42 Oilfield Review

The Source for HydraulicFracture Characterization

Les BennettJoël Le CalvezDavid R. (Rich) SarverKevin TannerCollege Station, Texas, USA

W.S. (Scott) BirkGeorge WatersOklahoma City, Oklahoma, USA

Julian DrewGwénola MichaudPaolo PrimieroSagamihara, Kanagawa, Japan

Leo EisnerRob JonesDavid LeslieMichael John WilliamsCambridge, England

Jim GovenlockChesapeake Operating, Inc.Oklahoma City, Oklahoma

Richard C. (Rick) KlemSugar Land, Texas

Kazuhiko TezukaJAPEXChiba, Japan

For help in preparation of this article, thanks to Gilles LeFloch, Montrouge, France; and Bill Underhill, Houston.DataFRAC, FMI (Fullbore Formation MicroImager),HFM (Hydraulic Fracture Monitoring), StimMAP andVSI (Versatile Seismic Imager) are marks of Schlumberger.PS3 is a mark of Vetco Gray, now owned by Schlumberger.SAM43 is a mark of Createch. Primacord is a mark of DynoNobel Incorporated.

Improved understanding of hydraulic fracture geometry and behavior allows asset

teams to increase stimulation effectiveness, well productivity and hydrocarbon

recovery. Although seismic methods for characterizing hydraulic fractures have

existed for years, new seismic hardware and processing techniques make this type

of monitoring significantly more effective than in the past.

Many of the world’s large, high-permeabilityreservoirs are now approaching the end of theirproductive lives. Increasingly, the hydrocarbonsthat fuel nations and economies will come fromlow-permeability reservoirs, and those tightformations require hydraulic fracture stimu-lation to produce at economical rates.

In the USA alone, operating companies spentroughly US$ 3.8 billion on hydraulic fracturing in2005.1 This huge expenditure is expected toincrease in the near future and to spreadthroughout the world. Companies need tools thathelp them determine how successfully theirhydraulic fractures have optimized wellproduction and field development. To do this,these tools should provide information abouthydraulic fracture conductivity, geometry,complexity and orientation.

While indirect well-response methods—fracture modeling using net-pressure analysis,well testing and production-data analysis—areused routinely to infer the geometry andproductivity of hydraulic fractures, measure-ments of the formation’s response to fracturingare now feasible to quantify fracture geometry,complexity and orientation.2 This articlediscusses the importance of characterizinghydraulic fractures when trying to optimizeproduction rates and hydrocarbon recoverywithin a field. We highlight a method ofmonitoring hydraulic fractures that uses seismictechnologies, including data acquisition, proces-sing and interpretation, and some associated

complexities. The microseismic hydraulic frac-ture monitoring technique is demonstrated incase studies from the USA and Japan, featuringtwo different fracturing environments.

Fracture StimulationFrom the first intentional hydraulic fracturestimulation of a reservoir in the late 1940s,engineers and scientists have sought tounderstand the mechanics and geometry ofhydraulically created fractures.3 Although anincrease in productivity or injectivity of astimulated reservoir may imply a successfultreatment, it does not necessarily mean that thereservoir and fracture models correctly pre-dicted the outcome.

Reservoir characteristics should always beconsidered when designing hydraulic fracturetreatments. In moderate- to high-permeabilityreservoirs, fractures are designed to improveproduction by bypassing near-wellbore formationdamage.4 In these reservoirs, the most importantfracture characteristic is dimensionless fractureconductivity—a function of the width, permea-bility and length of the fracture and of formation-matrix permeability. In permeable but weaklyconsolidated reservoirs, fracturing methods areused in conjunction with gravel packing toreduce the pressure drop and fluid velocitiesaround a wellbore during production, andtherefore mitigate sand production.5

Winter 2005/2006 43

1. Spears R: “Oilfield Market Report 2005,” Spears &Associates, Inc., http://www.spearsresearch.com/(accessed on October 14, 2005).

2. Barree RD, Fisher MK and Woodroof RA: “A PracticalGuide to Hydraulic Fracture Diagnostic Technologies,”paper SPE 77442, presented at the SPE Annual TechnicalConference and Exhibition, San Antonio, Texas,September 29–October 2, 2002.Cipolla CL and Wright CA: “Diagnostic Techniques toUnderstand Hydraulic Fracturing: What? Why? andHow?” paper SPE 59735, presented at the SPE/CERI GasTechnology Symposium, Calgary, April 3–5, 2000.

3. Brady B, Elbel J, Mack M, Morales H, Nolte K and Poe B:“Cracking Rock: Progress in Fracture Treatment Design,”Oilfield Review 4, no. 4 (October 1992): 4–17.

4. Meng HZ: “Design of Propped Fracture Treatments,”in Economides MJ and Nolte KG (eds): ReservoirStimulation. Schlumberger Educational Services:Houston, 1987.

5. Ali S, Norman D, Wagner D, Ayoub J, Desroches J,Morales H, Price P, Shepherd D, Toffanin E, Troncoso Jand White S: “Combined Stimulation and Sand Control,”Oilfield Review 14, no. 2 (Summer 2002): 30–47.

6. Meng, reference 4.7. Peterman F, McCarley DL, Tanner KV, Le Calvez JH,

Grant WD, Hals CF, Bennett L and Palacio JC: “HydraulicFracture Monitoring as a Tool to Improve ReservoirManagement,” paper SPE 94048, presented at theSPE Production Operations Symposium, Oklahoma City,Oklahoma, April 16–19, 2005.

8. Aly AM, El-Banbi AH, Holditch SA, Wahdan M, Salah N,Aly NM and Boerrigter P: “Optimization of GasCondensate Reservoir Development by CouplingReservoir Modeling and Hydraulic Fracturing Design,”paper SPE 68175, presented at the SPE Middle East OilShow and Conference, Bahrain, March 17–20, 2001.

9. Hashemi A and Gringarten AC: “Comparison of WellProductivity Between Vertical, Horizontal andHydraulically Fractured Wells in Gas-CondensateReservoirs,” paper SPE 94178, presented at the SPEEuropec/EAGE Annual Conference, Madrid, Spain,June 13–16, 2005.

In low-permeability reservoirs, by far themost common reservoir type to be fracturestimulated, industry experts have establishedthat fracture length is the overriding factor forincreased productivity and recovery.6 From areservoir-development standpoint, having areasonable understanding of hydraulic fracturegeometry and orientation is crucial fordetermining well spacing and for devising field-development strategies designed to extract morehydrocarbons.7 Reservoir modeling is alsoenhanced with improved knowledge of hydraulicfractures within a field.8

Natural fractures, often the primarymechanism for fluid flow in low-permeabilityreservoirs, severely compromise the ability topredict the geometry of hydraulic fractures andthe stimulation’s effect on production anddrainage. Understanding how hydraulicallycreated fractures interact with natural fracturesystems—open and mineral-filled—requiresknowledge of both hydraulic and naturalfracture types.

Hydraulic fractures tend to propagateaccording to the present-day stress directionsand preexisting planes of weakness, such asnatural fractures. The orientations of naturalfracture systems reflect ancient and possiblylocalized stress regimes.

In low-permeability reservoirs, the combinedeffects of natural and hydraulic fractures arelargely responsible for improved productivityfrom horizontal wells as compared with theproduction from vertical wells.9 The charac-teristics of both fracture types dictate the

preferred azimuth in which highly deviated andhorizontal wells should be drilled. Theoretically,in a horizontal well drilled parallel to themaximum horizontal stress direction, hydraulicstimulation produces a single longitudinalfracture along the horizontal wellbore. Thisscenario simplifies fluid flow out of the wellboreduring stimulation and into the wellbore duringproduction. However, depending on thecharacteristics and orientations of the naturalfracture systems, a transverse hydraulicfracturing strategy may actually result in higherproductivity, especially when multiple zones arebeing stimulated.10

While it is possible to have a good under-standing of existing natural fracture systems, ourability to determine hydraulic fracture geometryand characteristics has been limited. Geologicdiscontinuities such as fractures and faults candominate fracture geometry in a way that makespredicting hydraulic fracture behavior difficult.Clearly, the exploration and production (E&P)industry still has much to learn abouthydraulic fractures.

Characterization of the ComplexMore than simple curiosity drives petroleumindustry engineers and scientists to seekunderstanding of hydraulic fractures. Fracturestimulation is an expensive process, which canreap huge returns if done correctly. Yet tocomprehend hydraulic fracture propagation,accurate measurements of fracture growth,geometry and orientation are needed. These dataprovide a starting point for asset teams to assesspost-stimulation production performance andoptimize future stimulation treatments—tolower the cost or increase the effectiveness ofstimulation or both. This information can then beused to drive reservoir-development strategies.

Fractures from both horizontal and verticalwells can propagate vertically out of the intendedzone, reducing stimulation effectiveness, wastinghorsepower, proppant and fluids, and potentiallyconnecting up with other hydraulic fracturingstages or unwanted water or gas intervals. Thedirection of lateral propagation is largelydictated by the horizontal stress regime, but inareas where there is low horizontal stress

anisotropy or in reservoirs that are naturallyfractured, fracture growth can be difficult tomodel. In shallow zones, horizontal hydraulicfractures can develop because the vertical stresscomponent—the overburden weight—is smallest.A horizontal hydraulic fracture reduces theeffectiveness of the stimulation treatmentbecause it most likely forms along horizontalplanes of weakness—presumably betweenformation beds—and is aligned preferentially toformation vertical permeability, which is typicallymuch lower than horizontal permeability.

After a hydraulic fracture is initiated, thedegree to which it grows laterally or verticallydepends on numerous factors, such as confiningstress, fluid leakoff from the fracture, fluidviscosity, fracture toughness and the number ofnatural fractures in the reservoir.11 All hydraulicfracture models fail to predict fracture behaviorprecisely, and in many cases, models failcompletely, largely because of incorrectinformation and assumptions used in the models.Nevertheless, modeling is a necessary tool infracture engineering.

Stimulation engineers use hydraulic fracturesimulators to design and predict optimal fracturestimulation treatments. Basic inputs to thesemodels include fluid and proppant propertiesand volumes, closure stress, pore pressure,formation permeability and mechanical rockproperties, such as Poisson’s ratio and Young’smodulus. The risk of an inadequate treatmentoccurring is increased by estimating theseinputs. Asset teams can take steps to reduce thisrisk by using better models and by morethoroughly characterizing the reservoir andassociated stresses. These steps may includeacquiring petrophysical and mechanicalproperties from logs, obtaining borehole stressand natural fracture information from boreholeimages, and directly measuring the stresses byperforming the DataFRAC fracture datadetermination service.

Fracture modeling is a necessary part of thestimulation design and improvement process.However, even the most complex models fallshort in predicting reality.12 In the last 15 years orso, the industry has learned that hydraulicfractures are much more complex than the

biwing, single-plane cracks depicted in models.Investigation of actual hydraulic fracturegeometries, from minebacks, core-throughs andthousands of mapped fractures, has shown analmost limitless range of complexities, startingwith fracture asymmetry and the creation ofmultiple competing fractures.13

Given the complexities introduced by thepresence of natural fracture systems, reservoirheterogeneity and stress anisotropy, there islittle reason to believe that a hydraulicallyinduced fracture would maintain symmetry as itpropagates outward from the borehole.Asymmetrical hydraulic fractures formasymmetrical drainage patterns that should beconsidered when planning development drillingand modeling fluid flow within the reservoir. Inaddition, unexpected hydraulic fracture behaviorcan occur in depleted reservoirs or duringrefracturing operations.14

Assess and MonitorVarious methods are available to assess hydraulicfracture geometry before, during and afterfracture creation (next page).15 The accuracy ofindirect well-response techniques is linked tothe accuracy of the fracture and reservoir modelsthat generate the prediction. By far the mostcommon way to judge how well the treatmentwas delivered and its resulting geometry is toperform a net-pressure fracture analysis shortlyafter, or even during, the fracture treatment. Theresult of this analysis is closely linked to treatingpressure and therefore suffers when actualbottomhole pressure data are not available.Unfortunately, on a large percentage of jobs,treating pressure is measured at the surface—corrected for hydrostatic head and pipe friction.A more accurate treating pressure is measureddownhole, but even accurate treating pressuredata do not necessarily reflect fracture geometry.16

Another indirect way to deduce the geometryof hydraulic fractures uses post-treatmentproduction data. This method determines thewell productivity and is represented as aneffective fracture geometry that reflects theportion of the hydraulic fracture that is open,cleaned up and contributing to production. It mayrequire months to years of production history to

44 Oilfield Review

10. Brown E, Thomas R and Milne A: “The Challenge ofCompleting and Stimulating Horizontal Wells,” OilfieldReview 2, no. 3 (July 1990): 52–63.

11. Fracture propagation occurs when the stress-intensityfactor exceeds the degree of fracture toughness nearthe fracture tip. Fracture toughness, or the criticalstress-intensity factor, can be measured by performingcore burst tests in the laboratory.

12. Barree et al, reference 2.

Jeffery RG, Settari A and Smith NP: “A Comparison of Hydraulic Fracture Field Experiments, IncludingMineback Geometry Data, with Numerical FractureModel Simulations,” paper SPE 30508, presented at theSPE Annual Technical Conference and Exhibition, Dallas,October 22–25, 1995.

14. Dozier G, Elbel J, Fielder E, Hoover R, Lemp S, Reeves S,Siebrits E, Wisler D and Wolhart S: “RefracturingWorks,” Oilfield Review 15, no. 3 (Autumn 2003): 38–53.

15. Cipolla and Wright, reference 2.16. Barree et al, reference 2.

Wright CA, Weijers L, Davis EJ and Mayerhofer M:“Understanding Hydraulic Fracture Growth: Tricky butNot Hopeless,” paper SPE 56724, presented at the SPEAnnual Technical Conference and Exhibition, Houston,October 3–6, 1999.

13. Peterson RE, Warpinski NR, Lorenz JC, Garber M,Wolhart SL and Steiger RP: “Assessment of theMounds Drill Cuttings Injection Disposal Domain,” paper SPE 71378, presented at the SPE AnnualTechnical Conference and Exhibition, New Orleans,September 30–October 3, 2001.

Winter 2005/2006 45

perform the analysis, and the fracture geometrythat has been cleaned up may be vastly differentfrom the fracture geometry created hydraulically.The effective producing geometry is importantfor production estimation, but will, in general,underestimate the hydraulic fracture length.

Similar to the production analysis method,estimating fracture geometry from well testingmethods—buildup and drawdown—betterdefines the effective production geometry thanwhat has been created hydraulically.

Near-wellbore methods have been used toinvestigate the presence of hydraulic fractures.These include radioactive tracers, and temper-ature and production logs. While these tech-niques are widely used to detect the presence ofhydraulic fractures and estimate fracture height,their limitation is that they measure in a regionthat is at or near the wellbore and may not berepresentative of what is occurring away fromthe borehole.

Advances in radioactive isotope taggingduring injection and in the interpretationmethods that use hundreds of spectral channels

allow stimulation engineers to better discernfluid and proppant placement during multiple-stage stimulation treatments. Temperaturesurveys run after stimulation treatments identifynear-wellbore regions that have been cooled bythe injection of fracturing fluids and thereforeprovide an estimate of fracture height.Production logs—measurements such as fluidflow, fluid density and temperature—are used toidentify perforation intervals that are open andcontributing to flowback or production. Apositive flow response from a perforated interval

> Capabilities and limitations of indirect and direct hydraulic fracture diagnosis techniques. (Adapted from Cipolla and Wright,reference 2.)

• Cannot resolve individual and complex fracture dimensions• Mapping resolution decreases with depth (fracture azimuth 3° at 3,000-ft depth and 10° at 10,000-ft depth)

cannot determinemay determinecan determineTechniques

Main Limitations

Ability to Estimate

FractureDiagnostic

MethodGroup

Capabilities and Limitations of Fracture Diagnostics

Leng

th

Heig

ht

Asym

met

ry

Wid

th

Azim

uth

Dip

Volu

me

Cond

uctiv

ity

Far f

ield

, dur

ing

frac

turi

ngN

ear w

ellb

ore,

afte

r fra

ctur

ing

Mod

el b

ased

Surfacetiltmetermapping

Downholetiltmetermapping

Microseismicmapping

Radioactivetracers

Temperaturelogging

Productionlogging

Boreholeimage logging

Downholevideo

Net-pressurefracture analysis

Well testing

Productionanalysis

• Resolution in fracture length and height decreases as monitoring-well distance increases

• No information about proppant distribution and effective fracture geometry

• Limited by the availability of potential monitoring wells

• Measurement in near-wellbore volume• Provides only a lower limit for fracture height if fracture and well path are not aligned

• Limited by the availability of potential monitoring wells

• No information about proppant distribution and effective fracture geometry

• Dependent on velocity-model correctness

• Thermal conductivity of different formations can vary, skewing temperature log results

• Post-treatment log requires multiple passes within 24 h after the treatment

• Provides only a lower limit for fracture height if fracture and well path are not aligned

• Run only in open hole• Provides fracture orientation only near the wellbore

• May have openhole applications

• Run mostly in cased holes and provides information only about zones and perforations contributing to production

• Results dependent on model assumptions• Requires accurate permeability and reservoir pressure estimates

• Results dependent on model assumptions• Requires accurate permeability and reservoir pressure estimates

• Results depend on model assumptions and reservoir description• Requires “calibration” with direct observations

• Provides only information about zones or perforations contributing to production in cased-hole applications

suggests that the zone has been stimulated,especially if it compares favorably with apretreatment logging pass. However, flow intothe wellbore from a set of perforations may notmean that a specific interval has been treatedmore effectively because reservoir fluids can flowthrough communicating hydraulic fractures fromone zone to the next.

In an effort to better characterize hydraulicfracture behavior and geometry away from thewellbore, two HFM Hydraulic FractureMonitoring techniques have proved enormouslysuccessful. These far-field fracture-mappingmethods are surface and downhole tiltmetersand microseismic monitoring (above). Availablefor more than a decade, tiltmeters measure

hydraulic fracture-induced tilt, or deformation.By placing these devices in an array of shallowboreholes—20 to 40 ft [6 to 12 m] deep—deformation induced by fracture creation ismeasured. A map of deformation at the surfacecan be constructed from these surface data,allowing estimation of the azimuth, dip, depthand width of the hydraulic fracture.

Downhole tiltmeters are deployed in nearbymonitoring wells at a depth similar to that of thecreated fracture. Because this technique allowsthe sensors to be placed much closer to apropagating fracture than the surface method,the fracture geometry measurements tend to bemore accurate and include fracture azimuth,height, length and width.17 The success of

tiltmeter methods usually depends on the spatialrelationship between the tiltmeters—surface ordownhole—and the treatment well.

Mapping with surface tiltmeters haslimitations when attempting to characterizehydraulic fractures deeper than 10,000 ft[3,050 m]. As a general rule, downhole tiltmeterslose their effectiveness when the distance fromthe hydraulic fracture to the tiltmeter exceedsthree times the length of the created fracture.Another method, first investigated in 1982,monitors far-field fracture growth and geometryusing sensitive seismic receivers, such as theSchlumberger VSI Versatile Seismic Imager tool, deployed in nearby wells to detectmicroseismic events.18

46 Oilfield Review

17. Barree et al, reference 2.Cipolla and Wright, reference 2.

18. Albright JN and Pearson CF: “Acoustic Emissions as a Tool for Hydraulic Fracture Location: Experience at the Fenton Hill Hot Dry Rock Site,” SPE Journal 22(August 1982): 523–530.

20. Warpinski NR, Wolhart SL and Wright CA: “Analysis andPrediction of Microseismicity Induced by HydraulicFracturing,” paper SPE 71649, presented at the SPEAnnual Technical Conference and Exhibition, NewOrleans, September 30–October 3, 2001.

19. Arroyo JL, Breton P, Dijkerman H, Dingwall S, Guerra R,Hope R, Hornby B, Williams M, Jimenez RR, Lastennet T,Tulett J, Leaney S, Lim T, Menkiti H, Puech J-C,Tcherkashnev S, Burg TT and Verliac M: “SuperiorSeismic Data from the Borehole,” Oilfield Review 15,no. 1 (Spring 2003): 2–23.

> Tiltmeter and microseismic methods of far-field fracture monitoring. Tiltmeters (top) measure smallchanges in earth tilt. When these are mapped they show the deformation in response to the creationof hydraulic fractures. Tiltmeters can be deployed on surface or downhole in a monitoring wellbore.Microseismic monitoring (bottom) uses sensitive, multicomponent sensors in monitoring wells torecord microseismic events, or acoustic emissions (AEs), caused by rock shearing during hydraulicfracture treatments. The microseismic data are then processed to determine the distance andazimuth from the receiver to the AE and the depth of the AE.

Treatment well

Microseismic eventReceivers

Reservoir

Hydraulic fracture

Monitoring well

Downhole tiltmetersin monitoring well

Fracture-induced trough at surface Surface tiltmeters

Fractureracture

Dept

h

Winter 2005/2006 47

Tracking the CrackingMicroseismic events, or small earthquakes, occurwhen the normal stress is reduced alongpreexisting planes of weakness until shearslippage occurs. These shear movements emitboth compressional and shear waves that can bedetected by geophones. However, many believethe tensile cracking of rock that occurs duringfracture stimulation has a minimal contribution todetectable microseismic activity. Because thiszone of shearing accompanies the fracture tiparea, locating the source of these waves in spaceand time allows scientists and engineers toconstruct a map of the created fracture by plottingthe location of acoustic emissions (AEs) over timewhile fracturing. However, AEs may also occuraway from the fracture tip where there is fluidleakoff into the matrix or where stress changescause shear slippage in natural fractures.

To record compressional and shear waves,multicomponent—for example, three-component(3C)—geophones are placed in a monitoring wellto determine the location of microseismic events.The distance to the event can be calculated bymeasuring the difference in arrival timesbetween the compressional, or primary (P-(( )

waves, and shear, or secondary (S-(( ) waves. Also,hodogram analysis, which examines the particlemotion of the P-waves, may determine theazimuth angle to the event. The depth of theevent is constrained by using the P- and S-wavearrival delays between receivers observed at the monitoring well (above). This localizationtechnique requires an accurate velocity modelfrom which to calculate event locations, a low-noise environment, highly sensitive geophones torecord microseismic events and knowledge of theexact location and orientation of the receivers.Although this may seem simple, the process iscomplex and challenging.

The quality of hydraulic fracture charac-terization is directly linked to the quality of thevelocity model, or velocity structure, on whichthe interpretation is based. Initial velocitymodels typically are built using borehole soniclogs that describe the vertical velocity changes atwellbores. However, the time it takes for an AEto go from the source—near the hydraulicfracture—to the receiver and the direction fromwhich it comes into the receiver are influencedby the interwell geology. Borehole seismic

measurements, such as vertical seismic profiles(VSPs), provide detailed velocity informationaround the monitoring well. VSP surveys helprelate the time domain to the depth domain andtherefore help calibrate the velocity model. TheVSI tool used to acquire the VSP data alsorecords the microseismic events, ensuringconsistency in data acquisition, processing and interpretation.19

Reservoir-fluid type may also impact micro-seismic activity. Fluid factors can reduce stressand pore-pressure changes in the formation thatoccur during fracturing. Having gas in theformation instead of less compressible liquidsdecreases the area of microseismic activity.Consequently, some in the industry believe thatgas-filled reservoirs produce a narrower band ofmicroseismic events that more clearly definesthe geometry of the fracture.20

To locate AEs, a monitoring tool—typically anarray of eight 3C geophones for the VSI tool—isdeployed in a monitoring well within 2,000 ft[610 m] of the treatment well at roughly the samedepth as the treatment interval. The optimalplacement and geometry of the microseismic toolwithin the monitoring well are heavily dependent

> Locating acoustic emissions. The distance (D) to the event can be derived by measuring theDDdifference (∆T ) between the compressional, or primary (P-) wave and the shear, or secondary (S-)wave arrival times, Tp and Ts, respectively (top left). The valuett D is heavily dependent on the velocitymodel (bottom left), which is usually described by the tt P- and S-wave velocities, - Vp and Vs, respectively,of each layer in the model. The second coordinate, azimuth to the microseismic event, is determinedby examining the particle motion of the P-waves using hodograms (- top right). The depth of thettmicroseismic event, the third coordinate, is derived by examining the P- and S-wave arrival delaysbetween receivers, or moveout, at the monitoring well (bottom right).tt

Depth Determination

4,000 8,000 12,000 16,000

Velocity, ft/s

6,300

5,300

4,300

Dept

h, m

Velocity Model

Treatment well Monitoring well

Distance Determination

Azimuth-Angle Determination

P S

TpTT TsTT

∆T∆T = T TsTT – T .pTTD = ∆D T .p s / (s p s)

on the surrounding velocity structure, so accurateearth models help optimize the monitoringconfiguration.21 Unfortunately, the ideal spatialconfiguration between the treatment wellboreand potential monitoring wellbores occurs in only a small percentage of cases. Consequently,there is an ongoing effort to enable the recordingof AEs from treatment wells—a harsh and noisy environment.

Producing oil fields have many sources ofnoise that may have a negative impact on themicroseismic HFM technique, includingelectrical noise, nearby drilling activity andhydraulic fracturing jobs or fluid flowing fromperforations in the monitoring well. Much of thenoise can be eliminated on site or throughadaptive filtering during data processing.Improved seismic response can also be achievedthrough advances in acquisition technology.

For example, the Schlumberger microseismicHFM technique uses the VSI device, whichprovides excellent vector fidelity (right).22 TheVSI tool is deployed on wireline cable and usesthree-axis technology in each sensor package, orshuttle; eight sensor packages are typicallydeployed. The tool’s sensors were designed to beacoustically isolated from the main body of thetool but acoustically coupled to the casing duringthe HFM operation. This helps minimize thepotential for noise and maximize data qualitywhen recording very small microseismic events.The number of sensor sections and their spacingwithin the VSI configuration can be adjusted,depending on what is required.23

Optimal positioning of the sensor arrayshould be determined using network survey-design techniques.24 Once the VSI tool is set atthe appropriate depth in a monitoring well, theHFM engineer must determine the orientation ofthe tool to make use of particle-motion data fordetermining the azimuth angle. This isaccomplished by monitoring a perforation shot,string shot or other seismic source in thetreatment well, or in another well near thetreatment well.25 The utility of perforations orstring shots to calibrate velocity models has beendocumented.26 However, shot-based velocities areoften substantially different—sometimes higher,sometimes lower—than velocities calculatedfrom sonic data. These differences may be due toperforation-timing problems, imprecise locationsof perforations and receivers because ofinaccurate or nonexistent wellbore-deviationsurveys, reservoir heterogeneity between thetreatment and monitor wells, and inherent

differences between the velocity measurementsbeing compared—including anisotropy andinvasion effects.27

With the tool orientation determined, thesurface equipment is set up for continuousmonitoring, and when an event is detected,buffered data are recorded. On-site processinglocates the microseismic events, using one ofseveral available processing techniques, and theresults are transmitted to the fracturingoperations team at the treatment well location.The data are also sent to a processing center formore detailed interpretation.28

48 Oilfield Review

> Measuring acoustic emissions. The Schlumberger VSI Versatile Seismic Imager tool (left) usestttthree-axis (x, y, z) geophone accelerometers (right) that are acoustically isolated from the tool bodyttby an isolation string to acquire high-fidelity seismic data. The VSI device is mechanically coupled totthe casing or formation by a powerful anchoring arm. The coupling quality can be tested by using aninternal shaker before operations commence. Up to 40 sensor packages, or shuttles, can be linkedttogether to increase vertical coverage; however, eight shuttles are normally used in HFM operations.The tool is available in 3.375-in. and 2.5-in. diameters.

xz

y

Shaker Couplingcontacts

Threecomponents

Isolationspring

Texas Proving GroundIn the mining, waste disposal, geothermal andgas-storage industries, microseismic methodshave long been used to help understand thenature of hydraulically created fractures.However, recent improvements in tool design,processing and mapping accuracy, coupled withthe growing importance of low-permeability,hydraulically fractured reservoirs as a resource,have increased this technology’s utility in the oiland gas industry. The Barnett Shale reservoir inthe north-central Texas Fort Worth basin—oneof today’s most active gas plays in the USA—highlights the importance of direct and timelymicroseismic hydraulic fracture character-ization.29 Today, Barnett Shale fields produce

Winter 2005/2006 49

more than 1,200 million ft3/d3 [34 million m3/d],3

58% of thef total gas production from US gasshales (left).30

The Barnett Shale formation is anaturally fractured, ultralow-permeability—about 0.0002 mD—reservoir. Because of thisfextremely lowy permeability,w a large hydraulicfracture surface area is required to effectivelystimulate the reservoir. Consequently, largevolumes of fluidf are pumped at high rates duringstimulation treatments.

The Barnett Shale is a Mississippian-age,organic-rich, marine-shelf shalef deposit thatcontains fine-grained, nonsiliciclastic material.This formation overlies a major unconformitysurface that truncates the Ordovician-age rocksbelow. Throughout much of thef productive area,the Viola limestone creates a lower barrier tohydraulic fracturing and separates the under-lying, water-bearing Ellenberger formation fromthe Barnett Shale. Hydraulic fractures thatbreak through the Viola limestone typicallyresult in unwanted water production anddecreased gas production.

Stimulation of thef Barnett Shale has hadvariable effectiveness for reasons that are poorlyunderstood. The companies initially operatingy inthe Barnett Shale soon observed that thisreservoir did not respond to stimulation in thesame way asy conventional gas reservoirs. Unusualpost-treatment occurrences in which neigh-boring wells watered out indicated extremelylong hydraulic fracture growth, often inunexpected directions from treatment wells.Modern hydraulic fracture monitoring methods,most notably microseismicy monitoring, haveshown that Barnett Shale stimulation anddevelopment are complicated by naturalfractures and faults, which drastically influenceyhydraulic fracture behavior along with reservoirproductivity andy drainage. Moreover, the stressanisotropy iny the Barnett Shale is low, soattempts to model hydraulic fracture behaviorand geometry asy simple, single-plane events havebeen ineffective.

In the last five years, engineers and scientistshave learned more about the natural andhydraulic fracture systems in the Barnett Shaleformation. With that knowledge, they haveyadapted drilling strategies to improve gasproduction and recovery.31 One of thesef strategiesis the incorporation of horizontalf wells. Whileapproximately twicey as expensive as a verticalwell, horizontal wells typically generateyestimated ultimate recoveries that are threetimes greater than those of verticalf wells.

21. Le Calvez JH, Bennett L,t Tanner KV, Grant WD,t Nutt L,tJochen V, Underhill W and Drew J: “MonitoringMicroseismic Fracture Development tot OptimizeStimulation and Production in Aging Fields,” TheLeading Edge 24, no. 1 (January 2005): 72–75.

22. Vector fidelity is the property of multicomponent seismictreceivers to respond correctly to an impulse. A correctresponse occurs when a given impulse applied parallelto one of the three components registers only on thatcomponent andt when applied parallel to eachcomponent individuallyt registers the same magnitudeon each of the three components. The motion that istdetected by multicomponent seismict receivers ideallyis the same as the original impulse.Nutt L,t Menkiti H and Underhill B: “Advancing the VSPEnvelope,” Hart’s E&P 77, no. 10 (October 2004): 51–52.

23. Nutt ett al,t reference 22.24. Curtis A, Michelini A, Leslie D and Lomax A: “A

Deterministic Algorithm for Experimental Design Appliedto Tomographic and Microseismic Monitoring Surveys,”Geophysical Journal International 157, no. 2 (May 2004):595–606.

25. A string shot ist made up of Primacord detonating cordfired at strategict locations, for example near thetreatment depth,t to transmit at seismic wave withoutcreating a hole in the casing.

26. Warpinski NR, Sullivan RB, Uhl JE, Waltman CK andMachovoe SR: “Improved Microseismic FractureMapping Using Perforation Timing Measurements forVelocity Calibration,” paper SPE 84488, presented at thetSPE Annual Technical Conference and Exhibition,Denver, October 5–8, 2003.

27. Eisner L and Bulant P:t “Borehole Deviation Surveys AreNecessary for Hydraulic Fracture Monitoring,” preparedfor presentation at thet 86th EAGE Conference andExhibition, Vienna, Austria, June 12–15, 2006.

28. Durham LS: “Fracture ‘Groans’ Quietly Noisy:Microseismic Detection Emerging,” AAPG Explorer 25,no. 12 (December 2004): 16–18.

29. Frantz JH, Williamson JR, Sawyer WK, Johnston D,Waters G, Moore LP, MacDonald RJ, Pearcy M,Ganpule SV and March KS: “Evaluating Barnett ShaletProduction Performance Using an IntegratedApproach,” paper SPE 96917, presented at thet SPEAnnual Technical Conference and Exhibition, Dallas,October 9–12, 2005.Maxwell SC, Urbancic TI, Steinsberger N and Zinno R:“Microseismic Imaging of Hydraulic Fracture Complexityin the Barnett Shale,”t paper SPE 77440, presented atthe SPE Annual Technical Conference and Exhibition,San Antonio, Texas, September 29–October 2, 2002.Fisher MK, Wright CA,t Davidson BM, Goodwin AK,Fielder EO, Buckler WS and Steinsberger NP:“Integrating Fracture Mapping Technologies to OptimizeStimulations in the Barnett Shale,”t paper SPE 77441,presented at thet SPE Annual Technical Conferenceand Exhibition, San Antonio, Texas, September 29–October 2, 2002.

30. http://www.pickeringenergy.com/pdfs/TheBarnettShaleReport.pdf (accessedf November 30,r 2005).

31. Fisher MK, Heinze JR, Harris CD, Davidson BM, Wright CAtand Dunn KP: “Optimizing Horizontal CompletionTechniques in the Barnett Shalet Using MicroseismicFracture Mapping,” paper SPE 90051, presented at thetSPE Annual Technical Conference and Exhibition,Houston, September 26–29, 2004.

> Map of the north-central Texas Fort Worth basin showing Barnett Shaleactivity. There are currently more than 3,400 vertical and 300 horizontalwells producing from the Barnett Shale reservoir.

USAUU

Texas

Gainesville

Dallas

Fort Worth

Wichita Falls

O K L A H O M A

T E X A S

250 miles

0 25k

Producing wellsHorizontal-wellpermits

They have also been instrumental in opening upareas of the play where vertical wells have hadlimited success: in areas where the Violalimestone is absent and fracturing down into thewet Ellenberger is common. Optimum completiondesign in these wells is made more problematic

because of the complex nature of the hydraulicfracturing. Factors such as perforation-clusterspacing along laterals, stimulation stagingstrategies, fracture treatment sizing and offset-well placement all must be addressed to optimizeresource development.

Chesapeake Energy is one of severaloperators investigating the complexity offracturing the Barnett Shale from horizontalwellbores and its implications for acreagedevelopment. In February 2005, Chesapeakeused the StimMAP hydraulic fracture stimulationdiagnostics service in a vertical monitoring wellto determine fracture height, length, azimuthand complexity during a four-stage “slickwater”stimulation treatment on a horizontal well in theNewark East field.32 The design objective was toplace hydraulic fractures normal, or transverse,to the lateral. After perforating for each stage, apretreatment injection test was performed todetermine closure pressure and the rate ofpressure decline, which is a function of thedegree of natural fracturing because the matrixpermeability is too low to allow leakoff.

During all four stages, the primary fracturepropagation azimuth determined from micro-seismic monitoring was N60°E-S60°W, with anobserved preference for southwesterly growth(below and left). Most of the detected microseismicemissions were located to the southwest becauseof the monitoring configuration—bias existedbecause the monitoring well was positionedapproximately 2,000 ft to the southwest of thehorizontal treatment wellbore. In this case,formation heterogeneities were unlikely to bethe cause of the southwesterly bias. Chesapeakewas able to observe cross-stage communicationalong the lateral between Stages 1 and 2 andbetween Stages 2 and 3, which reduced theeffectiveness of those treatments.

50 Oilfield Review

32. Slickwater treatments use low proppant concentrations—in this case, less than 0.8 lbm/gal US [9.6 kg/m3]—allowing high-volume treatments at reduced cost. Thistype of treatment has been successful in the BarnettShale because it creates long fractures that connectwith crosscutting natural fractures, thereby increasingthe total effective hydraulic fracture length and drainagearea in a single well.

> Maps of microseismic events from the four-stage hydraulic fracture stimulation. The StimMAP displays include a three-dimensional (3D) view (top) anda plan view (middle). The treatment stages are color-coded: Stage 1 is purple, Stage 2 is blue, Stage 3 is green, and Stage 4 is yellow. Also included is asummary of each stage, including acoustically determined fracture system length, width and preferential azimuth (bottom). Depths are given relative totthe kelly bushing (KB).

91

517

369

444

Number ofN b fNumber ofNumber ofNumber ofNumber ofeventseventseventseventsevents

Z,360

Y,740

Y,025

X,358

PerforatedP f dP f t dPerforatedPerforatedPerforatedPerforatedintervali t lintervalintervalintervalintervaltop, MDt MDtop MDtop MDtop MDtop, MDp,

from KB, ftf KB ffrom KB ftfrom KB ftfrom KB ftfrom KB, ft,

Z,853

Z,227

Y,588

X,513

PerforatedP f dP f t dPerforatedPerforatedPerforatedPerforatedintervali t lintervalintervalintervalinterval

bottom, MDb tt MDbottom MDbottom MDbottom MDbottom, MD,from KB, ftf KB ffrom KB ftfrom KB ftfrom KB ftfrom KB, ft,

X,797

X,734

X,784

X,740

FractureFF tFractureFractureFractureFracturesystemtsystemsystemsystemsystemy

top, TVDt TVDtop TVDtop TVDtop TVDtop, TVDp,from KB, ftf KB ffrom KB ftfrom KB ftfrom KB ftfrom KB, ft,

Y,290

Y,305

Y,305

Y,309

FractureFF tFractureFractureFractureFracturesystemtsystemsystemsystemsystemy

bottom, TVDb tt TVDbottom TVDbottom TVDbottom TVDbottom, TVD,from KB, ftf KB ffrom KB ftfrom KB ftfrom KB ftfrom KB, ft,

493

571

521

569

FractureF tFractureFractureFractureFracturesystemtsystemsystemsystemsystemy

height, fth i h fheight ftheight ftheight ftheight, ftg ,g

1,918

1,728

1,556

1,521

SWSWSWSWSWSWextent, ftfextent ftextent ftextent ftextent, ft,

299

409

482

424

NENENENENENEextent, ftfextent ftextent ftextent ftextent, ft,

2,217

2,137

2,038

1,945

FractureF tFractureFractureFractureFracturesystemtsystemsystemsystemsystemy

length, ftl h flength ftlength ftlength ftlength, ftg ,g

1,143

2,275

1,138

527

FractureF tFractureFractureFractureFracturesystemtsystemsystemsystemsystemywidth, ftid h fwidth ftwidth ftwidth ftwidth, ft,

N60°E

N60°E

N60°E

N60°E

AzimuthA i hAzimuthAzimuthAzimuthAzimuth

Stage 2

Stage 3

Stage 4

WellW llWellWellWellWell

Stage 1

N

Hydraulic Fracture Data

Event rateTreating pressure, psiSlurry rate, bbl/min

N

Winter 2005/2006 51

During Stage 2, engineers on locationobserved that the bottomhole treating pressuresmatched those of Stage 1, so Chesapeake askedthe Schlumberger engineer to produce a quicksnapshot of the Stage 2 microseismic eventlocations. When compared to the Stage 1StimMAP results, the snapshot confirmed thatthe Stage 2 fracture was communicating with theprevious stage. To remedy this, three slugs ofproppant sand were pumped at a reduced rate todivert the treatment fluid away from theperforations that were taking the majority of thetreatment. Microseismic data confirmed that thetreatment had communicated with a complex setof parallel and conjugate natural fractures.

The Stage 3 perforation intervals werealtered to avoid a fault. Hydraulic fracturemonitoring confirmed that two primary fractureswere created on each side of the fault and werepossibly also affected by the presence of naturalfractures. Stage 4 did not appear to overlap withother stages.

In August 2005, Chesapeake used theStimMAP service on another horizontal well inNewark East field to determine the influence of afaulted karst zone on hydraulic fracturegeometry and orientation. Again, the stimulationinvolved four stages—slickwater treatments forStages 1, 3 and 4, and a CO2-fluid system forStage 2. The treatments were monitored from awell south-southwest of the east-southeast-oriented horizontal leg of the treatment well. Thedistance from the hydraulic fracturing to themonitoring well ranged from less than 500 ft[150 m] to more than 2,000 ft, depending on thelocation of the stage along the horizontalwellbore (below and right).

> Maps of microseismic events from another four-stage hydraulic fracture treatment. The StimMAP displays include a three-dimensional (3D) view (top)and a plan view (middle). The treatment stages are color-coded: Stage 1 is purple, Stage 2 is blue, Stage 3 is green, and Stage 4 is yellow. Also included isa summary of each stage, including acoustically determined fracture system length, width and preferential azimuth (bottom). Depths are given relative tomean sea level (MSL).

140

98

68

94

Number ofN b fNumber ofNumber ofNumber ofNumber ofNumber ofeventsteventseventseventsevents

Stage 1

Stage 2

Stage 3

Stage 4

WellW llW llWellWellWellWell

X,970

X,954

X,954

X,949

PerforatedP f dP f t dPerforatedPerforatedPerforatedPerforatede o a edinterval, TVDi t l TVDinterval TVDinterval TVDinterval TVDinterval, TVDinterval, TVD,from MSL, ftf MSL ff MSL ftfrom MSL ftfrom MSL ftfrom MSL ftfrom MSL, ft,

491

863

985

637

FractureFF tFractureFractureFractureFractureac u esystemtsystemsystemsystemsystemsystemy

height, fth i h fh i ht ftheight ftheight ftheight ftheight, ftg ,g

419

739

799

1,038

SSWSSSSWSSWSSWSSWSSWSSWextent, ftft t ftextent ftextent ftextent ftextent, ft,

264

178

676

630

NNENNENNENNENNENNENNEextent, ftft t ftextent ftextent ftextent ftextent, ft,

1,105

1,168

1,247

1,942

FractureFF tFractureFractureFractureFractureac u esystemtsystemsystemsystemsystemsystemywidth, ftid h fidth ftwidth ftwidth ftwidth ftwidth, ft,

N15°E

N15°E

N15°E

N15°E

AzimuthA i hA i thAzimuthAzimuthAzimuthAzimuth

X,744

X,483

X,670

X,682

FractureF tFractureFractureFractureFractureFracturesystemtsystemsystemsystemsystemsys ey

top, TVDt TVDtop TVDtop TVDtop TVDtop, TVDtop, TVDp,from MSL, ftf MSL ff MSL ftfrom MSL ftfrom MSL ftfrom MSL ftfrom MSL, ft,

Y,235

Y,346

Y,655

Y,319

FractureF tFractureFractureFractureFractureFracturesystemtsystemsystemsystemsystemsys ey

bottom, TVDb tt TVDbottom TVDbottom TVDbottom TVDbottom, TVDbottom, TVD,from MSL, ftf MSL ff MSL ftfrom MSL ftfrom MSL ftfrom MSL ftfrom MSL, ft,

683

917

1,475

1,168

FractureFF tFractureFractureFractureFractureac u esystemtsystemsystemsystemsystemsystemy

length, ftl h fl th ftlength ftlength ftlength ftlength, ftg ,g

758

551

400

393

NNWNNWNNWNNWNNWNNWNNWextent, ftft t ftextent ftextent ftextent ftextent, ft,

347

617

847

1,549

SSESSSSESSESSESSESSESSEextent, ftft t ftextent ftextent ftextent ftextent, ft,

E

Time

Legend

Hydraulic Fracture Data

Event rateTreatment time, minTreatment pressure, psi

Chesapeake knew the location of four faultsin the area from seismic images and well control,so engineers placed multiple perforation clusterswithin each stage to avoid directly fracturing intothe faults. Even with these precautions, fractureinitiation was influenced by the presence offaults near Stages 1, 2 and 4 (above). Stage 1most likely communicated with a fault. Themicroseismic and pressure evidence supportedthis scenario. The bulk of the microseismicevents occurred between the second and thirdset of perforations, and the instantaneous shut-inpressure for Stage 1 was significantly lower thanthat of the other three stages.

The StimMAP service achieved Chesapeake’sobjective of defining the orientation andgeometry of the hydraulically created fracturesin the treatment well. Engineers determined thatthe dominant fracture azimuth was N15°E. Whilefracture-height growth was largely symmetricaland upwardly contained within the BarnettShale, downward growth was observed in allstages. Laterally, Stage 3 demonstrated symmet-rical growth, whereas growth in Stages 1, 2 and 4appeared asymmetrical.33 The StimMAP interpre-tation also concluded that there was littlecommunication between the different stages.

Today, much of the effort to monitor hydraulicfracture growth is directed toward fracturestimulations in horizontal wells to assessfracture height and complexities associated withfracture interference. These issues cannot beaddressed in horizontal wells with the near-wellbore evaluation methods previouslymentioned. The ability to measure hydraulicfracture characteristics allows engineers tojudge the impact of completion and stimulationdesign changes—for example, varying theplacement or spacing of perforation intervalsalong the horizontal wellbore or alteringproppant carrier fluids. Because of improvedhydraulic fracture characterization, theeffectiveness of hydraulic fracture treatments inthe Barnett Shale has been linked to the openingof secondary natural fracture systems, whichincreases the width of the treated volume.

Testing Technologies, Models andLimits in JapanEven though microseismic monitoring tech-niques have been available for years, the quest toimprove velocity modeling, data acquisition,processing and interpretation continues. Japan Exploration Company (JAPEX) andSchlumberger collaborated on a project to test

the feasibility of microseismic monitoring in theYufutsu gas field, Hokkaido, Japan.34

The reservoir in the Yufutsu field is a naturallyfractured, Cretaceous-age granite and overlyingconglomerate located at depths from 4,000 m[13,124 ft] to 5,000 m [16,405 ft]. Within the field,there is no apparent correlation between gasproduction and well location or well orientation.However, JAPEX has determined that produc-tivity is controlled by the local stress conditionand by the distribution and orientation of severalnatural fracture systems across the field.35 Morespecifically, large-aperture natural fractures, or“mega” fractures, oriented parallel to themaximum horizontal stress, act as gas conduits,while small-scale fractures affect gas storage andmigration. Characterization of the fracturesystems has been successful at the wellbore,using borehole-imaging devices such as the FMIFullbore Formation MicroImager tool. However,to understand more about reservoir behavior andto improve reservoir modeling using a discretefracture network simulator, JAPEX needed toinvestigate a larger reservoir volume.36

A preliminary injection test using a four-levelVSI tool occurred in October 2003. In December2004, JAPEX installed tubing-deployed, perma-nent seismic monitoring technology, the Vetco

52 Oilfield Review

> Influence of faults on Barnett Shale stimulation. Chesapeake placed perforations along thehorizontal completion interval to avoid fracturing into four known faults. Even with theprecautions, the StimMAP hydraulic fracture stimulation diagnostics interpretation indicatedtthat the microseismic activity was concentrated around some of the fault planes and influencedby the presence of faults near Stages 1, 2 and 4.

Lower Barnett top

Ellenberger top

Monitoring well

Stage 4 Stage 3 Stage 2 Stage 1

EW

Winter 2005/2006 53

Gray PS3 system, in the SK-2D treatment well torecord production-induced AEs. JAPEX observedonly minimal microseismicity in the field,probably because of the lack of pressure drop inthe reservoir. However, microseismic activity wasinduced during injection operations thatinitiated shearing along preexisting naturalfractures. Consequently, recording and analyzingthese AEs using hydraulic fracture monitoringtechniques could help define the geometry andextension of the natural fracture systems. A VSPand a small-scale injection experiment wereconducted in February 2005, and a large-scaleinjection experiment was performed in May 2005 (right).

The VSP data were used to enhance the existingvelocity model and ultimately proved important inthe fracture analysis. Using a seismic airgun sourceplaced in a specially designed pit at surface and11 ⁄1 6⁄⁄ -in. Createch SAM43 seismic acquisition toolsdeployed within production tubing in near and farmonitoring wells, a 49-level VSP was recordedacross the pertinent interval in both wellssimultaneously. The VSP provided good quality z-component—vertical-component—data thatallowed Schlumberger and JAPEX scientists toevaluate the coupling quality of the Createch toolsand to find the optimal tool position for amicroseismic monitoring survey. Velocityinformation from the VSP survey was also used tocorrect the existing velocity model, which in turnimproved the accuracy of calculated AE locations.

Another objective of the project was toevaluate the hydraulic fracture monitoringperformance of the permanent, tubing-conveyedVetco Gray PS3 prototype system. An upper and alower sensor were deployed in the SK-2Dinjection well. The PS3 sensors were affected byelectrical noise. However, once the noise wasreduced by error-prediction filtering, P- andS-wave arrivals were observed. Although theprototype sensors also were affected by noisefrom pumping fluid in this completion, some ofthe AE events had sufficient signal-to-noiseratios to identify P- and S-wave arrivals. This testrepresented the first successful use of multiplesensors to map hydraulically induced AEs froman injection well.

Using criteria from multiple monitoringsensors for event discrimination, the 40-h, 500-m3 [3,145-bbl] fluid-injection program inFebruary produced 920 detectable events, ofwhich 40 exhibited detectable P- and S-wavephases at three or four sensors and werelocatable with reasonable confidence. Acomparison of event locations was made between

33. The large distance between the monitor well and thereservoir volume affected by Stage 4 may be responsiblefor the asymmetry observed in the event locations.

34. Drew J, Primiero P, Leslie D, Michaud G, Eisner L andTezuka K: “Microseismic Monitoring of a HydraulicInjection Test at the Yufutsu Gas Reservoir,” paper B,presented at the 10th Formation Evaluation Symposiumof Japan, Chiba, Japan, September 29–30, 2004.

> Geometry of the injection well, two monitoring wells and sensors with a map (inset) showing thettexperiment location.

Dept

h, m

–2,000

–2,500

–3,000

–3,500

–4,000

–4,5003

21

0Distance, km 0 1 2 3 4 5 6 7

Distance, km

Near monitoring well Injection wellFar monitoring well

3,141 m3,141 m 3,188 m3 188 mm

3,656 m3,656 m,

3,339 m3,339 3

Injection pointn pointn pointection 4 0134,013 m4,013 m13 m

2000 miles

0 200k

S e ao f

J a p a n

Pa

ci f

ic

Oc

ea

n

Ea

s tC

hi n

aS

ea

J A P A N

Sapporo

Far monitoring wellInjection wellNear monitoring well

10 miles

0 1k

35. Tezuka K, Namaikawa T, Tamagawa T, Day-Lewis A andBarton C: “Roles of the Fracture System and State ofStress for Gas Production from the Basement Reservoirin Hokkaido, Japan,” paper SPE 75704, presented at theSPE Gas Technology Symposium, Calgary, April 30–May 2, 2002.

36. Tamagawa T and Tezuka K: “Validation of Clusteringof Fractures in Discrete Fracture Network Model byUsing Fracture Density Distributions Along Boreholes,”paper SPE 90342, presented at the SPE Annual Technical Conference and Exhibition, Houston,September 26–29, 2004.

those calculated using the existing velocitymodel and those calculated using the VSP-refined velocity model (left). The revised velocitymodel significantly improved the source-locationcalculations, reducing uncertainty. The resultsusing the new model showed a tighter cluster ofactivity than was evident using the previousvelocity model, which had been built from VSPinformation obtained in other parts of the field.

The larger injection experiment in Maypumped 5,600 m3 [35,223 bbl] of fluid duringsix days in four different tests, or stages.37 Theexperiment produced 447 located events out of atotal of 2,515 detected events, some of whichoccurred after pumping had stopped (next page).

To determine the impact of monitoring frommultiple wells, the event locations calculatedusing only data from the near monitoring wellwere compared with the event locationscalculated using data from multiple monitoringlocations. The criteria for multiwell localizationwere that clear P-wave and S-wave arrivals couldbe picked at the near well, that at least oneP-wave arrival could be picked at the farmonitoring well and that at a minimum one P- orS-wave arrival could be picked from the PS3

treatment well data.The localization algorithm was then run on

both the single-well data and the multiwell data,using the new velocity model. With single-welldata, distance to the event was calculated usingthe P- and S-wave traveltime data, and angles ofray incidence were determined using hodogramanalysis. For single-well and multiwell processing,hypocenter estimates were made using theprobability density functions formed frommeasured and modeled time delays and angles.38

The single-well location cluster is more dispersedand more difficult to interpret than the multiple-well distribution, which also shows additional

54 Oilfield Review

> The impact of having a VSP-calibrated velocity model. A comparisonof the February 2005 test microseismic event localizations using thepreexisting velocity model (top) versus using the local VSP-calibratedvelocity model (bottom) shows a tighter clustering of events using theupdated model. This significantly reduces the uncertainty in defininghydraulic fracture geometry and orientation. In each of the displays, aplan view is shown on top, a north-to-south cross section is located in thelower left and a west-to-east cross section is shown in the lower right.

–147.5

N-S

, km

–147.5

–148.0

–148.5

–149.0

–149.5–47.5 –47.0 –46.5 –46.0 –45.5

W-E, km

N-S

, km

–147.5

–148.0

–148.5

–149.0

–149.5–47.5 –47.0 –46.5 –46.0 –45.5

W-E, km

–2,000

–2,500

–3,000

–3,500

–4,000

–4,500

Dept

h, m

–149.5 –149.0 –148.5 –148.0 –147.5N-S, km

–2,000

–2,500

–3,000

–3,500

–4,000

–4,500

Dept

h, m

–47.5W-E, km

–47.0 –46.5 –46.0 –45.5

–47.5 –47.0 –46.5 –46.0 –45.5

–2,000

–2,500

–3,000

–3,500

–4,000

–4,500

Dept

h, m

–149.5 –149.0 –148.5 –148.0N-S, km W-E, km

–2,000

–2,500

–3,000

–3,500

–4,000

–4,500

Dept

h, m

Preexisting Velocity Model

Local VSP-Calibrated Velocity Model

37. Primiero P, Armstrong P, Drew J and Tezuka K:“Massive Hydraulic Injection and Induced AEMonitoring in Yufutsu Oil/Gas Reservoir—AEMeasurement in Multiwell Downhole Sensors,”paper 50, presented at the SEGJ 113th Annual FallMeeting, Okinawa, Japan, October 16–18, 2005.

38. Michaud G, Leslie D, Drew J, Endo T and Tezuka K:“Microseismic Event Localization and Characterizationin a Limited-Aperture HFM Experiment,” ExpandedAbstracts, SEG International Exposition and 74th AnnualsMeeting, Denver (October 10–15, 2004): 552–555.Tarantola A and Valette B: “Inverse Problems: Quest forInformation,” Journal of Geophysics 50 (1982): 159–170.

Winter 2005/2006 55

> Examining acoustic emission (AE) magnitude and quantity during the second injection stage in Yufutsu gas field, Japan.This test started with a 2.5-h step-rate injection, followed by a series of 1-h high-rate injections, each followed by 1-hshut-in cycles. Next, a continuous injection rate of 14 bbl [2.2 m3] per minute was maintained for 19 h, with an exceptionfor scheduled pump maintenance. The middle plot displays estimated event magnitude. The size of the green ellipses isproportional to the signal-to-noise ratio. The number of microseismic events is shown on the bottom plot. Tubingpressure (blue) and pump rate (magenta) are displayed on both plots. A plan view (top) shows the located events thatwere attributed to this particular stage (black) of the total number of located events during the entire May 2005 injectionexperiment (gray). The beginning of the step-rate injection shows a pressure and rate threshold before AEs start tooccur and, while the number of events decreases during the shut-in periods, AEs still occur in large numbers afterpumping has stopped.

–147.8

–148.0

–148.2

–148.4

–148.6

–148.8

–149.0

–149.2

–149.4

–149.6

–46.6 –46.4 –46.2 –46.0 –45.8W-E, km

N-S

, km

0

10

20

30

40

50

60

Tubi

ng p

ress

ure,

MPa

and

pum

ping

rate

, bbl

/min

Estim

ated

mag

nitu

des

–3.5

–3.0

–2.5

–2.0

–1.5

–1.0

–0.5

0

0.5

09:00 12:00 15:00 18:00 21:00 00:00 03:00 06:00 09:00 12:00 15:00 18:00 21:00 00:00 03:00 06:00

May 2005 second step-rate and high-rate injections, time

May 2005 second step-rate and high-rate injections, time

Num

ber o

f eve

nts/

15 m

in

0

10

20

30

40

50

60

Tubi

ng p

ress

ure,

MPa

and

pum

ping

rate

, bbl

/min

09:00 12:00 15:00 18:00 21:00 00:00 03:00 06:00 09:00 12:00 15:00 18:00 21:00 00:00 03:00 06:000

5

10

15

20

25

30

35

40

45

50

activity significantly farther to the north of the point of injection (right). The comparisonbetween the two results highlights the challengeof monitoring hydraulic fracture behavior in thefield, where monitoring options can be limited toa single well.

One of the primary motivations for acquiringpressure and AE measurements whilemonitoring the Yufutsu stimulation was the useof that information to validate reservoir-simulation models. JAPEX has developed anumerical simulator, which simulates theshearing of rocks, the associated AEs and thepermeability enhancements during hydraulicsimulation.39 Comparison of simulated andmeasured AE event locations along with iterativematching of pressure histories was used to helpconfirm the validity of the simulations.

In addition to improving the characterizationof natural fracture systems and reservoirmodeling in the Yufutsu gas field, the injectionexperiments have confirmed the value of anaccurate velocity model and the advantages ofmonitoring AEs from multiple stations. Althoughlonger monitoring distances are less desirable,the experiment shows that monitoring can bedone from considerable distances, depending onthe geology. In this case, the farthest monitoringtool in the far monitoring well was about 2.5 km[8,200 ft] from the microseismic activity.

AE data provide information about the spatialdistribution of the fracture system. Advancedmapping techniques such as the double-difference method and multiplet analysis providesource locations so precisely that AE clusters andfracture-related structures can be extracted.40

For instance, the results of the double-differencemethod applied to the Yufutsu dataset givesmultiple linear structures, which are interpretedas a medium-scale fracture system, bridging the

56 Oilfield Review

> A comparison of event localization from one monitoring well and frommultiple monitoring locations. The AE data from the May 2005 injectionexperiment were located based on the hodogram analysis—to determineangle—and P- and S-wave traveltimes—to determine distance. Fracture-maps that used only data from the near monitoring well (top) werecompared with fracture maps that used data from three monitor welllocations (bottom). The use of multiple monitoring locations constrainedtthe number of possible event localization solutions to yield fewer, buthigher quality localizations, which produces a clearer representation oftthe activity.

N-S

, km

–147.5

–148.0

–148.5

–149.0

–149.5–47.5 –47.0 –46.5 –46.0 –45.5

W-E, km

N-S

, km

–147.5

–148.0

–148.5

–149.0

–149.5–47.5 –47.0 –46.5 –46.0 –45.5

W-E, km

–2,000

–2,500

–3,000

–3,500

–4,000

–4,500

Dept

h, m

–149.5 –149.0 –148.5 –148.0 –147.5N-S, km

–2,000

–2,500

–3,000

–3,500

–4,000

–4,500

Dept

h, m

–47.5W-E, km

–47.0 –46.5 –46.0 –45.5

–149.5 –149.0 –148.5 –148.0 –147.5 –47.5 –47.0 –46.5 –46.0 –45.5

–2,000

–2,500

–3,000

–3,500

–4,000

–4,500

Dept

h, m

N-S, km W-E, km

–2,000

–2,500

–3,000

–3,500

–4,000

–4,500

Dept

h, m

Fracture Map Using Single-Well Data

Fracture Map Using Multiwell Data

39. Tezuka K, Tamagawa T and Watanabe K: “NumericalSimulation of Hydraulic Shearing and Related AE Activityin Fractured Gas Reservoir,” paper A, presented at the10th Formation Evaluation Symposium of Japan, Chiba,Japan, September 29–30, 2004.

40. The double-difference method is a mapping techniquethat relates multiple pair of events relative to each other.Multiplets are clusters of nearly identical wavelets frommultiple events with a similar focal mechanism thatoriginate at the same, or very nearly the same, locationbut occur at different times.

41. Tezuka K, Tamagawa T and Watanabe K: “NumericalSimulation of Hydraulic Shearing in FracturedReservoir,” paper 1606, presented at the WorldGeothermal Congress, Antalya, Turkey, April 24–29, 2005.

42. Drew J, Leslie D, Armstrong P and Michaud G:“Automated Microseismic Event Detection and Locationby Continuous Spatial Mapping,” paper SPE 95513,presented at the SPE Annual Technical Conference andExhibition, Dallas, October 9–12, 2005.

43. Eisner L and Sileny J: “Moment Tensors of EventsInduced in Cotton Valley Gas Field from WaveformInversion,” paper P227, presented at the EAGE 66thConference and Exhibition, Paris, June 7–10, 2004.

First step-rate test

Second step-rate test,first high-rate test

Second high-rate test

Third step-rate test

The four injection stages

Winter 2005/2006 57

gap between the fault system and fracturesobserved on borehole images.

Another advantage of AE data is that theyprovide spatial constraints for reservoirsimulation. JAPEX developed the Simulator forHydraulic Injection and Fracture Treatment(SHIFT) to simulate hydraulic injectionexperiments.41 This simulator works on a discretefracture-network model and simulates theshearing of preexisting fractures, related AEactivity and permeability enhancement in adynamic process. It does this by coupling fluid-flow analysis and shear-induced fracture-dilationanalysis. The AEs and the injection pressuresobserved during the experiment were used forthe postjob matching analysis. The size,orientation and migration history of the AE cloudhelped constrain the model parameters. Inaddition, AE clusters can be used as

deterministic information to modify the fracturenetwork directly. The Yufutsu project involvingJAPEX and Schlumberger tested some of theinherent limits of hydraulic fracture monitoring.

New Microseismic Activity One of the major limitations in microseismicmonitoring methods is finding candidatetreatment wells that have a nearby monitoringwell, or wells, in which to install the VSI tool. Notonly does the monitoring well need to berelatively close to the treatment well, dependingon the acoustic properties of the surroundingrock, but it must also be well cemented andacoustically quiet during fracturing operations.Ensuring that the monitoring wellbore is in theappropriate condition prior to running the VSItool often requires significant time and expense.

Scientists continually search for the balancebetween dependable AE detection andlocalization, and expedient processing andinterpretation that provides useful answers atthe treatment site. With the advent of fastercomputers, a new method that uses coalescencemicroseismic mapping (CMMapping) hasachieved fast and reliable event localization forreliable real-time fracture mapping.42

Another challenge addressed by Schlumbergergeophysicists when detecting and locating AEs isthe identification and interpretation of multiplets.For example, multiplets have been observed tooccur during two different pumping stages.Identical microseismic responses arise from, andare mapped back to, the same source locations.Therefore, multiplets indicate the reactivation ofa fracture or fault for which activity was detectedearlier. During a multistage hydraulic fracturetreatment, this may indicate crossflow betweenstages, resulting in an ineffective stimulation. Thekey is being able to identify the occurrence ofmultiplets in real time so that actions can betaken while pumping. Schlumberger scientistshave developed a crosscorrelation method todetect crossflow between stages that also providesanother layer of quality control in real-time eventlocalization (left).

Scientists at Schlumberger CambridgeResearch are also developing a robust seismicinversion to determine the mechanisms of theobserved microseismic events, for example,shear or tensile mechanisms.43 This techniqueallows going beyond “dots in a box” and, forexample, quantifying stress changes resultingfrom microseismic events. This information isused to further constrain geomechanical modelsand provide companies with a betterunderstanding of hydraulic fracture propagationor stress changes in the fractured reservoir.

Hydraulic fracture mapping has much to offerthe E&P industry, especially in tight reservoirdevelopment. Accurate fracture models, cali-brated using direct measurements of hydraulicfracture geometry, lead to improved reservoirsimulation and development. After decades ofsearching for the best way to characterizehydraulic fractures, the industry has returned tothe best source for the answers to our questions—the hydraulic fractures themselves. —MGG

0.8

0.6

0.4

1.0

0.2

0

–0.2

–0.4

–0.6

0.8

0.6

0.4

1.0

0.2

0

–0.2

–0.4

–0.6–200 –150 –100 –50 0 50 100

Even

t num

ber

Event number

Event number

Even

t num

ber

50 450

100 200 300 400 500 600 700

50

100

150

200

250

300

350

400

450

100

200

300

400

500

600

700

Stage 1

Stage 5

Stage 2Stage 3Stage 4

–900

–950

–1,000

–1,050

–1,110

–1,150

–1,200

–1,250

–1,300

W-E, m

Dept

h, m

1 150

Crosscorrelation of Stages 1 (1–157),2 (158–471) and 3 (472–769)

Five Stages of Hydraulic Fracturing-InducedMicroseismicity: Depth View

Crosscorrelation of Stages 3 (1–298)and 4 (299–497)

Cros

scor

rela

tion

coef

ficie

ntCr

ossc

orre

latio

n co

effic

ient

> Detecting cross-stage hydraulic fractures using multiplets. The technique is based on the identificationof multiplets as a result of reactivation of fractures from a previous stage. In this example from Texas,tthe upper graph is a crosscorrelation of all microseismic events from Stages 1, 2 and 3 (top left).ttStage 1 includes Events 1 through 157, Stage 2 includes Events 158 through 471, and Stage 3 includesEvents 472 through 769. The crosscorrelation coefficient is color-coded, identifying microseismicevents in different stages that originate from the same fractures—multiplets. When Stages 3 and 4—Events 1 through 298, and 299 through 497, respectively—are crosscorrelated, the coefficient remainsvery low except where the stages correlate with themselves (bottom left). The event map reflects thisttobservation (right).tt