at least 1 year of post-fracture production. Both raw production and production normalized to reservoir quality were analyzed, and both indicate that the wells stimulated with ELWC proppant significantly outperformed the offset wells that used Brady sand. Several economic yardsticks then were evalu-ated in the full-length paper.

Canyon SandThe Canyon sandstone is produc-tive in several fields, including the Sonora, Sawyers, Shurley, and Aldwell fields, with first production occur-ring in 1952. The Canyon-sandstone trend occurs basinward of the western margin of the Eastern shelf in west Texas. The sandstones were deposited

from the Eastern shelf into the Val Verde basin. The reservoir is oil prone adjacent to the Midland basin and gas prone adjacent to the Val Verde basin. The vast majority of the sandstone members are fine- to medium-grained quartz lithic arenites.

Because the field has been on pro-duction for more than 50 years, and the operator continues to downspace (currently drilling on 20- to 40-acre spacing), the reservoir pressure in the trend continues to decline. The current operator has acquired a majority of the acreage in the Sonora field through a series of acquisitions since 1993, and more than 5,000 wells have been drilled during this time. Daily produc-tion has increased from 25 MMcf/D to 240 MMcf/D since 1993. This opera-

This article, written by Assistant Tech-nology Editor Karen Bybee, contains high lights of paper SPE 117538, “Apply-ing Science and Best Practices To increase Production and Optimize Economics in a West Texas Gas Field—A Canyon-Sand Case Study,” by Kelly Blackwood, SPE, (now with Encana Oil and Gas), and Kaylene Williamson, SPE, Highmount Energy; Terry Palisch, SPE, and Mark Chapman, SPE, Carbo Ceramics; and Mike Vincent, SPE, Insight Consulting, originally prepared for the 2008 SPE Eastern Regional/AAPG Eastern Section Joint Meeting, Pittsburgh, Pennsylvania, 11–15 Octo ber. The paper has not been peer reviewed.

Applying Science and Best Practices To Increase Production and Optimize Economics in a West Texas Gas Field

HYDRAULIC FRACTURING

For a limited time, the full-length paper is available free to SPE members at www.spe.org/jpt.

Fig. 1—Average per-well incremental gas production for first 90 days post-fracture.

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 900%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Ceramic minus Sand—RateCeramic vs. Sand—% Increase

Days Since Fracture

Aver

age

Per-W

ell I

ncre

men

tal P

rodu

ctio

n, M

cf/D

Incremental Production vs. Time Since Fractured

52 JPT • MARCH 2009

tor currently has 13 drilling rigs and 11 completion/workover rigs active in the field. Approximately 150 hydrau-lic-fracture treatments (30 wells) are conducted each month, using total proppant of approximately 12 million lbm. This high level of activity has provided an excellent opportunity for a field trial in which a large number of candidates with similar reservoir quality and pressure were available for evaluation.

Canyon-Sand CompletionsA typical well in this study is drilled vertically to approximately 6,500 ft true vertical depth (TVD). Fracture stimulations are performed rigless by means of 4.5-in. casing, and flow-through fracture valves/baffles are used to isolate between stages. All wells are stimulated at approximately 40 bbl/min with a low-pH crosslinked-fluid (25 to 35 lbm/1,000 gal) system and 40-quality CO2. An old rule of thumb used in the field is to pump 20/40 Brady sand when treating pay that is shallower than 7,000 ft and 20/40 Ottawa (White) sand when deeper than 7,000 ft. A typical well receives a total of 400,000 lbm of proppant in 4 to 6 stages at maximum slurry concentrations of 5 to 6 lbm of prop-pant added to 1 gal of fluid. Individual fractures are designed for 500 to 600 ft of half-length and 0.4 lbm/ft2 prop-pant concentration.

Rationale for Field TrialAlthough hydraulic fracturing is critical to the development of most unconven-tional reservoirs, many shortcuts and assumptions commonly made during the design of the stimulation treatment limit the productivity and profitability of the field. For example, it is not uncom-mon to use “photocopy engineering”—simply copying a treatment schedule used on a previous well or from another reservoir. In addition, many simplifying assumptions continue to be used that, while expediting the design phase, can prevent an engineer from achieving a truly optimal treatment.

One of these assumptions is often stated: “Conductivity is not important in tight gas/low-rate formations” or “Premium proppants are not need-ed in shallow or low-pressure, low-rate wells.” However, the belief that conductivity is unimportant in tight gas/low-rate/low-stress applications

ignores basic physics. Many authors have shown that conductivity is extremely important, even in applica-tions that previously were assumed to exhibit infinite conductivity.

Non-Darcy- and Multiphase-Flow Effects. The standard labora-tory conductivity test [International Organization for Standardization (ISO) test] is performed using 2% potas-sium chloride at rates equivalent to ½ teaspoon per minute. Unfortunately, these conditions ignore the effects of non-Darcy and multiphase flow. Most fracture engineers consider adjust-ments for these effects to be necessary in high-rate wells at normal depths, temperatures, and pressures. However, engineers also must recognize that although these Canyon-sand wells are low rate (100 to 400 Mcf/D), the low reservoir pressure (1,200 to 1,500 psi) and low temperature (165°F) create very high fluid velocities within the fractures as a result of gas expansion. In fact, 400 Mcf/D when evaluated at the expected bottomhole conditions would yield the same volumetric flow rate as a more prolific well producing 1.2 MMcf/D from a 4,500-psi, 200°F reservoir. In addition, because these Canyon-sand wells typically produce 1 to 2 B/D of condensate and 20 to 50 BWPD, multiphase-flow effects are significant. Laboratory measurements and modeling suggest that the effect of non-Darcy and multiphase flow in the proppant pack would reduce the effec-tive conductivity of the proppant pack for the typical Canyon-sand comple-tion by more than approximately 75%, with further reductions after consider-ing gel damage, stress cycling, and other realistic conditions.

Reduced Proppant Concentration. The standard ISO test is performed using a proppant concentration of 2 lbm/ft2. However, fracture model-ing suggests that the actual prop-pant concentration achieved in these Canyon-sand fractures is closer to approximately 0.4 lbm/ft2. This lower proppant concentration presents sev-eral conductivity challenges. First, it reduces the fracture width and lami-nar conductivity of the proppant pack by 80% (from 2 lbm/ft2 to 0.4 lbm/ft2). However, this five-fold reduction in width also requires a 500% increase in fluid velocity. Because non-Darcy

effects are related to the square of velocity, a 500% increase in velocity will increase the non-Darcy pressure losses by 2,500%. Narrow fractures also are more significantly damaged by proppant embedment, stress cycling, and filter-cake damage. Therefore, these narrow fractures provide only a small fraction of the conductivity advertised by suppliers.

Gel Damage/Cleanup. Generally, it is accepted that when gel is pumped in the pad and/or slurry stages of a fracture treatment, some degree of gel damage is unavoidable. However, stud-ies have shown that the type and size of the proppant will affect the magnitude of the gel damage. This damage can be caused by the residual gel in the center of the pack or by filter cake deposited on the fracture faces and then extruded into the proppant pack at closure. It has been shown that smaller proppants and more-angular proppants will not clean up as thoroughly as larger, more spherical, uniformly sized proppants. Typical 20/40 Brady sand has a mean particle diameter (MPD) of 600 µm and is relatively angular. 20/40 White sand, while similar in MPD, is more rounded than Brady. Premium resin-coated sands typically are more spheri-cal than their uncoated counterparts, with slightly larger MPDs. A 20/40 ELWC proppant typically has an MPD of 650 µm and is very spherical and more uniform in size.

Also adding to the concern of gel cleanup in the Canyon sand is the low reservoir pressure. Fracturing gels gen-erally are non-Newtonian and exhibit a yield point, or threshold pressure to initiate flow. With a limited supply of “energy” in these Canyon sands, the authors were concerned that por-tions of the fracture farther away from the wellbore may not clean up under normal conditions with low-conduc-tivity fractures.

Model Results. While there are sev-eral approaches to increase the con-ductivity of a hydraulic fracture, often the simplest method to evaluate in the field is upgrading the proppant. Other approaches (e.g., changing fluid sys-tems, using gel breakers, or increasing proppant concentration) may simulta-neously affect the fracture geometry, making it difficult to determine if the resulting gas production should

JPT • MARCH 2009

be attributed to fracture conductivity or to other changes. Modeling was performed to predict the benefit of upgrad-ing from Brady sand to several different proppants—20/40 White sand, 20/40 premium resin-coated sand, and 20/40 ELWC proppant.

While all proppants were considered, and each upgrade was expected to increase production, upgrading to 20/40 ELWC proppant provided the largest effect, an increase in gas production greater than 25% anticipated during the first year alone.

Field-Trial ResultsRaw-Data Analysis. During the first 90 days of production, it was evident that the ELWC-proppant wells produced, on average, significantly more gas than the Brady-sand wells. In fact, the average production (per well) for the ELWC-proppant wells during the first month was 60 to 80 Mcf/D higher than that of Brady-sand wells (Fig. 1), which equates to a 20 to 30% increase in production rate. This benefit sta-bilizes at approximately 10% incremental production after 90 days. After 1 year, the ELWC-proppant wells exceed Brady-sand wells by an average of more than 6 MMcf of incremental cumulative gas per well, or a 12% increase. Operators use a variety of benchmarks to compare well results. One simple comparison is to calculate the average IP30 for each set of wells, where IP30 represents the aver-age rate of the first 30 days of nonzero production. Using this measure, the average IP30 of the ELWC-proppant wells was approximately 320 Mcf/D, which is 25% more than the IP30 of approximately 250 Mcf/D for the Brady-sand wells. When comparing the IP30 for all 21 trial wells, it is clear that the ELWC-proppant wells outperform the Brady-sand wells by a significant margin, despite lower-quality pay and less thickness.

Another measure of well performance that often is used is estimated ultimate recovery (EUR). Typically, the EUR of a gas well is determined by use of decline-curve analysis of actual production data. For these Canyon-sand wells, the average production for each set of trial wells was matched to a hyperbolic decline curve. Interestingly, the decline rate of the two production curves essentially is identical. That the initial production rate is higher for the ELWC-proppant wells, with both groups having the same decline rate, sug-gests that the incremental production is not the result solely of rate acceleration but rather may be indicative of increased drainage volume. In fact, if the decline curves are projected 30 years, the ELWC-proppant wells would produce an incremental 30 MMcf (8%) per well, despite the lower reservoir quality. If the decline curves are projected to an economic limit of 10 Mcf/D, then the ELWC-proppant wells would produce an incremental 125 MMcf (16%) per well, although it takes more than 100 years for this value to be reached. JPT

54 JPT • MARCH 2009

Slickwater fracturing has increased over the past decade, with the advent of shale gas plays. Horizontal wells are now the standard, with as much as 1 million gal of water in as many as six to nine fracture stages per well. The objective is to create as much contact with the reservoir as possible. Additive packages have been minimized to reduce costs. Because of environmen-tal concerns and freshwater availabil-ity, the flowback and produced water are collected and used for subsequent fracture treatments. The full-length paper examines water-treatment tech-niques and evaluates the performance of additives that are used in slickwater fractures of shale reservoirs.

IntroductionThe number of slickwater fractures has increased because of higher nat-ural-gas prices and more experience in fracturing with lower-cost fluids. Slickwater fractures have been used in low-permeability and large net pays, and they require large amounts of water to obtain adequate fracture half-lengths. Before the Barnett shale in north Texas was fractured in 1997, many fractures were carried out with a crosslinked fluid and large amounts of proppants. The difficulty in clean-ing up the wells and the low return made many wells uneconomical.

Some wells were treated with slickwa-ter and no proppant. Initial produc-tion was higher but declined rapidly. Eventually, the state of the art evolved to high-rate slickwater fractures with various additives. The question to be addressed is how the various additives perform in shale and how to deter-mine which additives are necessary, particularly in light of the fact that most fractures are now conducted with produced and/or flowback water from previous fractures.

Water ReuseWater can be reused and makes up the majority of volume in a slickwa-ter fracture. There have been and are problems in obtaining sufficient water from municipalities or disposing of produced water. If there is a way to reuse the water or fluid, this will save money and solve environmental issues with disposal. Disposal costs can run as high as USD11/bbl. With the advent of horizontal wells with multiple staged fractures, there is a process requirement as great as 10,000 B/D with the entire job requiring as much as 100,000 bbl of water. Surface waters

alone can contain clays, sand/silt, iron, sulfates, and bacteria. Produced water and reused fracture waters can have various contaminants.

The presence of the many possible contaminants can affect the perfor-mance of the fracturing-fluid addi-tives. For example, surfactants and clay stabilizers can be adsorbed onto the colloidal solids. The presence of solids, residual gel, and bacteria also can impair formation permeability.

Shale waters typically have exces-sive barium and strontium that should not be precipitated because of natural-ly occurring radioactive material and solids creation. Water can be reused, especially if there are standards put in place that allow the water quality to be analyzed before it is reused. Salt becomes an issue when the content becomes too high. In this case, the salt water is diluted with “clean water,” as needed, to the desired specification.

Recently, governmental bodies in the states of New York and Pennsylvania have begun enacting enforcement measures to regulate the waters used in fracturing. These controls will establish limits for various materi-

This article, written by Assistant Tech-nology Editor Karen Bybee, contains high lights of paper SPE 119900, “Crit ical Evaluations of Additives Used in Shale Slickwater Fracs,” by P. Kaufman, SPE, and G.S. Penny, SPE, CESI Chemical, and J. Paktinat, SPE, Universal Well Services, originally prepared for the 2008 SPE Shale Gas Production Conference, Fort Worth, Texas, 16–18 November. The paper has not been peer reviewed.

Critical Evaluation of Additives Used in Shale Slickwater Fractures

HYDRAULIC FRACTURING

For a limited time, the full-length paper is available free to SPE members at www.spe.org/jpt.

Fig. 1—Comparison of commercially available friction reducers 60 sec-onds after injection into the flow loop.

56 JPT • MARCH 2009

als contained in additives and waters pumped into shale reservoirs.

AdditivesFriction Reducers. With the success of slickwater fractures, especially in the Barnett shale, friction reducers have grown in popularity. Friction reduc-ers are used to decrease the frictional pressure losses, allowing higher pres-sure from the same number of pump trucks. Common friction reducers are polyacrylamide based and have a usual loading range from 0.25 gal of friction reducer per 1,000 gal (gpt) of water to 1 gpt of water. There are three types of polyacrylamide friction reducers: anionic, cationic, and nonionic. They have thermal stability up to approxi-mately 400°F and readily decompose above 550°F. Chemical and thermal degradation of the polymer reduces its effectiveness.

Friction reducers can cause forma-tion damage and may require a break-er. Some breakers are delayed to allow the friction to be reduced in the tubing where it is most effective. Even low concentrations of 0.25 gpt of water results in 250 gal of potential polymer damage. Once past the perforations, the breaker will break the polymer to reduce damage and encourage poly-mer flowback. In a study of oxidative breakers on a freshwater- and brine-based polyacrylamide friction reducer used in slickwater fracture treatments, all breakers tested worked at 180°F to some extent. The results were generat-ed using a technique called molecular-weight cutoff and show that the per-sulfate breakers worked best at 180°F, and they were effective at 100°F at concentration of 5 and 10 lbm per

1,000 gal of water. Flow loop data showed no degradation of polymer at a persulfate concentration of 1 gal per 1,000 gal of water up to 105°F, nor did the breaker have detrimental effects on hydration of the polymer.

Care must be taken when selecting a friction reducer. Fig.1 illustrates the variations in commercially available friction reducers. Six friction reduc-ers were supplied by a major operat-ing company for independent evalu-ation. Flow was 5 gal/min through a 50-ft-long 0.5-in.-outside-diameter, 0.402-inside-diameter tubing. Total volume, including tank, was 5 gal. Pressure ports were 10 ft apart, and the concentration of friction reducer was 0.25 gal/1,000 gal of fresh water with 2 wt% potassium chloride (KCl). Polymers hydrate at different rates. At 20 seconds, the friction reduction between the lowest and highest values is roughly 50%. At 20 seconds the polymer has made two complete pass-es through the loop. At 10 minutes (Fig. 2) (600 seconds), or 60 passes through a progressing cavity pump, the difference between the lowest and highest values is roughly 20%.

Biocides. Biocides are used in slick-water applications to prevent bacteria growth and may alter the physical characteristics of fluids, such as vis-cosity, by degrading the polymer. The degradation of the polymer can be caused by free radicals, usually from oxygen. Therefore, potentially, there is a compatibility issue between oxygen scavenger or biocide. Compatibility of biocide with other chemicals such as corrosion inhibitors, scale inhibitors, polymers, and other well-treatment

fluids is critical. Some attributes nec-essary for a biocide are safety, cost effectiveness, compatibility with fluids or other additives, and ease of han-dling. In the well, bacteria can cause other problems. These include, but are not limited to, producing acid [acid-producing bacteria (APB)], sulfate-reducing bacteria (SRB), and produc-tion of carbon dioxide and oxygen.

Common biocides are quaternary amines, glutaraldehyde, and tetra-kis-hydroxylmethylphosphonium sulfate. A relative newcomer to oilfield appli-cations is tetrahydro 3,5-dimethyl-1,3,5-thiadiazinane-2-thione; this bio-cide has been shown not to interfere with friction reducer, is extremely effective in killing APB, and is a broad-spectrum biocide. It also is a longer-term-kill biocide that is compatible with oxygen scavengers.

Even if a biocide is effective in killing, testing should be conducted to determine compatibility of biocide with the friction reducer. The reuse of fracture water is of special concern because of the potential amount of bacteria in the produced water. Again, these bacteria can attack the polymer and reduce its effectiveness, and a reg-istered biocide is recommended in the fracture tank.

Scale Inhibitors. Calcium sulfate and carbonate and barium sulfate can cause scale problems if the concentra-tion is sufficiently high, the pressure differential is sufficiently high, and the temperature is sufficiently low. As the well is fractured, water dis-solves minerals in the shale, and if conditions are right, scale can form. If produced water is used, then the prob-lem of scale formation is exacerbated. The dissolved salts potentially can cause scale deposition. Many times the flowback fluid is diluted with fresh water to achieve the desired level of salts such as those containing barium and strontium.

Most available scale inhibitors are phosponates and organophospho-nates, which are anionic. This can cre-ate incompatibilities with some addi-tives such as friction reducers and clay stabilizers. Some novel solutions to this problem have been tried in recent history. The organophosponates can be reacted with calcium chloride to

Fig. 2—Comparison of commercially available friction reducers 10 min-utes after injection into the flow loop.

(Contd. on page 91)

JPT • MARCH 2009 91

The inauguration of the SPE Surat Section took place on 17 November 2008. Pictured from left are: Manoj Pandey, SPE Surat Section Secretary; Lal Chand Ram, General Manager, Oil & Natural Gas Corporation (ONGC); Amar Jha, SPE Surat Section Program Chairperson; N.K. Mitra, SPE India Council Chairperson and Director ONGC; Anil K. Johari, SPE Surat Section Chairperson; V.K. Yadav, General Manager, ONGC; and Amit Dave, SPE Surat Section Treasurer.

render them insoluble, to pump from the surface to the formation success-fully. The insoluble material can be made into the proppant matrix.

Proppants. The amount of proppants used in slickwater fractures is relatively small when compared to conventional fractures. Selection of proppants is criti-cal in a slickwater design. With the use of extremely large volumes of water (more than 5 million gal), proppant placement can be an issue. In a study of the placement of proppants in slickwa-ter applications, findings showed that a 20/40-mesh lightweight proppant with a specific gravity (SG) of 1.25 in viscosi-fied 9.4 lbm/gal brine had virtually no proppant settling. For proppant place-ment, there are several factors to con-sider when designing a fracture job: SG of the fluid and proppant, slurry-flow rate, and fluid viscosity.

Sand proppants with an SG of approximately 2.6 should be rather small to work in slickwater fractures.

The smaller the size of proppant, the greater the transport, assuming all other parameters are the same. The higher the SG, the shorter the trans-port, assuming the same size.

It must be considered that there is a tradeoff between strength and SG of the proppant. Sand typically has an SG of approximately 2.6 and ceramics have approximately 2.7. Recently, ultralight-weight proppants have come onto the market; they are a chemically modified walnut shell with resin coating with a measured 1.25 SG. There are limitations to this material, however; it is usable to 5,000-psi closure stress and a tempera-ture of 200°F. A new generation of prop-pant has an SG of 1.05, which is nearly buoyant in slickwater, thus allowing farther penetration into the fracture, and it is being used in the Barnett shale. Again, there are limitations to the use of the product: 7,000-psi maximum clo-sure stress and maximum temperature of 225°F. It is used in a partial-monolay-er application.

Another recent advance in proppants is lightweight 40/80-mesh ceramic that has an SG and mesh designed specifi-cally for slickwater applications. It has a crush strength of 2% fines at 7,500 psi, a 2.55 SG, and a roundness and sphe-ricity of approximately 0.8. This prop-pant exhibits a conductivity twice that of resin-coated and white sand.

Clay Stabilizers. There is always a question of whether or not a clay sta-bilizer is required in the water pumped into shales. An analysis of the shales in the northeast US found that clay was abundant, with the most abundant being illite; the majority of the other designation is quartz. The common method of stabilizing clays has been to add KCl, commonly 2 wt% KCl. Many flow tests and capillary-suction-time evaluations show that 2 wt% KCl has a marginal effect on swelling clays. To get the greatest benefit, tests show 4 wt% KCl often is better, but cost then becomes the issue.

HYDRAULIC FRACTURING (Contd. from page 56)

JPT

58 JPT • MARCH 2009

The full-length paper compares the strengths, weaknesses, and limitations of fracture modeling, production-data analysis (PDA), pressure-transient analysis (PTA), and numerical reservoir modeling in estimating effective frac-ture length and conductivity. The paper also evaluates how the complexities (in the hydraulic fracture) associated with non-Darcy flow, multiphase flow, and complex fracture geometries affect the results from the various techniques. The paper documents the significant differences in “effective” fracture length that, in many cases, can result from each technique.

IntroductionReliable estimates of fracture length (i.e., created, propped, and producing or effective) are necessary to consider design changes in subsequent frac-ture treatments to optimize the perfor-mance of hydraulically fractured wells, particularly in low-permeability reser-voirs. The created fracture length is the fracture length propagated during the fracture treatment, while the propped fracture length is the length supported by proppant after the fracture closes. The effective, or producing, length is

the length that is open or contribut-ing to hydrocarbon production after a fracture treatment. Increasing the effective fracture length usually means increased production.

Incorrect estimates of the effective fracture length can lead to less-than-optimal gas recovery and often contrib-ute to modifications of fracturing designs that may not result in improved well productivity. It has been known that the fracture lengths determined from frac-ture modeling, PDA, PTA, and numeri-cal reservoir modeling are not in agree-ment with the created fracture lengths obtained from fracture mapping.

Techniques for Determining Fracture PropertiesFracture Modeling. Fracture mod-eling (i.e., net-pressure analysis) can

provide information about fracture length, height, width, and conductivity. Fracture dimensions and conductivity can be estimated from fracture model-ing by matching the observed fractur-ing net pressures (fracturing pressure minus minimum rock stress, or clo-sure pressure). The limitations of this technique are that it typically provides nonunique solutions, can be unreliable if not calibrated, and requires baseline rock and stress data.

Net-pressure history matching can be implemented by adding new phys-ics to fracture models. With the right assumptions and physics, inferred fracture geometry can be more reli-able; however, inferred geometry from net-pressure matching does not always agree with directly mea-sured geometry.

This article, written by Assistant Tech-nology Editor Karen Bybee, contains highlights of paper IPTC 12147, “Re solving Created, Propped, and Effective Hydraulic-Fracture Length,” by C.L. Cipolla, SPE, E.P. Lolon, SPE, and M.J. Mayerhofer, SPE, Pinnacle Technologies, originally prepared for the 2008 International Petroleum Technology Conference, Kuala Lumpur, 3–5 December. The paper has not been peer reviewed.

Copyright 2008 International Petroleum Technology Conference. Reproduced by permission.

Resolving Created, Propped, and Effective Hydraulic-Fracture Length

HYDRAULIC FRACTURING

For a limited time, the full-length paper is available free to SPE members at www.spe.org/jpt.

Fig. 1—Net-pressure history match.

Uncalibrated-ModelNet-Pressure Match

Time, minutes

Observed Net (psi)Proppant Conc, Surf

Net Pressure (psi)Slurry Rate

80.0 96.0 112.0 128.0 144.0 160.0 0

600

1200

1800

2400

3000 50.00

0

600

1200

1800

2400

3000 100.0

40.00 80.0

30.00 60.0

20.00 40.0

10.00 20.0

0.00 0.0

JPT • MARCH 2009 59

PDA. PDA can be used to determine reservoir characteristics, completion effectiveness, and hydrocarbons in place. Conventional decline-type-curve analysis of production data is a viable alternative for evaluating well perfor-mance without shutting in the well. Unlike pressure-transient-test analysis, decline type curves do not rely upon identification of characteristic flow regimes for the analysis. As a result, unique estimates of fracture half-length cannot always be obtained, especially when using poor-quality production data. Other production analysis tech-niques consider the production data to be an extended drawdown test. Accordingly, these techniques use variable-rate pressure-transient-test-ing theory and superposition plotting functions to analyze the production data. Unlike conventional decline-type-curve-analysis techniques, these meth-ods allow identification of specific flow regimes. However, fracture properties still cannot be quantified without an estimate of reservoir permeability.

A critical element in the successful application of the decline-type-curve method is related directly to the fre-

quency and quality of the well produc-tion data. Some advantages of this tech-nique are that no shut-ins are required, production data are readily available, and analysis can be performed quickly.

PTA. Pressure-transient testing is an effective technique for evaluating the stimulation effectiveness of hydraulical-ly fractured wells. However, knowledge of reservoir permeability, either from a prefracture well test or from an indepen-dent source, usually is required to com-pute fracture properties, especially in low-permeability reservoirs. If a well is shut in for a sufficient time to reach the pseudoradial-flow period, then reservoir permeability can be determined from a post-fracture pressure buildup (PBU). Unfortunately, wells completed in tight gas sands usually require very long shut-in times to reach pseudoradial flow. The data required to characterize a hydrauli-cally fractured well can require very long buildup duration, and the buildup response can be distorted in early time by wellbore storage and in late time by superposition. If there is an estimate of reservoir permeability from an indepen-dent source or from a prefracture PBU,

then shorter-duration PBU tests in tight gas sands become practical.

PBU analysis is usually the most reliable method for determining the effective length of hydraulic fractures. Unfortunately, discrepancies in the results also are noted when using dif-ferent analysis models (i.e., bilinear, lin-ear, or pseudoradial). The application of well-test data to evaluate fracture perfor-mance can provide an independent esti-mate of fracture length and conductivity. Relative, or dimensionless, fracture con-ductivity (the ratio of fracture conduc-tivity to reservoir permeability) is easily measured using a post-fracture PBU.

Numerical Reservoir Modeling. Reservoir-simulation history matching has proved to be useful for determining effective fracture length. Uniqueness problems common to simulator history matching are minimized when trying to match all test data obtained on the well and the long-term production his-tory of the well. Numerical reservoir modeling is an excellent technique for handling simultaneous effects of frac-ture conductivity, reservoir boundaries, multiphase and non-Darcy-flow effects, and complex reservoir systems.

Case HistoriesTaylor Cotton Valley. The fracturing program in this field included different types of water fracture and linear-gel hybrid fracture treatments in the Taylor Cotton Valley at depths of approximate-ly 11,500 ft. Detailed production analy-sis was performed to evaluate well per-formance in conjunction with fracture-geometry measurements provided by microseismic fracture-mapping results, calibrated fracture modeling, and direct production-interference data, which provided interesting insights into effec-tive fracture lengths.

Microseismic fracture mapping indi-cated that created fractures are very long. There is compelling evidence that effective hydraulic-fracture lengths also are very long because immediate well interference can be detected in wells that are located along the fracture orientation with very large interwell distances. This means that there is a conductive hydraulic-fracture path in place over large distances, with frac-tures overlapping and linking wells. Production modeling indicates that the permeability feeding the long hydraulic fractures (perpendicular to fracture) is,

Fig. 2—Fracture profile.

60 JPT • MARCH 2009

on average, very low. The low perme-ability and long fractures will create elliptical drainage areas that should be taken into consideration for well-placement and -spacing strategies.

The evaluation of hydraulic-fracture treatments must always be performed in conjunction with a reservoir/produc-tion evaluation. The integration of all results from different techniques (in this case, fracture mapping, fracture modeling, and production analysis) is crucial for hydraulic-fracture optimiza-tion and for field-development strate-gies and well placement. Production modeling alone can be nonunique when reservoir permeability is not well defined. The reason for this nonunique behavior is that early-time flow data are partly influenced by cleanup of hydraulic-fracturing fluid and the fact that fracture length, fracture conduc-tivity, and permeability all affect the early-time flow behavior in a similar manner and, given real field data, can-not be distinguished clearly with com-monly available daily flow rates and flowing tubing pressures (FTPs). Using the calibrated-fracture-model fracture length of approximately 1,500 ft, pro-duction data were matched with a very-low permeability of 0.0005 md in the well. To match the continued steep decline in late-time production data, a drainage area of only 11 acres was used. It should be noted that this drain-age area is not represented by physical geological no-flow boundaries but may just be “apparent constant pressure” boundaries as a result of the extremely low permeability. The approach of lim-ited no-flow boundaries in the single-well modeling does not reflect this scenario fully, but it can describe the fast rate declines effectively. Additional reservoir modeling work illustrated the issue of nonuniqueness if effective frac-ture length were not known.

In many cases, the actual created hydraulic-fracture length is not mea-sured and fracture-model net-pressure matching is the primary method to esti-mate fracture length, both propped and created. However, net-pressure matching can be nonunique and will depend on the assumptions used in the analysis.

The fracture-modeling results emphasize that in many cases, signifi-cant differences in fracture geometry can result when the physics of frac-ture propagation is uncertain. In this example, a fracture length of 525 ft was

estimated assuming complex fracture growth, compared to the actual length of approximately 1,500 ft. If fracture geometry can be determined from an independent measurement such as microseismic mapping, then the frac-ture model can be calibrated by use of appropriate approximations for the physics that governs fracture growth in a given geologic environment.

The effective fracture length and con-ductivity can be estimated using pro-duction analysis; however, production analysis also suffers from nonunique solutions when sufficient data are not available. This analysis results in a very good match of the production data using a reservoir permeability of 0.001 md, fracture half-length of 1,230 ft, and a fracture conductivity of 100 md-ft. However, a good match to the produc-tion data also can be achieved using a reservoir permeability of 0.004 md, a fracture half-length of 535 ft, and a fracture conductivity of 100 md-ft. The production analysis illustrates the uncertainty that can result when reser-voir permeability is unknown, which is common in many tight gas reservoirs.

Canyon Formation. The Canyon sand in the Ozona field in west Texas is a deepwater turbidite deposit character-ized by numerous gas-productive mem-bers with permeabilities from less than 0.001 md to more than 0.1 md. Typical completions require fracture treatments using water-based fluids containing 50,000 to 250,000 lbm of 20/40 sand to achieve economic production rates.

Well OC-1 is a Canyon well perfo-rated from 6,240 to 6,275 ft. The well was drilled on 80-acre spacing, and a prefracture well test was performed to measure initial reservoir pressure and permeability. The well was swabbed dry and flowed for 7 days at 100 Mscf/D and 15-psi FTP. Quartz pressure gaug-es were placed in the well, and a bot-tomhole plug was set in the tubing to minimize wellbore storage. The log-log diagnostic plot of the 7-day PBU showed a permeability of 0.055 md (43 ft of pay) and reservoir pressure of 2,550 psi (virgin reservoir pressure).

The well was fracture stimulated down 41/2-in.casing with crosslinked gel with 30% carbon dioxide (CO2) and 20/40-mesh Ottawa sand. Diagnostic tests were performed using 2% potas-sium chloride water before the propped treatment and indicated a closure stress

of 0.67 psi/ft. Net pressures in excess of 650 psi were recorded during these initial injections, even though only a small volume of a low-viscosity fluid was injected. The high net pressure was an indication of complex fracturing.

The net-pressure data were history matched to estimate fracture geometry, and Fig. 1 shows the results. The figure shows that the measured net pressure ranged from 1,200 to 1,800 psi. The high level of net pressure was matched by assuming that multiple fractures were being propagated, to “simulate” com-plex fracturing. The predicted fracture geometry is shown in Fig. 2, indicating a propped fracture length of 210 ft with an average proppant concentration of 1 lbm/ft2. It should be emphasized that the high level of net pressure could result from any number of phenomena, and the assumption of multiple fractures is not a unique solution. Therefore, it is important to verify fracture geometry independently to ensure that the frac-ture model is reliable.

Well OC-1 was produced for 60 days and then shut in at the surface for a 14-day PBU. The average rate before the PBU was 450 Mscf/D at 550-psi FTP. The calculated fracture length is 180 ft, but the fracture conductivity (FcD=1.8) is very low. The low fracture conductiv-ity was not expected, on the basis of fracture-modeling results that indicated good placement of more than 1 lbm/ft2 of proppant. The measured FcD implies approximately 98% damage to the prop-pant pack, or only 2% retained conduc-tivity. Therefore, fracture-fluid damage is excessive, cleanup is slower than expected, or proppant placement must be poorer than predicted by the fracture model. The CO2-energized fluid was expected to improve cleanup. Finding the cause of the poor fracture conduc-tivity was essential to improve future fracture treatments.

A second post-fracture PBU was per-formed after 3 years of production. Production rate had declined from 500 to 100 Mscf/D in the 3-year period. The analysis indicates a fracture half-length of 220 ft (similar to the first test) but now shows good fracture conductivity. The results of the second post-frac-ture PBU now support the fracture-modeling estimates of both fracture length and conductivity. Therefore, fracture cleanup must be slower than expected, with no proppant-placement problems evident. JPT

JPT • MARCH 2009 61

There has been an increase in the number of wells drilled to depths greater than 20,000 ft in the Gulf of Mexico (GOM). Because of the high fracture gradient and friction in the wellbore tubulars, a conventional 1.0- to 1.04-specific-gravity (SG) fracturing fluid would require surface treating pressures greater than 15,000 psi, which exceeds the limit of the flexible treatment line. To solve this problem, a borate-crosslinked high-density frac-turing (HDF) fluid with SG of up to 1.38 was developed to reduce the amount of surface treating pressure required to achieve adequate bottomhole fractur-ing pressure.

IntroductionThe Tahiti field is in the GOM in the Green Canyon area where water depths range from 4,000 to 4,300 ft. The dis-covery well was drilled in 2002. Total depth was more than 28,000 ft. Initial evaluation indicated approximately 400 ft of net pay in the high-quality reservoir sand that was encountered. Subsequent appraisal drilling over the next 2 years resulted in confirmation of the size of the Tahiti field and its status as one of the most significant net-pay accumulations ever discov-ered in the GOM. The discovery well

was re-entered in 2004, and a well test was performed to verify deliverabil-ity, dynamic well data, and reservoir properties. A stacked frac pack in the Miocene M21A and M21B sands was planned for the well test. The Tahiti M21A sand averages 60 to 80 ft thick, and the M21B sand averages 120 to 150 ft thick. Permeability ranges from 600 to 800 md. The decision was made to complete both intervals with a single, high-rate frac pack. At the time, at a depth in excess of 25,800 ft, it was the deepest successful well test and frac-pack completion ever carried out in the GOM. The HDF fluid was a key component of the successful Tahiti well test. The well-test results led to the development of the Tahiti field, which began in February 2006.

While planning the Tahiti well test, several factors influenced the deci-sion to develop a suitable HDF fluid that would minimize surface treating pressures and allow the fracture job to be pumped at pressures below the

14,000-psi limit. Uncertainty regard-ing Miocene-pay-sand fracture gra-dients, and required treating rates, coupled with high friction losses in the treating string led to the desire to find an HDF fluid that would allow the fractures to be pumped at 40 to 45 bbl/min while staying within surface-treating-pressure limitations.

The initial Miocene-pay-sand frac-ture-gradient estimates ranged from 0.78 to 0.85 psi/ft. At 26,000 ft true vertical depth (TVD), the fracture-gra-dient estimate resulted in a 1,800 psi swing in treating pressure. Because the Tahiti Miocene sand had never been frac packed before, fracture-flu-id efficiency was an unknown, as was required treating rate to com-pletely fracture the M21A and M21B sands. The estimated net-pressure gain ranged from 600 to 1,500 psi. If all the variables tended to the high side, then the Miocene sand could not be fractured with a conventional 8.7-lbm/gal fracture fluid and stay within

This article, written by Assistant Tech-nology Editor Karen Bybee, contains highlights of paper SPE 116007, “Development and Use of High-Density Fracturing Fluid in Deep Water Gulf of Mexico Frac and Packs,” by L. Rivas, SPE, G. Navaira, SPE, and B. Bourgeois, SPE, Chevron, and B. Waltman, SPE, P. Lord, SPE, and T. Goosen, SPE, Halliburton, origi-nally prepared for the 2008 SPE Annual Technical Conference and Exhibition, Denver, 21–24 December. The paper has not been peer reviewed.

Development and Use of High-Density Fracturing Fluid for Deepwater Frac Packs

HYDRAULIC FRACTURING

For a limited time, the full-length paper is available free to SPE members at www.spe.org/jpt.

Fig. 1—Regained-permeability-test results.

62 JPT • MARCH 2009

the 14,000-psi surface-treating-pres-sure limit.

Fluid DevelopmentShear-History Testing. The first series of tests evaluated fluid sensitivity to the tubing shear experienced during the pumping operation. These tests simulated the tubing shear history that an element of fluid would experi-ence if it were pumped at 35 bbl/min from the surface to the perforations (510 sec−1 for 14.7 minutes). Spotting the fluid to the crossover tool at 8 bbl/min was simulated with a shear rate of 127 sec−1 for 52 minutes. Shifting the tool was simulated by shutting down for 5 minutes before adjusting the shear rate to 510 sec−1. Fig. 1 illus-trates the shear history and tempera-ture profile simulated with the Fann Model 50 viscometer testing. Fig. 1 shows that the fluid recovered from the shear, which is not always the case with borate fluids subjected to a shear history this long.

Regained-Permeability Testing. The HDF fluid tested was prepared using 11.5-lbm/gal sodium bromides as the base fluid. The formation core plugs were maintained at the test tem-perature throughout the test by use of a heating jacket. Once the core plugs were heated to temperature, synthetic formation brine was passed through the plugs in the production direction at a rate of 5 mL/min. Brine flow was continued until the pres-sure drop (DP) across the plugs re-mained constant.

An overburden pressure of 1,300 psi was maintained on the Hassler sleeve throughout the test. A backpressure regulator set at 200 psi was placed on the discharge side of the core plugs. A second backpressure regulator set at 1,100 psi was placed on the intake side of the core plug. This maintained a maximum differential pressure across the core plug of 900 psi. Once the differential pressure across the core plug reached 900 psi, fluid not passing through the core plug was discharged from the flow apparatus through the backpressure regulator on the intake side of the plug.

Three Tahiti core plugs approxi-mately 1.5 in. in diameter were used for the test. Two plugs were mounted in a stacked sequence in the Hassler sleeve to ensure linear flow through

the plugs. Synthetic brine was pumped through the plugs in the production direction, and the permeability was calculated. The third plug then was mounted in the stacked sequence to ensure linear flow through the plugs. A spacer was placed at the wellbore end of the third plug, with the void space provided by the spacer giving the treating fluid the capacity to form a dynamic filter cake. Synthetic brine was pumped through the plugs in the production direction, and the perme-ability was calculated.

The crosslinked HDF fluid with breaker was flowed through the appa-ratus in the treatment direction at a rate of 5 mL/min for a period of 1 hour at 90-psi pressure differential. The fluid formed an effective seal against the face of the core plug very rapidly, preventing fluid flow through the plug. The excess fluid flowing over the wellbore face of the core plug was discharged from the apparatus through the backpressure regulator on the intake side of the core plug. The plugs then were shut in, and the temperature was increased slowly to 180°F. After 12 hours, the temperature was increased to 208°F. After 48 hours, the synthetic forma-tion brine was flowed in the produc-tion direction at a rate of 5 mL/min. Flow was continued until the pressure differential across the plugs remained constant. A final permeability then was established and a regained perme-ability calculated.

To further evaluate the regained permeability of core away from the formation face, the first core in the injection sequence, where the frac-ture-fluid filter cake was deposited, was removed from the Hassler sleeve. The synthetic formation brine then was flowed through the remaining two cores in the production direction at a rate of 5 mL/min. Flow was con-tinued until the pressure differential across the formation plugs remained constant. A final permeability then was established and a regained perme-ability calculated.

The regained permeability for the stack of three core plugs was 52%. After the regained permeability for the stack of three core plugs was determined, the first core plug (i.e., the core with the fracturing-fluid filter cake) was removed and the regained permeability for the stack of remain-

S P E S E R V I C E D I R E C T O R Y

SPE Online www.spe.org

Awards ProgramContact: Tom WhipplePhone: 1.972.952.9452 • Email: twhipple@spe.org

Regional/Sectional Awards ProgramContact: Brian WigginsPhone: 1.972.952.9451 • Email: bwiggins@spe.org

Books, Papers, Software SalesPhone: 1.800.456.6863 or 1.972.952.9393Email: books@spe.org

Conference ProgramsContact: Wayne SpencePhone: 1.972.952.9449 • Email: wspence@spe.org

Continuing Education/Short CoursesContact: Chiwila Mumba-Black Phone: 1.972.952.1114 • Email: cmumba@spe.org

Distinguished Lecturer ProgramContact: Donna Neukum Phone: 1.972.952.9454 • Email: dneukum@spe.org

Dues, Membership Information, Address Changes, and Copyright PermissionPhone: 1.800.456.6863 or 1.972.952.9393Email: service@spe.org

Exhibit Space SalesContact: Joan PaynePhone: 1.972.952.9356 • Email: jpayne@spe.orgContact: Jim KlingelePhone: 1.713.779.9595, ext. 612 • Email: jklingele@spe.org Contact: Kirk ColliganPhone: 1.972.952.9516 • Email: kcolligan@spe.org

Insurance, Credit Card ProgramsContact: Alyssa MorenoPhone: 1.972.952.9439 • Email: amoreno@spe.org

JPT Employment/Professional ServicesContact: Mary Jane TouchstonePhone: 1.972.952.9351 • Email: mtouchstone@spe.org

JPT/JPT Online Advertising SalesContact: Jim KlingelePhone: 1.713.779.9595, ext. 612 • Email: jklingele@spe.org

Journal of Petroleum TechnologyContact: John DonnellyPhone: 1.713.779.9595, ext. 616 • Email: jdonnelly@spe.org

Meeting RegistrationPhone: 1.800.456.6863 or 1.972.952.9393Email: registration@spe.org

Peer ReviewContact: Stacie HughesPhone: 1.972.952.9343 • Email: shughes@spe.org

Periodicals, Book Publication Contact: Glenda SmithPhone: 1.972.952.9470 • Email: gsmith@spe.org

Professional Development ServicesContact: Tom WhipplePhone: 1.972.952.9452 Email: twhipple@spe.org

Section ServiceContact: Brian WigginsPhone: 1.972.952.9451 • Email: sections@spe.org

SPE WebsiteContact: Doug LauverPhone: 1.972.952.9519 • Email: dlauver@spe.org

SubscriptionsPhone: 1.800.456.6863 or 1.972.952.9393Email: service@spe.org

Technical Interest GroupsContact: Shasta StephensonPhone: 1.972.952.9514 • Email: sstephenson@spe.org

Young Member ProgramsContact: Melissa SchulteaPhone: +44.207.299.3300 Email: mschultea@spe.org

JPT • MARCH 2009

ing two core plugs was determined. This regained permeability was approximately 84%, indicating that most of the damage was confined to the first core plug. These regained-permeability values are in agreement with values reported in the literature for nonweighted, borate-crosslinked frac-packing fluids.

Compatability Testing. Samples were combined at room temperature at a 50:50 ratio and stirred for 1 minute. Test samples were observed for 15 minutes and checked 1 and 2 hours later. No precipitates were observed when zinc bromide completion fluid was mixed with a broken HDF fluid, a crosslinked HDF fluid, or a linear HDF fluid.

Treatment ResultsAs of July 2008, 17 jobs have been pumped with the HDF fluid. Of these jobs, 12 treatments were performed in the GOM. The development and first use of the HDF fluid was for the Tahiti well-test completion in 2004. The second use was in 2006. Subsequent treatments were performed on the Tahiti development-project comple-tions. Water depths ranged from 4,000 to 6,900 ft. The measured depth of the perforated interval ranged from 25,000 to 28,000 ft, and TVD ranged from 23,000 to 28,000 ft. Bottomhole pres-sure ranged from 19,000 to 19,800 psi, and bottomhole temperature ranged from 229 to 235°F. Treatments ranged from 15 to 45 bbl/min, with 50,000 to 490,000 lbm of total proppant pumped per treatment.

There were many positive results seen with the use of the HDF fluid. The largest and most obvious benefit is lower surface treating pressures. The HDF fluid substantially reduced surface treating pressures. Surface pressures were reduced by 22 to 39%. Actual surface pressures ranged from 5,700 to 10,800 psi. The surface treat-ing pressures were predicted with 83 to 99% accuracy. Friction for the HDF fluid was similar to that of 8.7-lbm/gal fracturing fluids. Without the reduc-tion in surface treating pressures, these treatments could not have been pumped because of pressure limita-tions on the current surface equip-ment and tubulars.

The HDF fluid also maintained physical properties under storage

conditions for 3 months at 80°F with-out compromise or degradation. The operator realized significant savings by storing the fluid leftover after a job and using it for the next job. The performance of the stored fluid also was similar to that of 8.7-lbm/gal fracturing fluids, showing no adverse effects during the minifracture, step-rate test, or main fracture. Typical redesign parameters were obtained, fracture models were calibrated, and job redesign was executed. The HDF fluid also allowed for better sus-pension of higher-strength, higher-SG proppants.

Well PerformanceFlowbacks to the rig were conducted after the first four development wells were completed. Three of the four wells had at least 3 months’ time between running the completions and opening the well to production. The fourth well was opened immediately after the completion activities.

On the basis of flowing data and pressure-transient analysis performed on the buildup at the end of the flowbacks, there were no adverse effects evident as a result of the HDF fluid being injected into this high-permeability sand. Also, by comparing the skins of the wells that were allowed to sit before flowback to that of the well that was flowed back immediately after completion, there appears to be no additional damage to the formation by allowing the HDF fluid to remain in the formation for an extended period of time.

The well that was flowed back immediately after completion exhib-ited a normal rate of cleanup (reduc-tion in water cut) as compared to other GOM well cleanups with simi-lar permeability. An examination of the oil flowback did not indicate any abnormal emulsions because of the HDF fluid. Overall, the performance was very good for these wells. The skin ranges seen after the flowbacks were in line with precompletion esti-mates and with the range of skin achieved on other GOM frac-packed completions. The cleanup of the wells was in line with the trends observed in other GOM frac-packed comple-tions. No abnormal emulsions were observed in the oil flowed back to the rig from these completions. JPT