Tom Bratton Denver, Colorado, USA Dao Viet Canh Nguyen Van Que Cuu Long Joint Operating Company (JOC) Ho Chi Minh City, Vietnam Nguyen V. Duc VietSovPetro Vung Tau City, Vietnam Paul Gillespie David Hunt Hydro Bergen, Norway Bingjian Li Ahmadi, Kuwait Richard Marcinew Satyaki Ray Calgary, Alberta, Canada Bernard Montaron Dubai, United Arab Emirates Ron Nelson Broken N Consulting, Incorporated Cat Spring, Texas, USA David Schoderbek ConocoPhillips Calgary Lars Sonneland Stavanger, Norway For help in preparation of this article, thanks to Lee Conn, MI LLC, Houston; Phil Christie, John Cook and Michael Williams, Cambridge, England; Adam Donald and Omer Gurpinar, Denver, Colorado; Peter Kaufman, Pittsburgh, Pennsylvania, USA; and John Lassek, Sugar Land, Texas. BorTex, ClearFRAC, CMR (Combinable Magnetic Resonance), ECLIPSE, FMI (Fullbore Formation MicroImager), Formation MicroScanner, GeoFrame, geoVISION, MDT (Modular Formation Dynamics Tester), Petrel, RAB (Resistivity-at-the- Bit), Sonic Scanner, Variable Density and VDA (Viscoelastic Diverting Acid) are marks of Schlumberger. Naturally fractured reservoirs present a production paradox. They include reservoirs with low hydrocarbon recovery: these reservoirs initially may appear highly productive, only to decline rapidly. They are also notorious for early gas or water breakthrough. On the other hand, they represent some of the largest, most productive reservoirs on Earth. The paradoxical nature of this class of reservoirs is the impetus behind the industrys efforts to learn more about them and model them with a reasonable amount of certainty. Nearly all hydrocarbon reservoirs are affected in some way by natural fractures, yet the effects of fractures are often poorly understood and largely underestimated. In carbonate reservoirs, natural fractures help create secondary porosity and promote communication between reservoir compartments. However, these high-permeability conduits sometimes short-circuit uid ow within a reservoir, leading to premature water or gas production and making secondary-recovery efforts ineffective. Natural fractures also occur in siliciclastic reservoirs of all types, complicating seemingly straightforward matrix-dominated production behavior. In addition, natural fractures are the main producibility factor in a wide range of less conventional reservoirs, including coalbed-methane (CBM), shale-gas, basement- rock and volcanic-rock reservoirs. Although natural fractures play a lesser role in high- porosity, high-permeability reservoirs such as turbidites, they commonly form barriers to ow, frustrating attempts to accurately calculate recoverable reserves and predict production over time. Ignoring the presence of fractures is not optimal reservoir management; eventually, frac - tures cannot be ignored because the tech nical and economic performance of the reservoir degrades. 1 The biggest risk in not characterizing natural fractures early is that such an oversight can severely limit future eld-development options. For example, a company that does not take advantage of the opportunities to evaluate natural fractures during the early development stage may waste resources on unnecessary inll drilling. Asset teams may never extract the hydrocarbons originally deemed recoverable because, without understanding the impact of natural fractures on production behavior, they have not adequately prepared the eld for secondary recovery. This article examines the impact of natural fractures on hydrocarbon reservoirs at different stages of reservoir development. The classi - cations of natural fractures and naturally frac - tured reservoirs (NFRs) are reviewed, along with factors that affect NFR behavior. We describe methods used over a range of scales to identify and characterize natural fractures and to model the inuence of fracture systems on production. Case studies from around the world highlight various approaches. Natural Fractures in Field Development The investigation of natural fractures should start during the exploration stage. Relevant surface outcrops of the reservoir section or reservoir analogs can form the basis of a lithological, structural and stratigraphic foundation from which geologists build conceptual models. These models often begin with knowledge of the regional stresses (next page). 2 The stress state is important in NFRs because the stress state 58732schD04R1.qxp:58732schD04R1 10/10/06 9:44 PM Page 4 Summer 2006 5 largely dictates whether fractures are open to conduct reservoir uids. In addition, the magnitude and direction of horizontal stresses play critical roles in hydraulic fracture design, the primary stimulation method for NFRs. Multicomponent (3C) seismic surveys acquired early in eld development yield important data for determination of azimuthal anisotropy, which is essential to characterize natural fractures and to place wells effectively. 3 For example, knowing the general orientation of fracture systems during well planning dramatically improves the chance that a well will intersect fractures. New wells present an opportunity to collect appropriate geological, geophysical and mechan - ical data from many sources, including infor - mation from logging tools, borehole seismic surveys, sampling devices and fullbore cores. Other valuable sources of information that can be acquired during the early stages of eld development include drillstem tests, initial ow tests, and buildup and drawdown tests. Properly assessing the role of natural fractures can result in early eld-development successes and can lay the groundwork for later development stages, including secondary-recovery projects. Information about natural fractures is also important during the well-construction stage. During overbalanced drilling and cementing operations, open natural fractures can cause lost circulation problems, loss of expensive drilling uids and the potential loss of wells. A less obvious cost may be associated with the reduced productivity that results when drilling uids and cement seal fractures that were once open and potentially productive. 4 Employing underbalanced drilling techniques and using less damaging 1. Nelson RA: Evaluating Fractured Reservoirs: Introduction, Geologic Analysis of Naturally Fractured Reservoirs, 2nd ed. Woburn, Massachusetts, USA: Gulf Professional Publishing (2001): 12. 2. For more on world stress data: http://www-wsm.physik. uni-karlsruhe.de/pub/introduction/introduction_ frame.html (accessed May 18, 2006). 3. Kristiansen P, Gaiser J and Horne S: How Multicomponent Seismic Can Be Used to Manage Fractured Carbonate Reservoirs, paper SPE 93762, presented at the 14th SPE Middle East Oil & Gas Show and Conference, Bahrain, March 1215, 2005. 4. Ehlig-Economides CA, Taha M, Marin HD, Novoa E and Sanchez O: Drilling and Completion Strategies in Naturally Fractured Reservoirs, paper SPE 59057, presented at the SPE International Petroleum Conference and Exhibition, Villahermosa, Mexico, February 13, 2000. > World stress map showing stress data compiled from various sources. In oil and gas regions, borehole measurements are an important source of present-day in-situ stress information. This basic information is used in modeling to help understand fracture networks in fields worldwide. (From the World Stress Map Project, http://www-wsm.physik.uni-karlsruhe.de/pub/casmo/content_frames/stress_maps_frame.html, used with permission.) Regime Thrust fault Normal fault Strike slip Unknown Method Focal mechanisms Breakouts Drilling-induced fractures Borehole slotter Overcoring Hydraulic fractures Geological indicators 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 5 drilling or cementing uids are possible ways to reduce lost circulation and its associated damage. However, in many cases, drillers options are more limited. When drilling weakened and depleted NFRs surrounded by low-permeability shales or overpressured zones, drillers must maintain a certain mud weight to support the shale or prevent a blowout from the overpressured zone. Through the years, innovative techniques have been developed to limit the risk, cost and damage caused by lost circulation problems. These include heating the drilling uid to alter the stress state around the borehole; treating the mud with specialized lost circulation materialsuch as berswhen losses start to occur; pretreating the drilling uid with particulate material; and strategically changing the stresses around the wellborefor example, by creating fractures. 5 In some cases, natural fractures are so large that drastic measures are required. For example, in some carbonate NFRs in central Asia, drilling- uid losses have reached 80,000 barrels [12,712 m 3 ] in long intervals of highly fractured and porous rock. The keys to addressing serious and recurring lost circulation problems are planning for losses, dening the target and having the required equipment and materials available when problems occur. 6 A detailed knowledge of the fracture system is essential to mitigation. Today, MWD tools can monitor critical drilling parameters in real time, allowing drilling engineers to mitigate lost circulation problems. In addition, LWD technology, such as the geoVISION imaging-while-drilling service and the RAB Resistivity-at-the-Bit tool, help identify natural fractures immediately after drilling past them. 7 Incorporating natural-fracture information and rock mechanical properties into cement-job designs reduces the risk of opening up natural fractures or accidentally fracturing the forma - tion, both of which could cause lost circulation. Once well construction and evaluation are nished, the focus moves to designing a completion and stimulation program to undo the damage caused by drilling and cementing. Some form of stimulation is required for most NFRs with a low-permeability matrix. Pumping reactive uidsacidizing, using various formulations of hydrochloric acid [HCl] or chelantsinto natural fractures is most common in carbonate reservoirs to remove near-wellbore damage, enhance connectivity and improve the conduc - tivity of the system. 8 During carbonate-rock stimulation using reactive uids, zones with the highest permeabilities commonly take most of the treatment uid, leaving the zones with lower permeabilities untreated. Consequently, diver - sion, leakoff and reaction-rate control are keys to success when acidizing carbonates. 9 Conventional approaches to diversion include particulate- and viscosity-based-diversion methods. Particulate diversion uses solids to bridge and restrict ow to highly permeable or fractured zones. For example, rock salt or benzoic acid akes are pumped to divert in the formation at the loss zone, and ball sealers are used to mechanically divert from inside tubulars at the perforations. Viscosity-based diversion uses foams, and acids or uids gelled with viscoelastic surfactants or polymers to divert treatment and provide uid-loss control within the formation. However, polymers have damaged reservoirs, prompting service companies to develop new surfactant-base uids. For example, the VDA Viscoelastic Diverting Acid system has been used to successfully stimulate fractured carbonate reservoirs all over the world, including Kuwait, Saudi Arabia, Mexico and Kazakhstan. 10 In addition, a new technique that uses both technologiesbrous particulate and viscosity diversionhas been developed for acidizing NFRs. Natural fractures in siliciclastic reservoirs are also occasionally acidized, typically using a combination of HCl and hydrouoric acid [HF]. Alternatively, hydraulic fracture stimulation of NFRs requires that the main fracture path be kept open and conductive with proppant. Controlling the leakoff rate and effective proppant placement, while minimizing damage to the natural-fracture network, are critical to achieve optimal stimulation and production. Natural fractures can signicantly limit the ability to place large volumes of proppant within a hydraulically created fracture. Various techniques are used to limit natural-fracture dilation and the corresponding uid losses during hydraulic fracturing. These include reducing fracture net pressure by rate-control or low-viscosity uids, and incorporating properly graded particulates to dynamically bridge dilating ssures, thereby reducing total leakoff volume. Additionally, conductivity damage within the created hydraulic fracture and natural- fracture system can be reduced by lowering the total volume of polymer usedfor example, by using low-polymer crosslinked frac gels, increas - ing breaker-to-polymer ratios through the use of encapsulated breakers, or by replacing the polymeric fracture uid with nondamaging viscoelastic surfactant uid systems such as ClearFRAC polymer-free frac uid. 11 The volume occupied by typical fractures open or mineral-lledwithin a vast matrix is usually relatively minuscule, yet the ability of fractures to signicantly impact uid-ow behavior in hydrocarbon reservoirs is enormous. It is not surprising that one of the greatest challenges facing reservoir experts is how to adequately simulate the effects of fractures on reservoir behavior. Understanding these reservoirs requires the acquisition and analysis of vast amounts of data, and usually begins with detailed, foot-by-foot characterization of the fracture and matrix systems. It is the interaction between these two systems that must be understood while reservoir properties change with continued production or injection. As eld development continues, other informationfor example, well-test data, production data, and passive and time-lapse seismic datahelps validate and improve reservoir models. The strategy a company uses to achieve eld- production and recovery potential is intertwined with, and increasingly directed according to, an ever-improving NFR model and simulation. During the primary-production stage, changes in reservoir pressure, and consequently effective stress, alter the uid ow within fracture networks. 12 Water or gas breakthrough is the most common negative implication of conductive fractures during the primary-production stage. Besides adding water production and disposal costs, producing high-mobility water leaves behind substantial volumes of low-mobility oil. Moreover, premature gas production can drain a reservoir of its energy, damage downhole pumps and complicate surface treatment of produced reservoir uids. Secondary-recovery techniques using uid injection also change eld pressure and effective stress dynamics, and therefore change fracture conductivity to uid ow. At this stage in eld development, asset teams should be familiar with the role natural fractures play in large-scale uid movement. Ideally, production and secondary-recovery strategieswell patterns and spacings, and selection of injection and production zonesshould reect the level of inuence that natural fractures have on hydrocarbon sweep as determined by simulation. 6 Oileld Review 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 6 Summer 2006 7 Classifying Fractures When developing and modeling fractured reservoirs, the ability to understand and predict the characteristics of fracture and fault systems is essential. 13 The complexity of natural-fracture systems is captured in the descriptive, genetic and geometric methods that geoscientists employ to classify natural fractures. Knowing fracture types enhances the simulation of uid ow through fractures, because various types of fractures conduct uid differently. To appreciate common classication schemes, a basic understanding of how natural fractures develop is needed. However, achieving this understanding requires more than extensive eld observation of natural fractures; it requires linking those observations with data from controlled laboratory experiments. 14 In the laboratory, fracture types are divided into two groups related to their mode of formation: shear fractures that form with shearing parallel to the created fracture, and tension fractures that form with tension perpendicular to the created fracture. In the laboratory, shear and tension fractures form in consistent orientation with respect to the three principal stress directions, namely the maximum compressive principal stress, 1 , the minimum compressive principal stress, 3 , and the intermediate stress, 2 (left). Shear fractures are created under high differential stress and in conjugate pairs, forming an acute angle with 1 . Tension fractures, a term sometimes used interchangeably with extension fractures, form perpendicular to 3 and at relatively low differential stresses, when the value of 3 , after adjustment for pore pressure the local effective stressis likely tensile. In the laboratory, it is common to observe the creation of tension fractures during compression experiments at low conning pressures and in association with shear fracturing. 15 Shear and tension fractures described from laboratory experiments have clear counterparts that occur naturally; shear fractures correspond to faults, whereas tension fractures corre spond to joints. 16 This mechanically based distinction provides a useful way to classify fractures. Most faulting occurs during signicant tectonic events when the differential stress is high. Tectonic faults typically occur over a broad range of scales, with displacements that range from millimeters to kilometers. Seismic images generally allow the detection of the larger faults, while borehole data are required to identify and characterize smaller faults. Tectonic faults typically cut unimpeded through stratigraphy and are therefore termed non-stratabound. Joints, or fractures having no visible displace - ment, form perpendicular to bedding. Joints can be either stratabound or non-strata bound. Stratabound joints stop at bedding surfaces and often develop a regular spacing and form well- organized connected networks in plan view. Commonly, there is a long and continuous set of joints, termed systematic joints, which are joined by a perpendicular array of cross joints that abut the systematic joints. 17 Non-stratabound joints occur on a wide range of scales and are spatially clustered. 18 5. Aston MS, Alberty MW, McLean MR, de Jong HJ and Armagost K: Drilling Fluids for Wellbore Strengthening, paper IADC/SPE 87130, presented at the IADC/SPE Drilling Conference, Dallas, March 24, 2004. Morita N, Black AD and Guh G-F: Theory of Lost Circulation Pressure, paper SPE 20409, presented at the SPE Annual Technical Conference and Exhibition, New Orleans, September 2326, 1990. 6. Ivan C, Burton J and Bloys B: How Can We Best Manage Lost Circulation? paper AADE-03-NTCE-38, presented at the AADE National Technology Conference Practical Solutions for Drilling Challenges, Houston, April 13, 2003. 7. Inaba M, McCormick D, Mikalsen T, Nishi M, Rasmus J, Rohler H and Tribe I: Wellbore Imaging Goes Live, Oileld Review 15, no. 1 (Spring 2003): 2437. Cheung P, Hayman A, Laronga R, Cook G, Flournoy G, Goetz P, Marshall M, Hansen S, Lamb M, Li B, Larsen M, Orgren M and Redden J: A Clear Picture in Oil-Base Muds, Oileld Review 13, no. 4 (Winter 2001/2002): 227. Bargach S, Falconer I, Maeso C, Rasmus J, Bornemann T, Plumb R, Codazzi D, Hodeneld K, Ford G, Hartner J, Grether B and Rohler H: Real-Time LWD: Logging for Drilling, Oileld Review 12, no. 3 (Autumn 2000): 5878. 8. Al-Anzi E, Al-Mutawa A, Nasr-El-Din H, Alvarado O, Brady M, Davies S, Fredd C, Fu D, Lungwitz B, Chang F, Huidobro E, Jemmali M, Samuel M and Sandhu D: Positive Reactions in Carbonate Reservoir Stimulation, Oileld Review 15, no. 4 (Winter 2003/2004): 2845. 9. Diversion is a technique used in injection treatments to ensure uniform distribution of treatment uid across the treatment interval. Injected uids tend to follow the path of least resistance, such as an open natural fracture, possibly resulting in the least permeable areas receiving inadequate treatment. By using some means of diversion, the treatment can focus on the areas requiring the most treatment. To be effective, the diversion effect should be temporary to allow full restoration of well productivity when the treatment is complete. 10. Al-Anzi et al, reference 8. Albuquerque MAP, Ledergerber AG, Smith CL and Saxon A: Use of Novel Acid System Improves Zonal Coverage of Stimulation Treatments in Tengiz Field, paper SPE 98221, presented at the SPE International Symposium and Exhibition on Formation Damage Control, Lafayette, Louisiana, USA, February 1517, 2006. 11. Samuel M, Polson D, Graham D, Kordziel W, Waite T, Waters G, Vinod PS, Fu D and Downey R: Viscoelastic Surfactant Fracturing Fluids: Applications in Low- Permeability Reservoirs, paper SPE 60322, presented at the SPE Rocky Mountain Regional Low Permeability Reservoirs Symposium and Exhibition, Denver, March 1215, 2000. Samuel M, Card RJ, Nelson EB, Brown JE, Vinod PS, Temple HL, Qu Q and Fu DK: Polymer-Free Fluid for Hydraulic Fracturing, paper SPE 38622, presented at the SPE Annual Technical Conference and Exhibition, San Antonio, Texas, October 58, 1997. Chase B, Chmilowski W, Marcinew R, Mitchell C, Dang Y, Krauss K, Nelson E, Lantz T, Parham C and Plummer J: Clear Fracturing Fluids for Increased Well Productivity, Oileld Review 9, no. 3 (Autumn 1997): 2033. 12. Lorenz JC: Stress-Sensitive Reservoirs, paper SPE 50977, Journal of Petroleum Technology 51, no. 1 (January 1999): 6163. 13. http://www.naturalfractures.com/ (accessed April 20, 2006). 14. Stearns DW and Friedman M: Reservoirs in Fractured Rock, in King RE (ed): Stratigraphic Oil and Gas Fields Classication, Exploration Methods and Case Histories, American Association of Petroleum Geologists, Memoir 16. Tulsa: AAPG (1972): 82106. 15. Engelder T: Stress Regimes in the Lithosphere. Princeton, New Jersey, USA: Princeton University Press (1993): 2425. 16. Pollard DD and Aydin AA: Progress in Understanding Jointing over the Past Century, Geological Society of America Bulletin 100, no. 8 (1988): 11811204. 17. Gross MR: The Origin and Spacing of Cross Joints: Examples from the Monterrey Formation, Santa Barbara Coastline, California, Journal of Structural Geology 15, no. 6 (June 1993): 737751. 18. Odling NE, Gillespie P, Bourgnie B, Castaing C, Chils J-P, Christensen NP, Fillion E, Genter A, Olsen C, Thrane L, Trice R, Aarseth E, Walsh JJ and Watterson J: Variations in Fracture System Geometry and Their Implications for Fluid Flow in Fractured Hydrocarbon Reservoirs, Petroleum Geoscientist 5, no. 4 (November 1999): 373384. > Principal stresses and the creation of fractures. The diagram shows the directions of the three principal stressesmaximum compressive principal stress, 1 , the minimum compressive principal stress, 3 , and the intermediate stress, 2 . The resultant fracturing is also indicated. Tension fractures (green) form parallel to 1 and 2 . The acute angle that forms between two shear fractures (red) is called the conjugate angle. The angle that forms between the shear fracture and 1 is called the dihedral angle. An obtuse angle forms between the shear fracture and 3 , while the shear fractures are parallel to 2 . 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 7 The origin of joints is often difcult to determine, but it is known from rock mechanics that they occur at low effective 3 . Truly tensile stress occurs at shallow depths, so some joints form close to the surface. However, at reservoir depths, joints can probably form only under high uid pressure, a process similar to hydraulic fracturing during well stimulation. As joints do not involve displacement that offsets bedding, they cannot be directly observed on seismic images, but can be located and characterized by well-log data and borehole images (above). While it is relatively simple for a geologist to distinguish faults and joints at an outcrop, the distinction is often less clear using subsurface data, as stratigraphic offsets may not be resolvable. Geologists may therefore have to rely on a number of criteria, such as fracture ll, orientation and spatial distribution, to deter mine whether fractures of a given set are likely to be faults or joints. It may be necessary in such cases to develop a pragmatic classication system based on observed properties of the fractures. Other types of fractures are created by volume-reduction mechanisms within the rock and not from external forces. These include desiccation cracks, syneresis fractures, thermal contraction fractures and mineral phase-change fractures. Of these, syneresis, or chicken-wire fractures, and mineral phase-change fractures in carbonates have the greatest importance in oil and gas production. Syneresis fractures are formed by a chemical process that causes dewatering and associated volume reduction. Carbonate rocks are easily dissolved in freshwater or aggressive uids and the dissolution is often concentrated to form caves or vugs. The resulting porosity is termed karst and is important in many fractured carbonate reservoirs. Maps of karst often show that the porosity is most strongly enhanced along the planes of preexisting fractures and so clarifying the underlying fracture system can often help in understanding karst systems. Because carbonates dissolve relatively easily under pressure, they have a tendency to form stylolitesuneven surfaces of insoluble residue that form perpendicular to 1 . Stylolites may cause local permeability reduc tion, or alternatively they may facilitate subse quent dissolution and perme - ability increase. Tension gashes, or fracturing associated with stylolites, are common (next page, top). 19 While tension gashes may contribute to permeability measured in core, their subsurface impact on reservoir producibility is thought to be minimal. A genetic classication system examines how fractures relate to the formation and the structure in which they are located. The creation of endogenetic fractures relates to the stresses during sedimentation, for example cleating in coals. Exogenetic fractures are formed after sedimentation and lithication, usually from tectonic stresses caused by folding and faulting. Once natural-fracture systems have been classied in both geologic and engineering terms, the next step is to investigate their impact on the reservoir. Classifying Fractured Reservoirs Most, if not all, reservoirs contain fractures. It is the degree to which fractures inuence uid ow through a reservoir that should dictate the level of resources needed to identify, characterize and model fractures. The effects of fractures can change throughout the productive life of the reservoir as pressures and uid types change during primary- and secondary-recovery stages. Moreover, fractures dont always conduct uid; they are often barriers to ow. Fractured reservoirs are classied based on the interaction between the relative porosity and permeability contributions from both the fracture and matrix systems (next page, bottom). 20 In Type 1 reservoirs, fractures provide both the porosity and permeability elements. Type 2 reservoirs have low porosity and low permeability in the matrix, and fractures provide the essential permeability for productivity. Type 3 reservoirs have high porosity and may produce without fractures, so fractures in these reservoirs provide added permeability. Type M reservoirs have high matrix porosity and permeability, so open fractures can enhance permeability, but natural 8 Oileld Review > Example of low-angle nonsystematic fractures in shales. FMI Fullbore Formation MicroImager images clearly show both fracturing (blue sinusoids in Track 3 and tadpoles in Track 4) and formation bedding (green sinusoids and tadpoles). Track 1 displays caliper, borehole orientation and gamma ray data. Tracks 2 and 3 show the FMI static and dynamic images, respectively. Dip tadpoles are presented in Track 4. X70.2 X70.4 X70.6 X70.8 X71.0 X71.2 X71.4 X71.6 X71.8 X72.0 0 120 240 360 FMI Static Image Resistive Conductive Orientation North 0 120 240 360 FMI Dynamic Image Resistive Conductive Orientation North gAPI 0 200 Gamma Ray Bit Size Caliper 2 mm 125 375 mm 125 375 mm 125 375 Caliper 1 deg 0 90 Bedding True Dip deg 0 90 Fracture True Dip deg 0 90 Borehole Drift D e p t h ,
f t 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 8 Summer 2006 9 fractures often complicate uid ow in these reservoirs by forming barriers. Fractures add no signicant additional porosity and permeability to Type 4 reservoirs, but instead are usually barriers to ow. Another reservoir class, Type G, has been created for unconventional fractured gas reservoirs, such as CBM, and fractured gas- condensate reservoirs. Most Type G reservoirs fall within or near the Type 2 reservoir classication. Before NFR classication can be done in any meaningful way, both natural-fracture and matrix systems within a reservoir must be understood, along with the complex ow interaction between those systems. Many factors affect uid ow within a NFR, including present-day stress orientation, natural-fracture directions, whether the fractures are mineral- lled or open, reservoir uid properties and phases, and the production and injection history of the eld. While many of these factors cannot be controlled, some problems can be mitigated. Field-development strategies can therefore be tailored to the natural-fracture systems to optimize production and recovery. The sooner this knowledge is acquired, the more prepared asset teams will be to make important eld- management decisions early in eld development. Evaluating Fractures and Fields There are many different ways to characterize natural fractures and to evaluate their role in reservoir exploitation. Dynamic methods seek to characterize the effects of fractures by measuring or directly describing the movement of uids through fractures and matrix. These dynamic methods include medium-scale interval, pressure-transient testing, which provides information on fractures and fracture-related ow, and estimates of fracture conductivity. 21 These tests can be obtained with the MDT Modular Formation Dynamics Tester. Another medium- to large-scale dynamic method uses injected tracers and water-composition analysis to determine direct communication attributed to fractures between zones and between wells. 19. Stylolites are wave-like or tooth-like, serrated, interlocking surfaces, most commonly seen in carbonate and quartz- rich rocks, that contain concentrated insoluble residue such as clay minerals and iron oxides. Stylolites are thought to form by pressure solution, a dissolution process that reduces pore space under pressure during diagenesis. For more on stylolites: Nelson, reference 1: 163185. 20. Nelson, reference 1: 101124. 21. Jackson RR, Xian C, Carnegie A, Gauthier P and Brooks AD: Application of Interval Pressure Transient Testing with Downhole Fluid Analysis for Characterising Permeability Distributions, In-Situ Flow Fractions and Water Cut, paper SPE 92208, presented at the SPE International Petroleum Conference, Puebla, Mexico, November 79, 2004. > Cross section of a stylolite. Stylolites are diagenetic features commonly found in low-permeability carbonate rocks. They form as irregular surfaces between two layers and are generally thought to be the result of pressure solution under a state of differential stress. Stylolites normally inhibit subsurface fluid flow, but are often associated with small fractures called tension gashes, which sometimes appear permeable on core tests. inch 0 0 1 1 cm > Naturally fractured reservoir classification system. Type 1 reservoirs, with fractures providing both primary porosity and primary permeability, typically have large drainage areas per well, and require fewer wells for development. These reservoirs show high initial production rates. They are also subject to rapid production decline, early water breakthrough and difficulties in determining reserves. Type 2 reservoirs can have surprisingly good initial production rates for a low-permeability matrix but can have difficulties during secondary recovery if the communication between the fracture and the matrix is poor. Type 3 reservoirs are typically more continuous and have good sustained production rates but can have complex directional permeability relationships, leading to difficulties during the secondary-recovery phase. Type M reservoirs have impressive matrix qualities but are sometimes compartmentalized, causing them to underperform compared with early producibility estimates, and making secondary-recovery effectiveness variable within the same field. Type 4 reservoirs would plot near the origin because the fracture contribution to permeability in Type 4 reservoirs is negative. (Adapted from Nelson, reference 1: 102.) T o t a l
p e r m e a b i l i t y ,
% Total porosity, % Increasing natural-fracture influence (decreasing matrix influence) 100% matrix permeability 100% fracture permeability 100% matrix porosity 100% fracture porosity Type 3 Type M (matrix only) Type 2 Type Type 4 1 Type G 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 9 Geometric methods measure specic attri - butes to identify and characterize natural fractures and assess their potential impact on production or injection. While traditional logging measurements, such as caliper and microresis - tivity logs, can allude to the presence of natural fractures, they are generally not quantitative. Today, various technologies have been developed to address NFRs. The most common small-scale, log-based fracture-evaluation techniques use ultrasonic and resistivity borehole imaging technologies that can be deployed by wireline or LWD methods. While the resolution of wireline-conveyed electrical borehole imaging tools is exceptional, the most detailed way to assess NFRs is by acquiring fullbore cores across intervals of interest. 22 Having access to fullbore core allows geologists and petrophysicists to examine specic properties that inuence a fractures ability to conduct uidsfor example, the presence of in- lling minerals. Another extremely valuable use of core data is to provide a ground truth from which to calibrate other fracture-analysis methods. However, fullbore coring can be expensive and poor core recovery can be a problem in highly fractured rock. Also, coring- induced fractures can be difcult to distinguish from unmineralized natural fractures. 23 Despite the difculties, there are now innovative ways to characterize NFRs using advanced technologies and processing techniques. The fractured granite basement rocks of the Cuu Long basin, offshore Vietnam, are mostly Type 1 reservoirsboth porosity and perme ability in the basement rock are provided by natural fractures (left). 24 However, in the fractured zones surrounding faults, secondary porosity has been documented because hydrothermal uids dissolve feldspars in the matrix. The result is a hybrid Type 2/Type 1 NFR. Since rst production in the early 1990s, common methods for measuring permeability the most daunting property to ascertain in these fractured basement reservoirswere performing well tests or acquiring and testing core. Well-test analysis of fractured reservoirs requires numerous assumptions that can lead to errors, while core analysis is typically pessimistic because the most highly fractured reservoir intervals often are not recovered and analyzed. Even though Cuu Long reservoirs rely solely on fractures to produce, their productivity can be astonishingsome individual wells can produce more than 20,000 bbl/d [3,180 m 3 /d] of oil. A series of geologic episodes, including an extensional phase during rifting, which created the basin, followed by two major phases of compression, has led to a complex but prolic natural-fracture network that can be divided into three fracture classessolution-enhanced and unenhanced bounding fractures, straight- walled fractures and discrete fractures (next page, left). 25 When not lled with clays, calcite and zeolites, the bounding network of fractures forms the main conduits for uid transmission and provides important storage volume for the basement reservoirs. 26 Some of the bounding fractures are enormous, measuring more than 1.5 m [4.9 ft] in fracture width. On the other hand, 10 Oileld Review 25. Bounding fractures are dened as fractures on which other fractures terminate. 26. Zeolites are microporous crystalline solids with well- dened structures. Generally, they contain silicon, aluminum and oxygen in their framework, and cations, and water or other molecules within their pores. From http://www.bza.org/zeolites.html (accessed April 30, 2006). 27. To compute fracture apertures, shallow-resistivity data are needed to calibrate, or scale, the FMI or Formation MicroScanner tool response. For more on the technique: Luthi SM and Souhaite P: Fracture Aperture from Electrical Borehole Scans, Geophysics 55, no. 7 (1992): 821833. 22. Lorenz JC and Hill R: Measurement and Analysis of Fractures in Core, in Schmoker JW, Coalson EB and Brown CA (eds): Geophysical Studies Relevant to Horizontal Drilling: Examples from North America. Denver: Rocky Mountain Association of Geologists (1994): 4757. 23. Lorenz JC, Warpinski NR and Teufel LW: Natural Fracture Characteristics and Effects, The Leading Edge 15, no. 8 (August 1996): 909911. 24. Li B, Guttormsen J, Hoi TV and Duc NV: Characterizing Permeability for the Fractured Basement Reservoirs, paper SPE 88478, presented at the SPE Asia Pacic Oil and Gas Conference and Exhibition, Perth, Australia, October 1820, 2004. > Location of the Cuu Long basin, offshore Vietnam. Fracture swarms in a granite outcrop along Long Hai Beach, Vietnam, are an offshore analog of the field (photograph). Fracture swarms run parallel to the beach for 300 to 400 m [984 to 1,312 ft]. The relative lack of published stress data makes it even more important to acquire usable stress data when possible (bottom right). (Stress map insert is from the World Stress Map Project, http://www-wsm.physik.uni-karlsruhe.de/pub/casmo/content_frames/ stress_maps_frame.html, used with permission.) 200 0 miles 0 200 km Hanoi Da Nang Ho Chi Minh City C u u L o n g B a s in V I E T N A M Vietnam S o u t h
C h i n a
S e a 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 10 Summer 2006 11 the majority of discrete fractures are relatively short, terminate at the bounding fractures, contribute the majority of the storage capacity to the bounding networks, and maintain apertures that mostly range from 0.01 to 0.1 mm [0.0004 to 0.004 in.]. In the elds of the Cuu Long basin, perme - ability is the driving factor for well productivity. Using FMI image data, geoscientists from Schlumberger, Cuu Long Joint Operating Company (JOC) and VietSovPetro developed a method to consistently calculate reservoir permeability and calibrate it to core analysis, well- testing results and production-log data. First, fracture interconnectivity is assessed using the image data and the BorTex texture classication tool in the Schlumberger GeoFrame integrated reservoir characterization system platform. This processing essentially maps out the conductive anomalies within the resistive granite matrix on the borehole image and computes a relative permeability indicator (RPI). In another processing step, fracture apertures and fracture density are calculated for hand-picked fractures on the FMI resistivity images. 27 These outputs, along with a calibration constant, are used to calculate fracture permeability (K f ). In Type 1 reservoirs, K f should equal reservoir permeability (K r ) for the same investigated volume. The RPI can then be scaled to K r to provide a continuous assessment of permeability. The limited amount of core taken in a zone of relatively low permeability was used to calibrate K r (above right). This image-based interpretation technique has been successful on numerous wells across the Cuu Long basin. For example, on one well, 300 m [984 ft] of the granite basement rock was penetrated at a top depth of around 3,900 m [12,800 ft]. A standard openhole-logging suite was acquired along with FMI images and only 3 m [9.8 ft] of fullbore core. After initial production, dynamic fracture-characterization methods were employed on two occasionsshortly after the well was completed and again after 17 months of productionand included well testing and production logging. > Fracture-classification system used in the Cuu Long basin. The FMI image (left) shows the two main fracture types. The fracture flow system relationships for each type are described for the discrete fracture system (top right) and the solution-enhanced system (bottom right). X,Y84 Fracture Flow-System Relationships Discrete Fracture System (Secondary) Tectonic only Low permeability Short length Short height stress Fine aperturesubject to principal Highly tortuous flow paths Secondary production conduits (behave like a matrix-porosity system) X,Y85 X,Y86 X,Y87 X,Y88 Solution-Enhanced System (Primary) Tectonic modified by hydrothermal and meteoric processes High permeability Long length Tall height Large aperture Linear to radial flow paths Primary production conduits 0 120 240 360 Orientation North D e p t h ,
m Resistive Conductive FMI Image > Calibration and validation of reservoir permeability (K r ) using 3 m of core data (left). The computed permeability log section on the right shows the high-permeability zones (yellow) that contributed to flow during production logging and well testing. These zones had an average permeability from well testing of 69 mD. The continuous computed K r across the same intervals averaged 92 mD. 0.1 1,000 mD Core Permeability Permeability 0.1 1,000 mD Core Permeability 0.1 1,000 mD 0.1 1,000 mD 1 m E s t i m a t e d
a v e r a g e
p e r m e a b i l i t y
f o r
t h e
c o n t r i b u t i n g
z o n e s
i s
6 9
m D . Permeability Permeability Fracture Permeability 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 11 A correlation between the calculated perme - abilities and actual reservoir performance was very good (below). Initially, oil owed from three zones as demonstrated by the production log, but there were several high-permeability zones that did not contribute. Experts at Cuu Long JOC and VietSovPetro suspected that the lack of contribution was caused by partial formation damage, since mud losses were recorded during drilling, for example from X,090 to X,100 m. Reassuringly, after 17 months of production, other zones began to contribute to production. Over time, the damaged zones cleaned up with assistance from the pressure drop in the wellbore. In addition, the water cut had increased since the start of production. This technique has helped to minimize the complicating effects that resistive fracture-lling minerals have on fracture characterization in the elds of the Cuu Long basin. However, conductive minerals in the fractures, found mainly in weathered zones at the top of the granite, still pose a dilemma because resistivity-based imaging tools cannot differentiate between conductive minerals and conductive drilling uid. In these zones, special attention is paid to corroborative datamud-loss records, gas shows and log data from the MDT or CMR Combinable Magnetic Resonance tools. Importantly, this fracture- characterization technique provides a detailed, depth-continuous permeability output that can help asset teams with individual well stimulation and completion and injection designs, and can be upscaled to reservoir models across an entire eld. 12 Oileld Review > Integrated fracture permeability analysis showing a comparison of computed permeability to production logging and well-testing results. Standard openhole-log data are displayed in Tracks 1 and 2, FMI images are shown in Track 3, fracture apertures calculated from the FMI data are presented in Track 4, K f and RPI are shown in Track 5, and K r with core-calibration points is displayed in Track 6. The yellow box in the Depth Track indicates the location of significant mud losses during drilling. Tracks 7 and 8 include the interpreted production-log results in the first well-testing job shortly after the well was drilled. Track 9 presents the interpreted production-log rate, showing zones that contributed water (blue) and oil (red) from the second well- testing job, performed after the well had been on production for 17 months. 0 120 240 360 FMI Dynamic Image Res. Cond. Orientation North 0.45 -0.15 2 20,000 m 3 3 /m ohm.m Neutron Porosity D e p t h ,
m X,000 X,100 1.95 2.95 2 20,000 ohm.m Bulk Density Shallow Laterolog g/cm 3 0.1 1,000 mD Core Permeability Permeability 0.009 0.5 Relative Permeability Indicator 1 10,000 mD Fracture Permeability 0.00001 0.1 Fracture Aperture cm Flow Rate 0 4,000 bbl/d Oil Rate, bbl/d Water Rate, bbl/d Oil Rate, bbl/d 1,058 784 169 483 132 232 860 452 119 301 106 854 276 Deep Laterolog 6 16 Caliper in. 0 200 Gamma Ray gAPI Permeability 0.1 1,000 mD 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 12 Summer 2006 13 Fractures in the Rocky Mountains Hydrocarbon production from low-porosity, low- permeability, hard-rock reservoirs depends on successfully connecting open fracture networks to the wellbore. Because the matrix quality is generally low, the surface area exposed to the wellbore along fracture planes often must be increased for required production volumes. This is accomplished by performing hydraulic fracture stimulation. Open natural fractures contribute to production but can also cause problems during drilling, cementing, completion and stimulation operations. Therefore, it is essential to identify fractured intervals for cement- and stimulation- staging considerations. A powerful combination of high-resolution borehole imaging and innovative acoustic measurements from the Sonic Scanner acoustic scanning platform adds dynamic elements to detailed fracture analysis from wireline tools. 28 Schlumberger geoscientists and petrophysicists in the Rocky Mountain region, USA, use Stoneley and dipole exural-wave data from the Sonic Scanner tool and FMI image data to clearly identify formation bedding, sedimentary features and fractures. 29 The improved low-frequency Stoneley responsedown to 300 Hzof the Sonic Scanner tool enables the detection of high- angle to vertical fractures. Also, using an attenuation technique called normalized differential energies (NDE) makes it possible to differentiate natural fractures from drilling- induced fractures, even when they are oriented in the same directionusually parallel to the present-day maximum horizontal-stress direction. However, when the stress-related anisotropy direction differs only slightly from the fracture- induced anisotropy direction, the new tool is still able to differentiate the two because of the improved ability to resolve small amounts of anisotropynow 2%, versus 5% with the previous technology. Frequency-content and signal-strength variations occur in naturally fractured intervals. Another processing technique called slowness- frequency analysis (SFA) allows the interpre - tation of dipole exural-wave frequency and amplitude data and shows the quality of the estimation of shear slowness from exural-wave dispersion analysis up to several feet into the formation from the wellbore. In the Type 2 reservoirs in the Rocky Mountains, porosities range from 3 to 7% and matrix permeabilities are in the microdarcies. The FMI tool enables the calculation of fracture aperture, fracture porosity, fracture density and fracture trace length at the wellbore. 30 Combining independent fracture-characterization methods from the Sonic Scanner Stoneley-wave and shear-wave analysis with FMI image interpretation shows an unambiguous assessment of fracturing across the interval (above). Armed with this log-based characterization of the fractures, the asset team can better judge the optimal way to cement, complete and stimulate this potentially productive interval. Experts in the Rocky Mountain region have developed a hard-rock completion solution that combines Sonic Scanner data with FMI data to optimize hydraulic fracture design. The solution incorporates natural-fracture characterization including the determination of fracture aperture, fracture permeability and fracture extentand the analysis of maximum and minimum hori - zontal stresses. All this information is captured in the mechanical earth model that is used by stimulation designers to optimize hydraulic fracture design. 28. Arroyo Franco JL, Mercado Ortiz MA, De GS, Renlie L and Williams S: Sonic Investigations In and Around the Borehole, Oileld Review 18, no. 1 (Spring 2006): 1433. 29. Donald A and Bratton T: Advancements in Acoustic Technique for Evaluating Open Natural Fractures, Transactions of the SPWLA 47th Annual Logging Symposium, Veracruz, Mexico, June 47, 2006, paper QQ. 30. Hornby B and Luthi S: An Integrated Interpretation of Fracture Apertures Computed from Electrical Borehole Scans and Reected Stoneley Waves, in Hurst A, Grifths C and Worthington P (eds): Geological Applications of Wireline Logs II, Geological Society Special Publication 65. London: The Geological Society (1992): 185198. > Fracture characterization using Sonic Scanner and FMI data. Borehole anisotropy analysis includes the slowness-frequency analysis (SFA) and slowness-time-coherence (STC) projections for the fast inline and slow inline waveforms. In Track 2, the anisotropy magnitude and direction vary with depth, ranging from more than 16% (red) to less than 2% (blue). The high-anisotropy zones correspond to zones with fractures visible on the FMI image in Track 7. Minimum and maximum offline energy differences are displayed in the Depth Track from the shear-wave anisotropy analysis. Large differences between measured Stoneley slowness and slowness modeled for an elastic, impermeable formation are observed in Track 3. Fracture-aperture computations from the Sonic Scanner reflection and attenuation analysis in Track 4 are compared with fracture apertures calculated on hand-picked fractures in Track 5 from the FMI image in Track 7. Track 6 displays the Stoneley Variable Density log. 300 s/ft 0 T-Based Fast Shear X,100 X,200 X,300 X,400 0 100 0 100 Maximum Energy Minimum Energy D e p t h ,
f t Modeled Stoneley s/ft 250 150 S-Se Washout Caliper in. 4 14 Bit Size in. 4 14 T Stoneley s/ft 250 150 Stoneley Aperture Fracture Width in. 0 0.5 Stoneley Permeability Fracture Trace Length 1/ft 10 0 Fracture Porosity ft 3 /ft 3 0.1 0 Fracture Permeability mD 100,000 10 10 s 0 20,440 Stoneley Variable Density Log 0 120 240 360 Resistive Conductive FMI Image Offline Energy Orientation North -90 90 deg Fast Shear Azimuth 2 4 6 16 0 Anisotropy Flag, % Shear Difference 300 s/ft 0 T-Based Slow Shear 200 % % Time-Based Anisotropy 0 300 s/ft 0 Compressional T 300 s/ft 0 Stoneley T 0 T-Based Anisotropy 200 Azimuth Uncertainty 58732schD04R1.qxp:58732schD04R1 10/10/06 9:51 PM Page 13 Coalbed Methane Reservoirs There may be no other NFR as difficult to stimulate as a coalbed methane (CBM) reservoir, an unconventional but growing source of methane. Beginning with its deposition as peat, coal is a unique reservoir rock. To be productive, coalbed reservoirs require natural fractures. Vertical natural fractures in coal are called cleats, and these form during coalifi - cation. Systematic coal cleats are classified geometrically with the primary, more continuous fracture set called face cleats and the secondary, less continuous fracture set called butt cleats (below). Coal fractures can also be classied genetically. Endogenetic fractures, or classic cleats, are created under tension as the coal matrix shrinks because of dewatering and devolatilization during coalication. These cleat sets are orthogonal and nearly always perpen - dicular to bedding. In contrast, exogenetic fractures form due to tectonism, and regional stress elds dictate their orientation. Shear fractures also are observed in some coals. Cleats are the primary permeability mechanism in virtually all CBM reservoirs, so understanding cleating and natural fracturing in coals is critical during all facets of CBM reservoir development. Methane is stored in coal by adsorption, a process by which the individual gas molecules are bound by weak electrical forces to the solid organic molecules that make up the coal. Coals ability to store methane largely reduces 14 Oileld Review > Examples of Canadian coals on FMI images and outcrops. The FMI image (top left) and a photograph from a representative outcrop (bottom left) of the Alberta Plains coal show both face and butt cleats. Shear fractures, face cleats and butt cleats are shown on both the FMI image (top right) and the outcrop photograph (bottom right) of the Alberta Foothills coal. Interestingly, shear fractures usually degrade coal permeability. Alberta Foothills Coal gAPI 0 150 Gamma Ray mm 125 375 Bit Size Caliper mm 125 375 0 90 0 120 240 360 deg Bedding True Dip Face Cleat deg 0 90 FMI Dynamic Image Resistive Conductive Orientation North M e a s u r e d
d e p t h ,
m Face cleat Shear fractures X,X20 X,X21 X,X22 Face cleat Butt cleat Shear fracture Bedding Foothills Coal Alberta Plains Coal gAPI 0 150 Gamma Ray mm 375 Bit Size Caliper mm 375 deg 0 90 Bedding True Dip Face Cleat 0 120 240 360 deg 0 90 M e a s u r e d
d e p t h ,
m Orientation North FMI Dynamic Image Resistive Conductive X,X59 X,X60 125 125 Face cleat Butt cleat Plains Coal 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 14 Summer 2006 15 the need for conventional reservoir-trapping mechanisms, making coals gas contentwhich increases with increasing coal rankand the degree of cleating or natural fracturing the overriding considera tions when assessing an area for CBM production potential. 31 This storing ability gives coals unique early- time production behavior that is related to desorption, not pressure depletion. Coals may contain water or gas, or both, in the cleat and natural-fracture systems, in addition to gas sorbed onto the internal surface of the coal matrix. Any water present in the cleat system must be produced to reduce the reservoir pressure in the cleat system before signicant volumes of gas can be produced. Dewatering increases the permeability to gas within the cleats and fractures, and causes the gas in the matrix to desorb, diffuse through the matrix and move into the cleat system, resulting in CBM production proles that are unique by comparison with other fractured reservoirs. In most CBM reservoirs, water production is initially high. As the water moves out of the cleats and fractures, gas saturation and production increase and water production decreases. The speed at which the reservoir dewaters depends on several factors, including original gas and water saturations, cleat porosity, relative and absolute permeability of the coal, and well spacing. When permeability to gas eventually stabilizes, the coal is considered dewatered and gas production peaks. From this point, both water and gas production slowly decline, with gas being the dominant produced uid. In just a few years of development, CBM gas production in Alberta, Canada, has surpassed 300 million ft 3 /d [8.50 million m 3 /d]. Most of this production comes from the Horseshoe Canyon and Mannville coal zones, and a small percentage less than 1%comes from the Ardley coals in the Upper Cretaceous Scollard formation (above). The less-exploited Ardley coals, however, are a signicant potential CBM resource, exceeding 40 trillion ft 3 [1.13 trillion m 3 ]. Burlington Resources, now ConocoPhillips, has investigated the Ardley coals using the FMI tool. 32 In two wells, the borehole images have allowed geoscientists at ConocoPhillips and Schlumberger to determine the present-day stress regime from drilling-induced fractures, which are oriented northeast to southwest, in the direction of maximum horizontal stress. This direction is consistent with previous assessments. 33 The FMI images have also provided insight into the nature and direction of > Maps showing the distribution of Alberta coals (left), thickness of the Ardley coal (top right) and published stress data (bottom right). The map on the left shows the area of the Ardley coal play (red) where the Ardley coal thickness is greater than 12 m [39.4 ft]. (Stress map insert is from the World Stress Map Project, http://www-wsm.physik.uni-karlsruhe.de/pub/casmo/content_frames/ stress_maps_frame.html, used with permission.) Alberta Alberta Calgary Edmonton km miles 0 0 200 200 Edmonton km miles 0 0 100 100 Ardley coal thickness 0 to 6 m 6 to 12 m >18 m 12 to 18 m 31. Anderson J, Simpson M, Basinski P, Beaton A, Boyer C, Bulat D, Ray S, Reinheimer D, Schlachter G, Colson L, Olsen T, John Z, Khan R, Low N, Ryan B and Schoderbek D: Producing Natural Gas from Coal, Oileld Review 15, no. 3 (Autumn 2003): 831. 32. Schoderbek D and Ray S: Reservoir Characterization of Ardley Coals, Scollard Formation, Alberta: Borehole Image Interpretation, presented at the AAPG Annual Meeting, Calgary, June 1619, 2005. 33. Bell JS, Price PR and McLellan PJ: In-Situ Stress in the Western Canada Sedimentary Basin, in Mossop GD and Shetson I (compilers): Geological Atlas of the Western Canada Sedimentary Basin. Calgary: Canadian Society of Petroleum Geologists and Alberta Research Council (1994): 439446. 58732schD04R1.qxp:58732schD04R1 10/10/06 9:52 PM Page 15 cleating within the Ardley coalsthe Val DOr, the Arbour, the Silkstone and the Mynheer zones (above). Interpretation of FMI images suggested that within the Scollard formation, the Silkstone coal had the most productive potential and the Arbour coal had some potential. ConocoPhillips integrated public and proprietary cleat-orientation information from mines and outcrops. In addition, geoscientists performed a detailed examination of six unoriented conventional cores that were cut from the Ardley coals ve to ten years earlier. To supplement the regional study of Ardley coal cleating, these cores had to be oriented after the fact, years after their acquisition. To accomplish this, ConocoPhillips utilized a technique developed by Applied Paleomagnetics called paleomagnetic core orientation, which requires that whole cores be reassembled and that plugs cut from the core be selectively demagnetized. 34 The cores are oriented using the secondary magnetization of magnetite found in nearly all rocks. This magnetization points to present-day geographical north and represents the average geomagnetic eld over the past 780,000 years, which is the time since the last geomagnetic polarity reversal. Once the north direction on the 16 Oileld Review 34. Bleakly DC, Van Alstine DR and Packer DR: Core Orientation 1: Controlling Errors Minimizes Risk and Cost in Core Orientation, Oil and Gas Journal 83, no. 48 (December 2, 1985): 103109. Bleakly DC, Van Alstine DR and Packer DR: Core Orientation 2: How to Evaluate Orientation Data, Quality Control, Oil and Gas Journal 83, no. 49 (December 9, 1985): 4654. Hamilton WD, Van Alstine DR and Butterworth JE: A Fracture-Orientation Comparison Between Core- Based and Borehole-Imaging Techniques: Paleomagnetic, Electronic Multishot, and FMI, presented at the AAPG Annual Convention, San Diego, California, May 1922, 1996. 35. Barkved O, Bartman B, Compani B, Gaiser J, Van Dok R, Kristiansen P, Probert T and Thompson M: > Images of Ardley coals. The FMI tool successfully identifies cleating, or lack of cleating, in the four Ardley coal zones. The FMI static image of the Val DOr coal appears very bright (top left), indicating a high degree of mineralization. The Arbour coal image (top right) indicates no large cleats, while the Silkstone coal image (bottom left) shows abundant face cleats, mostly striking northeast to southwest. The Mynheer coal is dominated by shale interbeds (bottom right). Mineralized Val DOr X52.0 X52.5 X53.0 Well-cleated Silkstone Y12.5 Y13.0 Y13.5 Y14.0 Small-scale cleating Arbour X86.5 X87.0 X87.5 Shaly Mynheer Y31.5 Y32.5 Y32.0 gAPI 0 200 Gamma Ray mm 375 125 mm 375 125 mm 375 125 mm 375 125 mm 375 125 mm 375 125 Bit Size Caliper 2 Caliper 1 deg 0 90 Borehole Drift 0 360 deg 0 90 Bedding True Dip deg 0 90 Fracture True Dip D e p t h ,
m 0 120 240 360 FMI Dynamic Image Resistive Conductive Orientation North FMI Static Image Res. Cond. Orientation North gAPI 0 200 Gamma Ray Bit Size Caliper 2 Caliper 1 deg 0 90 Borehole Drift 0 360 deg 0 90 Bedding True Dip deg 0 90 Fracture True Dip D e p t h ,
m 0 120 240 360 FMI Dynamic Image Resistive Conductive Orientation North FMI Static Image Res. Cond. Orientation North The Many Facets of Multicomponent Seismic Data, Oileld Review 16, no. 2 (Summer 2004): 4256. Caldwell J, Christie P, Engelmark F, McHugo S, zdemir H, Kristiansen P and MacLeod M: Shear Waves Shine Brightly, Oileld Review 11, no. 1 (Spring 1999): 215. 36. Fracture intensity is a qualitative description of the degree of natural fracturing that is usually derived from seismic traveltime attributes. 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 16 Summer 2006 17 reassembled core is determined, the results from the detailed analysis can then be oriented, yielding orientation data comparable to outcrop and mine studies and FMI image analyses (right). All sources of data indicated that a dominant northeast-to-southwest face-cleat system might be open because of its favorable alignment with the present-day maximum horizontal stress. The butt-cleat system in the Ardley coals is much less persistent and is aligned less favorably with respect to present-day stresses. The lack of butt cleats in Ardley coals is in contrast to Horseshoe Canyon and Mannville coals. Horizontal wells drilled perpendicular to the face-cleat system may require hydraulic fracturing of multiple intervals within the horizontal section to effectively stimulate the coals and optimize production potential. A more effective stimulation promotes the dewatering of the cleat systems and speeds up gas desorption. The challenging permeability scenario will also inuence well-design considerations, such as drilling updip to maximize drainage. Exploration for coalbed methane in the Ardley coals of the Scollard formation is in its infancy. ConocoPhillips plans to integrate the results of this cleat study with hydrogeological and structural interpretations to develop its future exploration strategy. A Seismic Net to Capture Fractures The ability to characterize fracture systems in the early eld-development stage reduces economic risk because it enables asset teams to determine optimal horizontal well directions to maximize production and recovery. So far, most of the discussion on fracture characterization has dealt with the investigation of fractures using relatively high-resolution techniques as compared to seismic methods, which use wavelengths up to 100 m [328 ft] to detect the presence of natural fractures using azimuthal anisotropy analysis. 35 These techniques do not detect individual faults or fractures, but rather exploit the average response across a large volume of rock. For example, measuring traveltime differences between the fast and slow shear waves, together with the polarization direction of the fast shear wave, helps to infer the fracture intensity and fracture orientation, respectively. 36 Seismic fracture-characterization methods include velocity anisotropy determina - tion, azimuthal amplitude variation with offset, and normal moveout (NMO) variation with azimuth (right). > Determining major cleat directions in the Ardley coals. Paleomagnetic core orientation was used to supplement the ConocoPhillips Ardley coal database. Rose diagrams showing the cleat strike data from the rotated core analysis are displayed on the left side of the map, while the rose diagrams from the FMI interpretation are shown to the right of the map. Overall, the data support a northeast to southwest face-cleat strike. Coal cleats Natural extension fracture Natural shear fracture High-angle induced fracture Low-angle induced fracture Cleats from FMI tool Rose Diagram Symbols Well 6 Well 1 Well 3 Well 4 Well 2 Well 5 FMI cleats 0 30 60 90 120 150 180 210 240 270 300 330 5% 10% 15% 5% 10% 15% 20% 20% Ardley coal thickness 6 to 12 m >18 m 12 to 18 m 5 2 4 3 6 1 0 30 60 90 120 150 180 210 240 270 300 330 2% 4% 6% 8% 10% 2% 4% 6% 8% 10% > Seismic azimuthal anisotropy methods. The diagrams show land and marine seismic acquisition methods used to detect fracture-induced anisotropy. The fracture diagram (top left) shows vertical fractures striking north-south in the example, causing shear-wave splitting that helps determine the fast-shear direction (north-south red polarization vectors) and the slow-shear direction (east-west blue polarization vectors). The sinusoid shows how anisotropy can be determined from compressional and shear velocity variations with azimuth (top right). The land seismic diagram (bottom left) shows the rays for common midpoint gathers from two source-receiver directions. The seabed seismic diagram (bottom right) demonstrates the effects of seismic anisotropy by showing two rays: a south-going fast ray from a source position to the north of the seabed receiver cable; and a west-going slow ray from an east source position above the seabed receiver cable. In 3D surveys, all azimuth directions are interrogated. V e l o c i t y North South East West Azimuth NE SW F a s t s h e a r, N S S lo w sh e a r, W E F a s t F a s t N S Slow E W F a s t N S Natural fractures Natural fractures Seabed receiver cable S lo w Slow E W 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 17 Seismic investigations of NFRs include those from multioffset, multiazimuth vertical seismic proles (VSPs). Walkaway- and walkaround-VSP techniques permit velocity anisotropy and amplitude variation with offset and azimuth (AVOA) analyses at higher resolutions than with surface seismic methods and can be used to calibrate surface seismic results. Integrating all available data to optimize the VSP conguration is important for extracting high-quality anisotropy information. This information can then be used to design 3D surface seismic surveys to cover areas remote from well control. 37 Through the years, geophysicists have noted that compressional- (P-) wave velocities exhibited azimuthal variations when processing some 3D seismic surveys, especially those in areas of high tectonic stress. 38 The fast P-wave direction aligns with the maximum compressional stress direction, parallel to natural fractures resulting from the stress. In this simple scenario, the slow P-wave direction would be aligned perpendicular to the fracture strike, and the fracture-lling uid would affect the velocity. Azimuthal variations in other seismic attributes, such as reection amplitudes, have also been observed and exploited to determine fracture azimuth. The advantage of examining amplitude variations is that it detects local azimuthal variations in contrast to velocity-based techniques, which respond to the accumulating effects of overlying strata. 39 Consequently, AVOA analysis is a higher vertical resolution depiction of a NFR than that obtained with velocity-based methods. Reection amplitude, or reectivity, depends on the effective elastic properties of the fractured rock at the seismic scale. Because both P- and shear (S-) velocities change with azimuth in a fractured medium, an AVO response will be inuenced by fracture properties, including fracture azimuth. While AVOA processing and interpretation are fairly simple where there is a single alignment in an otherwise homogeneous medium, multiple fracture directions for example near faultsand additional sources of anisotropy may signicantly complicate the analysis. 40 Another approach examines the azimuthal variation of P-wave normal moveout (NMO) velocity. 41 A minimum of three azimuthal measure- ments is required to construct an ellipse in the horizontal plane that shows NMO velocities in all azimuthal directions. Although most seismic fracture-analysis methods assume a simple geometryhorizontal beds and vertical fracturesthe NMO technique allows some further assessment where beds are dipping and where natural fractures may not be vertical. However, this technique also suffers from velocity- related degradation of vertical resolution. A carbonate reservoir study in a eld in southwest Venezuela compared seismic-based fracture-orientation results with fracture orientations based on FMI images. 42 Different seismic data types were used in the study, including 2D three-component (3C) P- and S-wave data, and 3D P-wave data. The study found that most of the results from the rotation analysis of the converted-wave 2D-3C data, and the AVOA and NMO analyses of the 2D and 3D P-wave data determined the general direction of the regional maximum horizontal stress. However, results varied between the different methods because of local structural variations. With the 3D P-wave data, the AVOA technique appeared more robust than the NMO analysis. The Venezuelan study also found that there were quantiable advantages to acquiring land 3C data, including the ability to estimate fracture orientation and fracture density, or intensity. Acquiring multicomponent seismic data in a marine setting requires sophisticated four- component (4C) seabed acquisition equipment. 43 Marine seismic studies have been successful in identifying anisotropy direction and magnitude at the specic target horizon by effectively removing the inuence of the overburden in a layer-stripping approach. 44 Passive seismic methods that detect the reservoir response to production or injection can also be thought of as dynamic fracture- and fault- characterization techniques. Natural fractures and faults emit microseismic eventsmostly due to shear readjustmentsin response to changes in effective stress following eld production and injection, and especially during hydraulic fracture stimulation operations. 45 Sensitive seismic sensors positioned in nearby wellbores detect these acoustic emissions, which in this method serve as the seismic source (above). Special processing methods estimate event locations, producing a continuous time- based record of production- or injection-induced activity. Seismic methods represent medium- to large-scale fracture detection and character- ization methods, and therefore have implications in the effort to model the interwell volume of these complex reservoirs. 18 Oileld Review > Tracking acoustic emissions induced by fluid production or injection. Producing from or injecting into rocks in the subsurface changes net stress in fractures and faults, inducing small shear events that emit acoustic signals (red stars). These emissions can be recorded in nearby monitoring wells that contain sensitive multicomponent seismic recording equipment. Special localization processing creates a record of the events in space and time. These acoustic emissions are located in 3D space and help identify fracture and fault directions. Monitoring well Natural fractures Production or injection well Production or injection well 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 18 Summer 2006 19 Regardless of the technique, information cultivated from seismic data contributes to reservoir modeling that guides primary- and secondary-recovery planning. However, in many elds, wells from which to draw detailed fracture information are too few and too widely spaced to populate the model volume. Geologists gather detailed fracture dataorientation and possibly spacingfrom analog outcrops. However, this process rarely captures a comprehensive descrip - tion of the fracture network for modeling purposes and sometimes overestimates fracture intensity. Geoscientists at Hydro and Schlumberger in Norway have developed a way to capture the detailed quantitative information needed to make NFR models from outcrop analogs. This method uses a combination of high-resolution optical photography, radar technologies and an automatic surface-extraction technique now widely used for mapping faults in 3D seismic datasets. 46 Hydro and Schlumberger experts have tested this new technique using a well-studied NFR outcrop analog in the Guadalupe Mountains, New Mexico, USA. For several years, Hydro, together with the University of Texas at Dallas, has been using detailed 3D photorealistic models for high- resolution mapping of outcrop analogs. 47 Photorealistic models are derived from the mapping of high-resolution 2D photographs onto 3D outcrop scans using light detection and ranging (LIDAR) technology. 48 LIDAR equipment transmits laser lightvisible electromagnetic radiationto a target and receives back the reected signal for analysis to determine certain properties of the target. The most common type of LIDAR is used for precise range nding accurate to 2 mm [0.08 in.]and the returned radiation intensity can help dene other characteristics of the target. Digitizing sufcient detail of sedimentary architecture from photorealistic models for the building of reservoir models is a straightforward process. However, manual digitizing and analysis of fractures from these datasets are impractical, because several hundred thousand to millions of fractures are commonly present. The new automated approach to outcrop mapping is organized to take advantage of the 3D directional information inherent in LIDAR data and couple it with the detailed information within high- resolution 2D image data. To achieve this, the LIDAR and photographic data are rst analyzed separately. Because the outcrops naturally weather along fractures, fault planes and bedding, the major fracture sets and bed boundaries are captured by vector analysis of the LIDAR data (above left). The orientations of target 37. Peralta S, Barrientos C and Arroyo JL: The Specialized Use of the VSP to Dene Fracture Orientation and to Help in a Multicomponent Survey Design, Transactions of the SPWLA 47th Annual Logging Symposium, Veracruz, Mexico, June 47, 2006, paper SS. Leaney WS, Sayers CM and Miller DE: Analysis of Multiazimuthal VSP Data for Anisotropy and AVO, Geophysics 64, no. 4 (July-August 1999): 11721180. 38. Corrigan D, Withers R, Darnall J and Skopinski T: Fracture Mapping from Azimuthal Velocity Analysis Using 3D Surface Seismic Data, Expanded Abstracts, SEG International Exposition and 66th Annual Meeting, Denver (November 1015, 1996): 18341837. 39. Hall SA and Kendall JM: Constraining the Interpretation of AVOA for Fracture Characterization, in Ikelle L and Gangi A (eds): Anisotropy 2000: Fractures, Converted Waves and Case Studies. Tulsa: The Society of Exploration Geophysicists (2000): 107144. 40. Sayers CM: Misalignment of the Orientation of Fractures and the Principal Axes for P and S Waves in Rocks Containing Non-Orthogonal Fracture Sets, Geophysical Journal International 133, no. 2 (May 1998): 459466. Sayers CM and Dean S: Azimuth-Dependent AVO in Reservoirs Containing Non-Orthogonal Fracture Sets, Geophysical Prospecting 49, no.1 (January 2001): 101106. Williams M and Jenner E: Interpreting Seismic Data in the Presence of Azimuthal Anisotropy; or Azimuthal Anisotropy in the Presence of the Seismic Interpretation, The Leading Edge 21, no. 8 (August 2002): 771774. 41. Grechka V and Tsvankin I: 3-D Description of Normal Moveout in Anisotropic Inhomogeneous Media, Geophysics 63, no. 3 (MayJune 1998): 10791092. For more on normal moveout (NMO): http://www. searchanddiscovery.com/documents/geophysical/liner/ images/liner.pdf (accessed May 7, 2006). 42. Perez MA, Grechka V and Michelena RJ: Fracture Detection in a Carbonate Reservoir Using a Variety of Seismic Methods, Geophysics 64, no. 4 (July August 1999): 12661276. 43. 4C marine seismic data are typically acquired using three orthogonally oriented geophones and a hydrophone within an ocean-bottom sensor. Provided the system is in contact with the seabed, the 3C geophones measure shear waves. The hydrophone measures compressional waves. 44. Gaiser J, Loinger E, Lynn H and Vetri L: Birefringence Analysis at the Emilio Field for Fracture Characterization, First Break 20, no. 8 (August 2002): 505514. 45. Bennet L, La Calvez J, Sarver DR, Tanner K, Birk WS, Water G, Drew J, Michaud G, Primiero P, Eisner L, Jones R, Leslie D Williams MJ, Govenlock J, Klem RC and Tezuka K: The Source for Hydraulic Fracture Characterization, Oileld Review 17, no. 4 (Winter 2005/2006): 4257. 46. Pedersen SI, Randen T, Snneland L and Steen : Automatic 3D Fault Interpretation by Articial Ants, paper Z-99, presented at the 64th EAGE Conference and Exhibition, Florence, Italy, May 2730, 2002. 47. http://www.aapg.org/explorer/2004/06jun/lasers.cfm (accessed July 3, 2006). 48. For more on photorealistic models: http://www. utdallas.edu/~aiken/LASERCLASS/TSPSphotoFINAL.pdf (accessed June 30, 2006). > Using 3D data from light detection and ranging (LIDAR) technology to map major fracture sets. The digital photograph is photorealistically mapped onto a surface derived from the LIDAR data (top). The major fracture patterns are apparent from both image (center) and vector analysis. The Y- component of the surface normal vector (bottom) shows vertical features that are mostly fractures. The height of the vertical outcrop face ranges from about 20 to 25 ft [6.1 to 7.6 m]. Z X Y 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 19 surfaces are described using the three directional components of the normal vector. Radiation intensity is then corrected for both the distance to the LIDAR apparatus and the angle of the outcrop surface. A 3D LIDAR model grid is created and populated with the directional and intensity data. The corrected LIDAR intensity and directional- component data can then be partitioned into value ranges for mapping and analysis. Although there is good detail in the LIDAR data, an even higher level of information is contained in the photographs (above). However, before an automated structural interpretation of the photographic data is accomplished, the digital image must be ltered for noiseanything in the image that does not represent part of the rock exposure, such as vegetation or scree. Next, an attribute or combination of attributes is selected and the Automated Structural Interpretation process, adapted from what is now used in Petrel software, can begin enhancing surfaces. The process uses an adaptation of the technique developed for fault interpretation in 3D seismic volumes. At rst, a fault or fracture may appear only as a trend within the data, but as signal-to-noise character - istics are improved along the surfaces, a more dened plane is mapped by agents using the principles of swarm intelligence (next page). A large number of process agents are deployed in the data volume, making decisions based on precoded behavior. Like ants, the agents traverse the various surfaces emitting an electronic pheromone along the trail, from which an estimate of the surface orientation is made and stored; in this case fractures and bedding are 20 Oileld Review 49. Ali AHA, Brown T, Delgado R, Lee D, Plumb D, Smirnov N, Marsden R, Prado-Velarde E, Ramsey L, Spooner D, Stone T and Stouffer T: Watching Rocks Change Mechanical Earth Modeling, Oileld Review 15, no. 2 (Summer 2003): 22-39. 50. Will R, Archer R and Dershowitz B: Integration of Seismic Anisotropy and Reservoir-Performance Data for Characterization of Naturally Fractured Reservoirs Using Discrete-Feature-Network Models, paper SPE 84412, presented at the SPE Annual Technical Conference and Exhibition, Denver, October 58, 2003. 51. Rawnsley K and Wei L: Evaluation of a New Method to Build Geological Models of Fractured Reservoirs Calibrated to Production Data, Petroleum Geoscience 7, no. 1 (February 2001): 2333. > Using components of a photorealistic model, photography and innovative software to map bedding, fractures and faults. A high-resolution digital photograph of an analog outcrop in the Guadalupe Mountains (top) is processed. The software detects and enhances the discontinuities on the photograph (middle). The white coding indicates a high level of discontinuity, and the black coding represents a low level of discontinuity. Both bedding (green) and fracture faces (red) are mapped (bottom). The height of the vertical outcrop face ranges from about 20 to 25 ft [6.1 to 7.6 m]. 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 20 Summer 2006 21 picked. The result is a 2D map of linear outcrop featuresmostly fractures and beddingbut at a higher resolution than that extracted from the LIDAR data. Once the innovative processing is made on the high-resolution digital photographs and LIDAR data, the results are recombined into the 3D photorealistic model for manual verication and analysis. At this stage, the 2D maps derived from the photos are transformed into 3D data as they are projected onto the photorealistic outcrop model as a series of planes and attributes. The results of the photographic and LIDAR analysis are displayed as attributes in an editing window, and compared to the photorealistic model by the interpreter for quality control. Following editing of the data, the structural geologist is able to begin the process of quanti - tative fracture interpretation. Because bedding is automatically mapped as a part of the process, the interpreter is able to perform quantitative analysis of fracture extent, density and orienta - tion on a layer-by-layer basis, thus estab lishing a mechanical stratigraphy. The analyzed joint planes and their relationship to bedding and faults can then be used as the basis for a discrete fracture network model. Such models can be analyzed in terms of representative fracture volumes and ow heterogeneity related to the fracture systems. Modeling the Effects of Fractures There are perhaps no other simulation tasks as challenging in todays oil and gas elds as constructing valid NFR models to simulate reservoir uid ow with a reasonable degree of certainty. The challenges span multiple disciplines and multiple scales, and must always be addressed with limited information. The ultimate aim in reservoir simulation is to estimate and predict the distribution and ow of uids within the reservoir in response to production or injection. Natural fractures make achieving this aim considerably more difcult. Some experts simplify the challenges of NFR uid-ow simulation into three categories. First, a model must resolve the uid pathways by determining fracture connectivity. Connectivity depends on fracture length, orientation and intensity, which come from subsurface data and outcrop analogs. Second, knowledge of fracture- system permeabilities, permeability variation across the eld, and the interaction between fractures and the matrix is essential. Third, the uid pressure, or capillary pressure, and the relative permeabilities in the reservoir must be captured. Additionally, a good understanding of the in-situ stress regime is needed for credible NFR simulation. This information comes from a variety of sourcesincluding logging measure - ments, borehole breakout and leakoff testsand is used in mechanical earth models. 49 The complexity of NFRs represents a real challenge in reservoir simulation. The most geologically realistic models are discrete fracture network (DFN) models. In these models, each fracture is represented as a plane in the reservoir with attached properties such as aperture and permeability. DFNs are able to represent the geometric complexity of fractured reservoirs with a high level of detail. Fluid ow can be simulated through DFNs using nite- element methods, and the effects of matrix ow can also be incorporated. Creating a plausible model, however, places great demands on geoscientists, and the fracture system must be parameterized in all its detail. This model is typically built from high-quality data near wellboresfor example, borehole image data, core analysis and pressure-transient dataand is expanded to the interwell region using geostatistical techniques. DFN models can also be guided by seismic anisotropy fracture- characterization results and production data. 50 Well and seismic data are generally not sufcient to provide information about fracture extent and connectivity and so outcrop analogs become crucial sources of information. Today, the generation of DFNs still has limitations. DFNs are computationally intensive, so it is not possible to model all of the fractures within a reservoir in this way. While a DFN could be used for an individual well test history-match, commercially available DFNs can handle only single-phase ow and thus cannot model secondary-recovery mechanisms. 51 It is possible to represent only the largest fractures geometri - cally in cellular models, while smaller fractures have to be represented as modied cell properties. However, the physics of ow between fractures and matrix in cellular models can be represented using the nite-difference method and using dual-porosity and dual-porosity/ dual-permeability techniques. It is difcult to provide a link between the geologists view of a fractured reservoir and a cellular representation. One method for dealing with this problem is to create small-scale DFN models that represent the details of the fracturing and to upscale them to cellular grid blocks using either static or dynamic methods. For example, a joint system was mapped from a > Automatic fracture and fault delineation. One or several attributes are selected for the generation of Cube B from the seismic Cube A. Conditioning by the Petrel Automated Structural Interpretation module is applied to Cube B using swarm intelligence, which enhances the fracture and fault features to produce the resulting Cube C. The fault surfaces are then extracted as separate objects as shown on Cube D. These surfaces can then be incorporated into geologic models. A B C D 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 21 helicopter photograph of a eld outcrop (left). Joints were picked on the photograph using what is now the Petrel Automated Structural Interpretation technique. The results were used to build a DFN, capturing the entire complexity of the network. With an assumed aperture, the upscaled permeabilities in three different directions were determined using a pressure solver and input to a cellular simulation model. 52 Flow simulation in cellular models is performed in two ways: by nite-difference and by streamline simulation. 53 Finite-difference simulators typically offer a wide range of functionalities and are preferred in long-term, mature project environments. Also, nite- difference simulators have been more suited for simulating uid ow not dominated by reservoir heterogeneities in models with fewer uncertain - ties. Streamline simulators, such as the three- phase ECLIPSE FrontSim module, are better for accessing dynamic reservoir behavior in large, multimillion-cell models. Streamline simulators are faster to run and allow asset teams to quickly validate upscaled reservoir models with dynamic data (below left). Armed with a suitable ow simulator, asset teams can now examine connectivity across the reservoir and consider strategies to maximize hydrocarbon recovery. As more data come into the model, each portion of the reservoir model can be ne-tuned. This may involve improving structural and mechanical earth models, matrix and fracture models, and matrix-fracture exchange models (next page). 22 Oileld Review > Streamline simulation. Streamline simulators, such as the ECLIPSE FrontSim software, allow reservoir engineers and geoscientists to quickly simulate fluid flow in heterogeneous reservoirs. These simulators are especially useful when simulating the effects of fractures or other high- permeability conduits on waterflooding for secondary recovery. In this example, the streamlines and reservoir layers are color-coded according to water saturation, S w . G03 G05 G11 G09 G04 G14 G12 G13 G01 34-5 G07 G06 G02 1.0 0.1 S w > Example of an automatically generated fracture pattern from an outcrop in a 50-m by 50-m [164-ft by 164-ft] area (top left) incorporated into a discrete fracture network (DFN) model. A constant aperture was assigned to the fractures, and the permeability was upscaled using a pressure solver. The upscaled permeability in the X-direction, Block K xx , is scaled according to the color bar (left). Histograms (bottom) show Block K xx and the fracture porosity for each 10-m by 10-m [32.8-ft by 32.8-ft] cell. The rose diagram (top right) shows the orientation of 1,669 fractures interpreted by what is now the Petrel Automated Structural Interpretation process. 10 m Length-weighted orientation of 1,669 fractures 100 90 80 70 60 50 40 30 20 10 0 P e r m e a b i l i t y
( K x ) ,
m D 10 2 . 5 7 . 5 1 2 . 5 1 7 . 5 2 2 . 5 2 7 . 5 3 2 . 5 3 7 . 5 4 2 . 5 4 7 . 5 5 2 . 5 9 8 7 6 5 4 3 Permeability (K x ), mD Permeability, X direction 2 1 0 F r e q u e n c y ,
n u m b e r
o f
b l o c k s 3 2 1 0 . 0 1 7 0 0 . 0 1 7 4 0 . 0 1 7 8 0 . 0 1 8 2 0 . 0 1 8 6 0 . 0 1 9 0 0 . 0 1 9 4 0 . 0 1 9 8 0 . 0 2 0 2 0 . 0 2 0 6 0 . 0 2 1 0 0 . 0 2 1 4 0 . 0 2 1 8 Fracture porosity, % Fracture Porosity 0 F r e q u e n c y ,
n u m b e r
o f
b l o c k s Y (N) X (E) Z 52. A pressure solver is a tool in modeling software that enables the calculation of pressure at every point in a model. 53. Alaka JO, Bahamaish J, Bowen G, Bratvedt K, Holmes JA, Miller T, Fjerstad P, Grinestaff G, Jalali Y, Lucas C, Jimenez Z, Lolomari T, May E and Randall E: Improving the Virtual Reservoir, Oileld Review 13, no. 1 (Spring 2001): 2647. 54. Ahr WM, Allen D, Boyd A, Bachman HN, Smithson T, Clerke EA, Gzara KBM, Hassall JK, Murty CRK, Zubari H and Ramamoorthy R: Confronting the Carbonate Conundrum, Oileld Review 17, no. 1 (Spring 2005): 1829. Akbar M, Vissapragada B, Alghamdi AH, Allen D, Herron M, Carnegie A, Dutta D, Olesen J-R, Chourasiya RD, Logan D, Stief D, Netherwood R, Russell SD and Saxena K: A Snapshot of Carbonate Reservoir Evaluation, Oileld Review 12, no. 4 (Winter 2000/2001): 2041. 55. Kossack CA and Gurpinar O: A Methodology for Simulation of Vuggy and Fractured Reservoirs, paper SPE 66366, presented at the SPE Reservoir Simulation Symposium, Houston, February 1114, 2001. Gurpinar O, Kalbus J and List DF: Numerical Modeling of a Large, Naturally Fractured Oil Complex, paper SPE 59061, presented at the SPE International Petroleum Conference and Exhibition, Villahermosa, Mexico, February 13, 2000. Gurpinar O, Kalbus J and List DF: Numerical Modeling of a Triple Porosity Reservoir, paper SPE 57277, presented at the SPE Asia Pacic Improved Oil Recovery Conference, Kuala Lumpur, October 2526, 1999. 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 22 Summer 2006 23 Typically, models are tested and calibrated using historical pressure and production data history-matchingand must be updated and ne-tuned with new information. An asset teams ability to quickly update reservoir models and run multiple simulations has been enhanced, and continues to improve, with the availability of increased computing power. Fracture Breakthroughs Some of the largest hydrocarbon reservoirs in the world are naturally fractured carbonate reservoirs in the Middle East, Mexico and Kazakhstan. 54 In many cases, these reservoirs have three porosity systems: fracture, matrix and vuggyboth connected and isolatedand involve multiphase uid ow, adding to the modeling difculties. The challenges facing operators in these elds are daunting. Declining hydrocarbon productivity, increasing water production and signicant volumes of unswept oil are the most obvious reasons for concern. Closer examination has revealed inherent difculties in modeling heterogeneous, dual- and triple-porosity reservoirs with multiphase uid ow. In these cases, it has been useful to develop special relationships for relative permeabilities and capillary pressure that take the complexities into account. 55 On March 25, 2006, Schlumberger, in an alliance with King Fahd University of Petroleum and Minerals, ofcially opened the Schlumberger Dhahran Center for Carbonate Research (SDCR) to engage in collaborative projects focusing on carbonate reservoirs, the majority of which are NFRs. Scientists at this state-of-the-art research center will focus on the development of technologies that address the challenges of exploiting these complex reservoirs, including research in land seismic technologies, geology, rock physics and uid dynamics. In the past, available static and dynamic data have dictated an asset teams approach to NFR characterization, modeling and simulation. Today, a better understanding of NFR complexities, improved measurements and interpretation techniques across a wider range of scales, faster and vastly improved modeling capabilities and exciting new research will make the industrys progress in fractured reservoirs natural. MGG > Modeling naturally fractured reservoirs. A workflow example describes the major elements involved in NFR modeling during the project startup (green background), model creation (yellow background) and model fine-tuning (blue background) phases. The numbers at the bottom indicate where in the workflow that model fine-tuning should take place, in order of preference. Engineering Evaluation Geological Evaluation Fracture Indicators Log-Derived Properties Reservoir and Dynamic Data for Flow Modeling Data Collection, Verification and Validation Pressure Production Injection analysis Well summaries Pressure- treatment analysi permeability (K r ) s Reservoir Capillary pressure (P c ) Single-well models Project Objectives Status of the Field Status of Data Project Time/$$ Well location Unique identifier Directional survey Well completion Production Wellbore facilities Pressure Well test PVT Core analysis Relative permeability Logs Image logs Dipmeter logs Sedimentology Seismic data Seismic navigation Velocity control Drilling records Reports Previous studies Cores Sedimentology Facies model Stratigraphy Correlations Synthetics Seismic interpretation Faults and horizons Structure modeling Dipmeter interpretation Petrophysical evaluation Regional structural setting Structural framework Curvature Lithology Drilling events Well tests Production behavior Image logs Sonic logs Total and matrix petrophysics Model layering influence Model grid influence Fault Model Fracture Index Flow Simulation Grid Discrete Fracture DFN Verification 3D Property Distribution Reservoir- Management Model Predictive Reservoir Model K r , P c Two- phase Three-phase Flow Model Yes No Dual System 1 2 4 3 NFR Modeling PVT, production, pressure, well completion, injection, test database, K r c , P, prediction objectives Fracture Model Matrix Model Streamline Simulator Connectivity Review Structural Framework Fracture Intensity Direction Multiphase Reproduce Historical Behavior 58732schD04R1:58732schD04R1 8/14/06 7:29 PM Page 23 Model