Strategic Alliance, Multidisciplinary Teamwork Enhance Field Development In Cotton Valley Trend

Holly G. Krus, Ken Haley, Larry Britt
Amoco Exploration & Production Co.
Houston

Ronald H. Benson, Kevin W. England, Ron Holt, Nicholas C. Piskurich
Schlumberger Cos. Houston

Robert A. Woodroof Jr. ProTechnics International
Houston

A strategic alliance and multidisciplinary teamwork improved economic performance of a field in the East Texas basin's Cotton Valley trend.

Optimizing the development phase, and thus improving the profitability, of Amoco Exploration & Production Co.'s Glenwood natural gas field was the task assigned to a multidisciplinary reservoir management team (RMT) in 1996 (Fig. 1[94621 bytes]). The team comprised personnel from Amoco, Schlumberger companies Dowell Wireline & Testing and GeoQuest, and ProTechnics International.

Members were chosen largely from an existing Amoco/Schlumberger strategic alliance that had been in place for 4 years. The team's primary emphasis was to strategically locate and fracture-stimulate the remaining wells to be drilled in Glenwood field's initial development phase.

This article illustrates the team process used and decisions made that led to field cost savings and production improvements before yearend 1996. Thorough data collection and evaluation were critical project elements and enabled the field's hydraulic fracturing program to be modified, thus improving incremental production and eliminating ineffective fracturing costs.

Establishing a team

Over the course of drilling 10 wells in Glenwood field between December 1994 and January 1996, many questions were raised about the efficiencies of the completions being used, including:

• Is the entire pay interval effectively stimulated?

• Is the stimulation achieving the fracture length required to efficiently drain the reservoir given the chosen well pattern?

• To answer these questions, among others, and generally improve the remaining development of this field, the RMT was formed in February 1996.

• The first major hurdle of the RMT was overcome early in the program by having senior management commit resources and people to the project, from its planning stages through execution. Original team membership consisted of both Amoco and Schlumberger personnel.

• At the first RMT meeting, a common set of project objectives was outlined in order to minimize duplication and focus effort during project execution. The seven primary objectives driving all project activities were to:

• Develop a process for real-time post-appraisal.

• Optimize fracture treatments and determine spacing requirements.

• Develop scorecards to monitor progress.

• Assess the impact of completion costs in the entire well investment.

• Improve communication flow.

• Identify and pursue projects with maximum value per dollar spent.

• Identify cutting-edge technology that will benefit field development.

Having RMT members from varying disciplines, companies, and cultures proved advantageous. Team members held expertise in field operations, production, reservoir, research, and service company engineering of many types (Fig. 2 [152731 bytes]). Experience of the individuals ranged from 3 to over 20 years.

All project issues were worked through as a team. Individual or company-specific objectives were put aside, as was the traditional client-vendor approach, resulting in a cooperative effort versus an adversarial atmosphere. Each project step was questioned and analyzed thoroughly by the team for the good of the field.

The RMT met monthly to report on progress made on obligations taken at prior meetings. The meetings also proved to be excellent training sessions for all members. As the project evolved, the group was expanded to include members from ProTechnics International, a company specializing in completion diagnostics, as its personnel offered additional insight into the Cotton Valley play.

Early field development

Glenwood field development began in late 1994 and was targeted for the Cotton Valley sand at about 11,000 ft. Each well was expected to recover 1-2.5 bcf of gas from the lower section of the Taylor interval of the Cotton Valley sand.

At project start, the specific reservoir characteristics of the Cotton Valley sands in the Upshur County portion of the play were unknown. Thus, reservoir parameters from nearby Willow Springs field, Gregg County, were used as a benchmark. Log responses taken in the early Glenwood wells were very similar to those from Willow Springs, and the same reservoir was being completed in both fields (Fig. 3 [76164 bytes]).

The exact method used to design, execute, and evaluate Willow Springs wells was applied to the initial development wells in Glenwood field. While the first few fracture stimulations went as planned, the field crew soon found it could not pump the treatment to completion. In previous Cotton Valley developments, very few problems of this type had been encountered.

Problem analysis revealed that fracture gradients and rock stresses varied throughout Glenwood field, unlike the uniform properties found at Willow Springs (Table 1 [17075 bytes]). Concern grew around all the rock properties assumed for Glenwood field. Were other field assumptions in error?

To solve the fracture treatment pumping problem, diagnostic radioactive tracer surveys were run, and the results were integrated with Nolte-Smith net pressure plots. Data showed that the hydraulic fractures were growing excessively high, limiting proppant placement in the lower pay intervals. Several process improvements were used to solve this problem.

First, the perforation density was increased in the lower pay zones. These perforations were broken down to allow for better proppant placement in these areas. Second, wellhead isolation tools were used during fracture treatments to provide for higher maximum allowable surface treating pressures. Fracture treatments were conducted after these changes were successfully pumped to completion.

Still, the analyses conducted to solve the fracture treatment problem indicated that the reservoir was not fully understood. The problem required additional evaluation of fracture design parameters.

The RMT was formed at this point to design the optimum field development program, and thus maximize field profitability.

The RMT knew from the beginning that in low-permeability sands, such as the Cotton Valley, fracture effectiveness impacts hydrocarbon recovery. A less than optimum fracture treatment, for example, might be reaching only 75% of the well's net pay or might achieve only 75% of the expected fracture length. In this case, a Cotton Valley well might recover only 1.125 bcf of gas versus its 1.5 bcf potential, dropping the internal rate of return from 19% to 9% (Fig. 4 [11481 bytes]).

Further, if fracture treatment is less than optimum, certain offset locations may not be drilled due to the false assumption that the reservoir rock is the limiting factor. For these reasons, obtaining an optimized fracture treatment was foremost for Glenwood field's RMT.

Data acquisition

An initial function of the RMT was to recommend a plan for acquiring the unknown reservoir parameters needed for fracture design optimization. These parameters included formation flow capacity (kh), fracture length (xf), fracture height (hf), and fracture conductivity (kfw).

The RMT's resulting "data well" concept involves strategically placing a well during the very early stages of field development for the purpose of gathering all pertinent reservoir data. In fact, three data wells were chosen for Glenwood field, and the additional data to be gathered were specifically aimed at optimizing the input parameters for the fracture modeling program being used. Table 2 outlines the data acquisition plan.

As it happened, operational issues made it impossible to acquire the complete suite of new Glenwood field data from just three wells. A complete data suite was finally acquired from a total of nine wells over a 3 month period.

Over this period, as input data assumptions were gradually replaced with actual, measured data, each successive well and fracture job showed improvement. The final 26 wells drilled in the field through September 1996 (45 total, counting the 10 pre-RMT wells) reaped the benefits of the data acquisition process.

Thoroughness in data acquisition was a constant theme. When a single fracture parameter could be obtained from multiple sources, all sources were evaluated.

For example, the compressional and shear transit times provided by the Dipole Shear Imaging (DSI or dipole sonic) tool were used with density data to compute rock stress, Young's modulus, and Poisson's ratio. A Modular Formation Dynamic Tester (MDT) tool was also used with a dual packer module to determine closure stress.

In order to characterize downhole conditions, prior to fracture treating, bottomhole pressure gauges were placed in several wells just below the perforations. Pressures were monitored throughout the fracture treatment, and the gauges were recovered with coiled tubing during the clean-out trip.

Radioactive tracer surveys were used in selected wells to determine minimum fracture height at the wellbore. As many as three isotopes were used to determine placement of the pad fluid, and early and late proppant stages.

After perforating of several wells, slug tests were recorded with downhole gauges. Typically, these tests took 3 days to determine formation skin, formation flow capacity, and reservoir pressure.

Knowledge of rock stresses, permeabilities, and downhole pressures allowed the pseudo-3D fracture model selected to be used with confidence. The benefits of this model were not felt early in the project because of the reservoir property assumptions. Actual knowledge of the minimum near- wellbore fracture heights from radioactive tracer logs also enabled a comparison between modeled and tracer-derived heights.

The RMT went so far as to try to compare rock stress signatures and actual fracture orientation in one well by running the DSI tool before and after fracture stimulation. Unfortunately, the data measured through casing were not of adequate quality to be useful.

Data evaluation

Following data collection, data evaluation included the study of fracture height, fracture length, pad size, leak-off coefficients, conductivity requirements, proppant stress, fluid clean-up, and formation permeability.

Bottomhole treating pressure data were collected on several wells to evaluate fracture geometry and the effectiveness of the current hydraulic fracture stimulation treatments. The use of bottomhole treating pressure data in the East Texas Cotton Valley formation for this purpose goes back to 1979 and has been a routine fracture treatment evaluation tool.

Fig. 5 [62277 bytes] shows a log-log plot of net treating pressure versus time for two wells in the 1996 Glenwood program. As shown, the net treating pressure increased normally for over 200 min, then began increasing on a unit slope indicative of a tip screen-out. Further, the fracture stimulation treatments in these wells were pumped nearly 90 min without any appreciable growth in fracture length.

This tip screen-out pressure behavior is analogous to a wellbore storage phenomenon in that the fracture has ceased growing in length (assuming a constant height), and the slurry is being stored in an ever-widening fracture. Increasing fracture width in this manner is critical to the success of high-permeability fracturing; however, in low-permeability formations, such as the Cotton Valley, the critical dimension is fracture length.

Collecting bottomhole pressure data also made evident that calculating bottomhole treating pressure from measured surface pressure was not accurate, in some cases, due to under or over-estimated friction pressure, and, therefore, could not be used to make real-time job changes.

Fig. 6 [77358 bytes] shows a plot of net treating pressure versus pump time from an example well. As shown, the calculated bottomhole treating pressure varied considerably from the measured bottomhole treating pressure. Because of this type of discrepancy, problems could be occurring downhole that were not visible at the surface, or a fracture treatment might be stopped due to a false assumption at the surface that downhole problems were occurring.

Based on this analysis, two actions were possible. Either the pad fraction could be increased to allow continued fracture growth throughout the entire stimulation, or the total slurry volume could be reduced. Choosing the latter solution would bring a reduced slurry cost; however, the former solution was preferred given the importance of fracture length to the ultimate recovery of wells in tight formation gas reservoirs.

Fig. 7 [56949 bytes] displays a plot of cumulative gas recovery versus time as a function of fracture length for a typical East Texas Cotton Valley tight formation gas well having 0.01 md permeability. As shown, increasing the fracture half-length from 500 to 1,000 ft increases recoverable reserves from 2.5 bcf to 3.0 bcf. Increasing the fracture half-length to 2,500 ft results in a cumulative recovery of 3.7 bcf, which is 1.2 bcf more gas than from a well with a 500 ft fracture half-length. Clearly, fracture length directly impacts gas recovery from tight formation gas reservoirs.

Computer model

To further assess the importance of fracturing on tight formation gas reservoirs, and ultimately Glenwood field's depletion plan, the RMT used a computer model to determine the optimum dimensions of fracture half-length, fracture conductivity, aspect ratio (the ratio of the length of the drainage area parallel to the fracture to the length of the drainage area perpendicular to the fracture), and well spacing. This included performing numeric, multiwell reservoir simulations for various fracture dimensions, contouring the pressures, and determining the aspect ratios for each well.

Fig. 8 [20129 bytes] shows the resulting relationships between aspect ratio, fracture half-length, and dimensionless fracture conductivity (Fcd). As shown, the aspect ratio for a 1,000 ft fracture is 1.5:1 and is independent of Fcd.

Further, this figure shows the effect of Fcd on aspect ratio as fracture half-length increases. The importance of fracture conductivity and/or dimensionless fracture capacity increases as the fracture half-length increases to 2,500 ft. For a 2,500 ft fracture with an Fcd of 1, the aspect ratio is 2:1. Given an Fcd of 20, the aspect ratio for the same fracture is 14:1.

Results of the model study showed definitively how fracturing is critical to Glenwood field's ultimate depletion plan. Combining this knowledge with the significant hydraulic fracturing issues raised by the bottomhole treating pressure data shown in Fig. 5 led the RMT to believe a more-extensive data collection and interpretation effort was warranted.

The extended data collection program called for further evaluation of stresses, rock properties, and fracture fluid leak-off. This was accomplished with DSI logs and in situ stress tests. In addition, conventional cores were collected, and static rock properties were evaluated.

Fig. 9 [61871 bytes] displays a geomechanical profile developed from dipole sonic and density log data. A gamma ray log, stress profile, Young's modulus, and Poisson's ratio are shown in tracks 1 through 4, respectively. The stress profile was developed from an analysis of shear and compressional wave travel times and calibrated with an in-situ stress test and fracture closure pressure information. The log-derived elastic properties were calibrated with static (triaxial) rock properties measurements conducted on conventional cores from the area. Next, the stress profile, rock properties, and fracture treatment designs were input, and a history matching procedure was conducted using the measured bottomhole treating pressure data. Results of this evaluation were compared with radioactive tracer information, which verified fracture height at the wellbore.

Fig. 10 [67171 bytes] shows a plot of the resulting history-matched treating pressure data. A good match of the data was obtained once all of the collected information was merged into the evaluation. With this history match, subsequent fracture stimulations were modified. These modifications, based on the evaluation of both prejob and postjob inputs, led to design changes, ultimately improving overall stimulation effectiveness.

The elevated fracture closure pressures, however, raised concerns over whether the proppant (20/40 Ottawa sand) could withstand pressure cycling, which was due to production operations, without crushing. As a result, a lab study assessed the effects of fracture closure stress and stress cycling on fracture conductivity using static triaxial testing.

The first series of tests was conducted with a 1 psf proppant concentration (1/8 in. pack) using the net confining stress conditions encountered in the field. The confining stress was released, simulating well shut-in, and then reapplied. Two additional stress cycles were conducted and evaluated. The second set of tests was conducted in a similar fashion with a 2 psf proppant concentration (1/4 in. pack).

Fig. 11 [17111 bytes] shows the results of these tests as a plot of fracture conductivity versus stress cycle. For a 1 psf proppant concentration, a fracture conductivity of 600 md-ft was achieved. As the fracture stresses were cycled, simulating wellbore shut-in, the fracture conductivity was reduced 58% to 250 md-ft. Similarly, for a 2 psf proppant concentration, stress cycling reduced fracture conductivity from 1,300 md-ft to 550 md-ft, a 58% reduction.

These test results indicated that although fracture conductivity was negatively impacted by stress cycling, catastrophic failure did not occur, and adequate fracture conductivity remained with the 20/40 Ottawa sand. A more expensive, man-made, intermediate-strength proppant did not have to be used.

Value

The value of the rigorous data collection and evaluation program outlined and implemented by the Glenwood field RMT was measured by well performance improvements.

Fig. 12 [70076 bytes] shows a plot of cumulative gas recovery versus production time, comparing the average Amoco well in Glenwood field to the average offset well. The Amoco wells are outperforming offset wells.

By using actual, acquired data, the team discovered that the fracture half-lengths being generated were as much as 20% less, in some cases, than the desired fracture half-lengths. This decrease in fracture half-length amounted to unnecessary expenses totaling as much as $45,000/well in ineffective hydraulic fracture treatment cost.

With bottomhole data and mechanical rock properties data in hand, the hydraulic fracture design was modified to increase fracture half-length. These changes in the stimulation design improved incremental production from one well by 126.5 MMcf over a 2 year period (Fig. 13 [28117 bytes]). At a natural gas price of $2.50/Mcf, this incremental production equates to incremental gross revenue of $316,250.

Further, evaluating field rock properties with the aid of a mechanical properties log and lab conductivity tests resulted in a decision to run Ottawa sand in the field versus more expensive, man-made proppants. This saved nearly $200,000/well.

Observations

At the close of 1996, the Glenwood field RMT had realized many tangible and intangible benefits for Amoco and had assembled technical and process observations of benefit to others inside and outside of Amoco.

Technical observations included:

• Fracture length is the key design parameter for low-permeability fracturing.

• The importance of fracture conductivity on a well's aspect ratio increases as the fracture length increases.

• Although fracture conductivity can be hurt by stress cycling, catastrophic failure did not occur with 20/40 Ottawa sand at Glenwood field. Adequate fracture conductivity remained, so intermediate-strength proppant was not needed.

• Calculated bottomhole treating pressure data may not always be indicative of downhole conditions, and therefore should be used with caution when making real-time job changes.

• Process observations included:

• A multidisciplinary team working with a common goal enabled a more efficient resolution of technical issues at Glenwood field.

• A team of this nature should be in place at the onset of any new field development to maximize economic impact.

• An aggressive data collection and evaluation effort has been critical to the field optimization and well spacing goals achieved at Glenwood field.

• The "data-well" concept should be used in Amoco's new field developments to properly determine unknown reservoir parameters and fully optimize field development.

• In the early stages of new field development, the pace should be slow enough to make necessary design modifications. Once critical design parameters are determined, the development pace can be increased.

• Each team member, regardless of experience level, gained valuable knowledge outside of his or her area of expertise as part of the RMT.

• Better and more open communication improved the quality, timeliness, and implementation of decisions for Glenwood field.

The RMT concept applied to Glenwood field proved that with open and honest communication between all team members-oil company and service company personnel alike-and the application of new and appropriate technology, improved well or field economics can be quickly achieved.

Acknowledgments

The team thanks the following people for their valuable support during the Glenwood field program and in preparing this article: Oscar Esparza, Amoco technologist; Pam Jones, Amoco technologist; Roy Dove, Schlumberger GeoQuest log analyst; and members of the Schlumberger Production Enhancement Group.

Bibliography

Hickman, T.S., "A Rationale for Reservoir Management Economics," Journal of Petroleum Technology, October 1995, p. 886.

Holstein, E.D., Berger, A.R., "Measuring the Quality of a Reservoir Management Program," Journal of Petroleum Technology, January 1997, p. 52.

Nolte, K.G., Smith, M.B., "Interpretation of Fracturing Pressures," SPE paper 8297, SPE Annual Technical Conference and Exhibition, September 1979.

Nolte, K.G., "Determination of Fracture Parameters from Fracturing Pressure Decline," SPE paper 8341, SPE Annual Technical Conference and Exhibition, September 1979.

Nolte, K.G., "Principles of Fracture Design Based on Pressure Analysis," SPE Production Engineering, February 1988, pp. 22-30.

Raza, S.H., "Data Acquisition and Analysis: Foundational and Efficient Reservoir Management," Journal of Petroleum Technology, April 1992, p. 466.

Satter, A., Varnon, J.E., Hoang, M.T., "Integrated Reservoir Management," Journal of Petroleum Technology, December 1994, p. 1057.

Stiles, L.H., Magruder, J.B., "Reservoir Management in the Means San Andres Unit," Journal of Petroleum Technology, April 1992, p. 469.

Thakur, G.C., "Implementation of a Reservoir Management Program," SPE paper 20748, SPE Annual Technical Conference and Exhibition, New Orleans, September 1990.

Torvund, T., "The Oseberg Reservoir Management Planning: A Case History From the Oseberg Field," OTC paper 6140, Offshore Technology Conference, Houston, May 1989.

Wiggins, M.L., Startzman, R.A., "An Approach to Reservoir Management," SPE paper 20747, SPE Annual Technical Conference and Exhibition, New Orleans, September 1990.

The Authors

Holly G. Krus is a petroleum engineer for Amoco Exploration & Production Co. She joined the company in 1993 and currently provides production and reservoir engineering support to Amoco's East Texas Cotton Valley fields. She has a BS in mechanical engineering from Texas Tech University and is an SPE member.

Ken Haley is a staff petroleum engineer for Amoco with 15 years of experience, mainly in the U.S. Gulf Coast. He has a BS in chemical engineering from Ohio State University and an MS in petroleum engineering from the University of Houston. He is an SPE member.

Larry Britt is a petroleum engineering associate for Amoco. He joined the company in 1979 and worked in numerous engineering positions prior to his current assignment as hydraulics fracturing team leader at Amoco's Technology Center, Tulsa. He has a BS in geological engineering from the University of Missouri-Rolla and is an SPE member.

Ronald H. Benson is an alliance coordinator assigned to Amoco for Schlumberger Well Services. He joined Schlumberger in 1973 and served in varying engineering and management positions before receiving his current assignment. He has a BS in mechanical engineering from the Georgia Institute of Technology and is an SPE member.

Kevin W. England is area engineer, southeastern U.S., for Dowell, a division of Schlumberger Technology Corp. He joined the company in 1984 and has held various U.S. engineering positions. He currently focuses on completion and stimulation technology. He has a BS in petroleum engineering from the University of Tulsa and an MS in advanced techniques in the production of hydrocarbon reservoirs from l'Ecole Nationale Sup?rieure du P?trole et des Moteurs, Rueil-Malmaison, France. He is an SPE member.

Ron Holt is a Dowell Schlumberger alliance coordinator assigned to Amoco's southeastern business unit. He joined the company in 1987 and served in various U.S. engineering and management positions before receiving his current assignment as coordinator for the various Schlumberger companies' efforts for Amoco's southeastern business unit. He has a BS in petroleum engineering from Montana Tech and is an SPE member.

Nicholas C. Piskurich is an alliance engineer for Dowell Schlumberger assigned to Amoco. He joined the company in 1990 and held various U.S. engineering positions before receiving his current assignment, which focuses on stimulation and completion support to Amoco's southeastern business unit. He has a BS in petroleum engineering from Pennsylvania State University and is an SPE member.

Robert A. Woodroof Jr. is the technical manager for ProTechnics International, Houston, responsible for technical data output products and managing its Houston Computing Center. He joined the company in 1995 after spending 23 years with Western Co./BJ Services in various R&D and technical management positions. He has a BS in chemistry from the University of Texas at Arlington and has published on acid corrosion inhibition, well stimulation, and radioactive tracing. He currently serves as vice-chairman of the SPE's Gulf Coast Section.