38 Oilfield Review

Reducing Uncertainty with Fault-Seal Analysis

Kip CervenyBP AlaskaAnchorage, Alaska, USA

Russell DaviesRock Deformation Research (RDR) USA Inc.Dallas, Texas, USA

Graham DudleyRichard FoxBPAberdeen, Scotland

Peter KaufmanCambridge, Massachusetts, USA

Rob KnipeRock Deformation Research LtdUniversity of LeedsLeeds, England

Bob KrantzConocoPhillipsHouston, Texas

For help in preparation of this article, thanks to Karen Dawe,Geological Association of Canada, St. John's, Newfoundland,Canada; Jayne Harnett, RDR, Leeds, England; and DavidMcCormick, Cambridge, Massachusetts, USA.FMI (Fullbore Formation MicroImager), MDT (Modular Formation Dynamics Tester), OBMI (Oil-Base MicroImager),Petrel and RFT (Repeat Formation Tester) are marks of Schlumberger.

Oil and gas reservoirs in faulted siliciclastic formations are difficult to exploit. By

integrating seismic data, detailed core information, and wellbore and production

data, geoscientists can now model fault behavior and incorporate the results into

reservoir fluid-flow simulators. This integrated process improves prediction of fault

behavior, and reduces the uncertainty and risk associated with complex traps.

A fault can be a transmitter of or barrier to fluidflow and pressure communication. Categorizingfault behavior within these extremes is important for hydrocarbon drilling, explorationand development. Modern fault-seal analysismethods utilize seismic data, structural and microstructural information from high-resolution core analysis, and wellbore and production data to predict fault behavior and toreduce uncertainty and risk in faulted silici-clastic reservoir exploitation.

Sealing faults may be a primary control onthe trap in many hydrocarbon reservoirs, butthey may also transform a relatively large andcontinuous hydrocarbon reservoir into compart-ments that then behave as a collection of smallerreservoirs. Each compartment may have its ownpressure and fluid characteristics, hamperingefficient and effective field development and subsequent hydrocarbon recovery.

Faults that do not form a seal may prevent oiland gas from accumulating as hydrocarbons formand migrate through structures in the sub-surface. Open and permeable faults within anestablished reservoir may also cause serious lost-circulation problems during drilling operations.The loss of drilling mud can be expensive anddangerous, and can result in the abandonment ofwells. Whether detrimental or beneficial, faultsand their behavior need to be understood bygeologists and engineers to successfully exploreand extract hydrocarbon reserves.

Recent developments in fault-seal predictionhave focused on two separate but interrelatedaspects of faulting: fault architecture and fault-rock properties. The fault architecture refers tothe fault shape, size, orientation and inter-connectivity. It also refers to the distribution ofthe overall fault displacement into multiple subfaults. Horizontal fault length may rangefrom millimeters, in the case of microfaults, tohundreds of kilometers. For example, the SanAndreas fault in California, USA, is more than800 miles [1,290 km] long. Detailed studies inoutcrops and in the subsurface have shown thatlonger faults usually comprise interconnectedshorter faults. The fault clusters form a fault-damage zone or an interconnected halo of faultsat a range of scales that may have a large cumulative impact on reservoir behavior. Thedisplacement of the major and minor fault segments within the reservoir juxtaposes thereservoir across the fault against dissimilarlithologies, which may impact the fluid flow.

The rock properties that develop within thefault zones affect a fault’s ability to seal. Theseproperties are affected by the local facies, reservoir-fluid types and saturations, pressuredifferentials across faults, fault-zone architec-tures, burial and fault histories, andjuxtaposition of the lithologies across faults.1 Inaddition, pressure and phase changes duringreservoir development compound the complexityof analyzing fault-seal behavior.2

Winter 2004/2005 39

Modern methods in fault-seal analysisimprove the prediction of fault behavior in thesubsurface and reduce the uncertainty inexploiting faulted siliciclastic reservoirs. Thisarticle summarizes methods of fault-seal prediction and the associated uncertainties. Abrief introduction to basic fault theory helpsdefine the fundamental causes, types and characteristics of faults before presenting amore detailed characterization of the process offault-seal behavior and prediction. Also dis-cussed are oilfield technologies that are used tomeasure and predict fault characteristics. Case

studies from Hibernia, Newfoundland, Canada,and Prudhoe Bay, Alaska, USA, demonstrate howa better understanding of fault sealing improvesclastic reservoir simulation and development,thereby reducing uncertainty and risk.

Basic Fault Mechanics, Architecture and PropertiesWhen rocks or rock layers are subjected to tectonic stress, they bend, break, or do both. Inits simplest form, a fault is a planar break, or failure surface, in rock across which there isobservable displacement, or slip. Contraction

1. Facies designations represent the overall characteristicsof a rock unit that reflect its origin and differentiate theunit from others around it. Mineralogy and sedimentarysource, fossil content, sedimentary structures and texture distinguish one facies from another.

2. Davies RK and Handschy JW: “Introduction to AAPG Bulletin Thematic Issue on Fault Seals,” American Asso-ciation of Petroleum Geologists Bulletin 87, no. 3 (March2003): 377–380.Yielding G, Øverland JA and Byberg G: “Characterizationof Fault Zones for Reservoir Modeling: An Example fromthe Gullfaks Field, Northern North Sea,” AmericanAssociation of Petroleum Geologists Bulletin 83, no. 6(June 1999): 925–951.Jev BI, Kaars-Sijpesteijn CH, Peters MPAM, Watts NLand Wilkie JT: “Akaso Field, Nigeria: Use of Integrated 3-D Seismic, Fault Slicing, Clay Smearing and RFT PressureData on Fault Trapping and Dynamic Leakage,” AmericanAssociation of Petroleum Geologists Bulletin 77, no. 8(August 1993): 1389–1404.

and extension induce shear failure in rocks. Thedirection of the principal stresses dictates theorientation of the failure plane, or fault. Thestrength of the rock controls the magnitude ofthe shear stress necessary to break the rock.

Although oversimplified, the Andersoniantheory of faulting, developed by geologist E.M. Anderson in 1951, is still widely used as abasis to describe the fundamentals of fault orientation in failure.3 Anderson described thethree basic fault types—normal, reverse andwrench, or strike-slip—relative to the maximumregional stress orientations. This theory assumesthat one of the principal stresses—σ1, σ2 or σ3

in order from greatest to least magnitude—orthe lithostatic load, is always vertical, and thatthe others are orthogonal and horizontal. Thetheory predicts that faults will form as two con-jugate planes with the following threerelationships between fault orientation andprincipal stresses:• faults form at ± 30° to the σ1 direction• faults form at ± 60° to the σ3 direction• the line formed by the intersection of

conjugate fault planes will be parallel to σ2.These relationships are significant because if

geologists know the principal stress directions,they can predict fault orientations. If the relative magnitudes of the principal stresses arealso known, geologists can predict fault types(above left).

At the seismic map scale, however, faults arerarely planar because of perturbations in thestress field caused by heterogeneities andanisotropy in the rocks. More commonly, faultsare composed of separate segments with distincttips defined by lines of zero displacement. Thelinkages may occur as hard links where thefaults tips connect, or soft links where the fault-tip geometry is influenced by an adjacent faultthat lacks a physical connection.4 The displace-ment of the stratigraphy across a fault varies ina systematic pattern from zero displacement atthe fault tips to a maximum near the fault center. Anomalies in the systematic distributionin throw reflect the complexities in the lithologyand adjacent fault segments.5 Fault complexitiespreclude a simple interpretation of the fault orientation, geometry and architecture.

A fundamental step in evaluating faultbehavior and sealing properties is mapping thefaults and constructing fault-plane throw andjuxtaposition maps at the seismic scale (left).6

The limits of seismic resolution, however, intro-duce uncertainty in the throw mapped acrossthe fault and do not allow the mapping

40 Oilfield Review

> Relating fault types to stress orientation. The Andersonian theory explains the three main fault typesrelative to the principal stress orientation. These include the normal-fault style, in which σ1, thelargest in-situ stress, is vertical (top); the reverse-fault type, in which σ1 is horizontal, and σ3, thesmallest in-situ stress, is vertical (middle); and the wrench, or strike-slip, fault type, in which both σ1and σ3 are horizontal (bottom).

Normal-Faulting Stress Regime

Reverse-Faulting Stress Regime

Wrench-Faulting Stress Regime

σ1

σ2 σ3

σ3

σ2 σ1

σ2

σ3 σ1

> Interpreting faults from seismic data and modeling using software tools. Complex fault architecturein exploration and development scenarios can be made more understandable with the use ofpowerful mapping and imaging software such as the Petrel workflow tools application. In thisexample, color-coded stratigraphic intervals in the hanging wall and footwall are juxtaposed againstthe modeled fault surfaces in three dimensions.

5 15 25 35 45 55 65

Clay content, %

0.6

km0 1.0

0 miles

Winter 2004/2005 41

of faults whose throw is less than the seismicresolution. The total mapped throw across aseismic-scale fault may also include the summedthrows of numerous faults that are too small tobe detected individually at the seismic scale.The volume of closely spaced fault segments isknown as the fault-damage zone.

The mapped throw across a seismic-scalefault displaces the rock layers on a single faultor on a collection of multiple faults, each ofwhich is below the seismic resolution. The offsetinfluences the fault sealing and properties of thefault rocks within the fault zone. A sealing faultmay result, for example, if a fault intersectingdifferent lithologies places permeable, reservoir-quality rocks against less permeable rock, suchas shale. This is known as a juxtaposition seal. A fault seal may also form if the reservoir is still juxtaposed against itself—where the throw is less than the reservoir thickness—oragainst another reservoir. This occurs becausethe rock within the fault zone may develop lower permeability.

Different fault rocks develop under differentdeformation conditions, and their sealing properties are related to the conditions of deformation and lithologic factors, such as claycontent.7 Faults that cut porous sandstones withlow clay content—less than 15%—may developlow-permeability seals from porosity reductionassociated with the mechanical crushing of thequartz grains. These are called cataclastic ordeformation bands. Disaggregation bands can alsodevelop in clean sandstones, but without the asso-ciated reduction in porosity and grain crushing.

Faults in impure sandstones form phyllosilicate-framework fault rocks (PFFR),with higher clay contents—from 15 to 40%—thatreduce the porosity and permeability by com-pacting and mixing the clay particles and quartzgrains. Clay smear occurs along faults that cutrocks with greater than 40% clay. The clay layersor shales are dragged and deformed along thefault plane, forming a low-permeability barrier tofluid flow. Cementation may also occur along afault plane, forming nearly impermeable barriersto flow. These cemented zones, however, arerarely continuous unless they are associated witha regional change, such as an increase in temper-ature above 90°C [194°F] at which the rate ofquartz precipitation increases (above right).8

The most common faults found in oil and gasfields are normal faults, and most have somecomponent of oblique movement. Complex,three-dimensional (3D) fault geometries stemfrom the nucleation, growth and linking of

3. Anderson EM: The Dynamics of Faulting and Dyke Formation with Applications to Britain. Edinburgh, Scotland: Oliver and Boyd (1951): 206.For more on failure-plane orientation: http://www.naturalfractures.com/1.1.3.htm (accessed January 15, 2005).

4. Walsh JJ and Watterson J: “Geometric and KinematicCoherence and Scale Effects in Normal Fault Systems,”in Roberts AM, Yielding G and Freeman B (eds): TheGeometry of Normal Faults, Geological Society of London,Special Publication 56. Bath, England: The GeologicalSociety Publishing House (1991): 193–203.

5. The throw of a fault is the generalized elevation differenceof the same bed on the opposing sides of the fault, or thevertical component of displacement. Fault displacement

> Fault-rock classification relating clay content, fragmentation and lithification. The original host rocksinclude clean sandstones with less than 15% clay, impure sandstones with 15 to 40% clay, andclaystones and shales with greater than 40% clay. Fragmentation and lithification progress throughouta fault’s history and produce one of the three types of fault rocks from each host as shown in thelower portion of the diagram. Photographs at bottom illustrate different forms of fault rocks, including(A) disaggregated and cemented (left), (B) phyllosilicate-smear framework (center) and (C) clay-smearfault rocks (right).

Cataclasite

(intermediate

clay content)

Proto-cataclasite

(intermediate

clay content)

Ultra-cataclasite

(intermediate

clay content)

Disaggregation

zone(intermediate

clay content)Disaggregation

zone(clay-poor)

Proto-cataclasites

(clay-poor)

Cataclasite

(clay-poor)

Ultra-cataclasite

(clay-poor)

Ultra-cataclasite

(clay-rich)

Cataclasite

(clay-rich)

Proto-cataclasite

(clay-rich)

Disaggregation

zone(clay-rich

hydroplastic)

Clay content

Quartz-rich

lithologies

(deformation

band) fault-

rock series

Phyllosilicate

framework

fault-rock

series

Clay-smear

series

~15%

40%

100%

10%

50%

90%

Lithification

Poorly lithified

Partly lithified

Well lithified

Fragm

entat

ion

Cleansandstones,

silts

Impuresandstones,

silts

Claystones,

shales

Lithificationstate

Fragm

entat

ion

Clay content

A

B

C

A

C

cm

1 mm

CataclasiteUndeformedsandstone

B

in.cm

is the full distance across which a bed is separated oneach side of a fault and defined as:displacement = throw/sine (fault-plane dip).

6. Knipe RJ: “Juxtaposition and Seal Diagrams to Help Analyze Fault Seals in Hydrocarbon Reservoirs,” American Association of Petroleum Geologists Bulletin 81,no. 2 (February 1997): 187–195.

7. Davies and Handschy, reference 2.8. Fisher QJ and Knipe RJ: “Fault Sealing Processes in Sili-

ciclastic Sediments,” in Jones G, Fisher QJ and Knipe RJ(eds): Faulting, Fault Sealing and Fluid Flow in HydrocarbonReservoirs: Geological Society Special Publication 147.Bath, England: The Geological Society Publishing House(1998): 117–134.

faults, and give rise to damage zones. An under-standing of fault-damage zones is crucial inmodeling fault behavior and its impact on reservoir performance.

Fault-Zone Architecture CharacteristicsAn appreciation of fault-damage zone complex-ity can be obtained through careful study offaults in outcrops. Surface exposures allow geoscientists to observe fault architecture indetail and in a 3D spatial context and scale not afforded by subsurface investigation. Importantly, much of what determines fault-sealing properties occurs at subseismic scalesand within the fault-damage zone. Consequently,

the study of damage zones in outcrops hasbecome crucial in modeling fault seals and inpredicting how they affect subsurface fluid flow.

The damage zone is the volume of deformedrocks around a major fault that has resultedfrom the initiation, propagation, interaction andbuildup of slip along small faults between faultblocks.9 The deformed volume radiating awayfrom a main fault segment can be divided intoinner and outer damage zones. The inner dam-age zone typically consists of intensely deformedfault rocks that are difficult to map discretely,while the outer zone has a high density of small-throw faults that often maintain an orientationsimilar to the principal fault segment.

The damage-zone geometry can also bedefined along the strike of a fault, or faults, asthree distinct zones (below left). The first zoneis called the tip-damage zone and is associatedwith the stress concentration at the tip of themain fault segment, where the displacementgoes to zero. The second zone is called the link-ing-damage zone, and refers to the volumeaffected by the interaction between two subpar-allel, noncoplanar fault segments. Thewall-damage zone, the third zone, is locatedalong the fault surface and is a result of damagefrom continued fault slip or from damage by pre-viously abandoned fault tips as fault propagationcontinued through time.10 Secondary, subseismic-scale faults, natural fractures and cementationmay occur in all three zones.

Intensive investigation of fault exposures,like the Moab fault in southeast Utah, USA, hasallowed geoscientists to characterize fault-damage zones and make analogies to majorfaults in the subsurface. The Moab fault hasbeen extensively studied by geoscientists,including scientists from Schlumberger-DollResearch (SDR), Ridgefield, Connecticut, USA,and Rock Deformation Research (RDR) Ltd,Leeds, England.11 Located in the northeast

42 Oilfield Review

> Fault-zone classifications. A 3D conceptual diagram illustrates the inner and outer damage volumesassociated with faults. The map view overlay of the fault propagation through the host rocksindicates three distinct zones, which are the tip-damage zone (red), the linking-damage zone (lightblue) and the wall-damage zone (green). Fault tip lines are shown in black.

Tip-damage zone

Linking-damagezone

Wall-damagezone

Inner damage zone

Outerdamage zone

> High-precision mapping of the Moab fault-damage zone. Global positioning systems (GPS)and rover units, which determine exact surveylocations, were used to record discrete featuresand position them with high precision.Secondary structural elements, such as faultsand fractures, were tagged with key geologicattributes to capture the complexity and scale ofthe fault damage.

Winter 2004/2005 43

portion of the Paradox basin, the Moab fault is anormal fault approximately 28 miles [45 km]long with a northwest to southeast strike. Thefault comprises several linked segments. Thelongest segment has a throw of 3,150 ft [960 m]to the south, as observed from surface displace-ment and erosion of Pennsylvanian to Cretaceoussedimentary rocks.12 The Moab fault was activefrom at least the Triassic period until at least themid-Cretaceous period. The canyon landscapesurrounding Moab is ideal for mapping the faultexposure in three dimensions (above).

SDR and RDR scientists set out to capturedetailed outcrop data along a segment of theMoab fault-damage zone at Bartlett Wash as ananalog to similar structures expected but notimaged in the subsurface. Within the study area,

the throw along the main fault segment is 690 ft[210 m]. The older, Jurassic-age Slick Rockmember of the Entrada sandstone is wellexposed on the footwall and exhibits a densenetwork of small-throw faults within a narrowzone adjacent to the main fault segment.

The geoscientists employed a sophisticatedmapping technique, using a high-precision, differ-ential global positioning system (GPS) and roverunits to map discrete features to within 0.8 in.[2 cm] (previous page, right). Data coordinateswere tagged with key geologic attributes at manystations to capture the complexity and scale ofthe fault-damage zone. The positions and geome-tries of major and secondary structural elements,such as faults and natural fractures, were alsorecorded. Scientists created a digital geologic

model to use as an analog for subsurface fault interpretation to facilitate visualizationthrough innovative techniques, such as virtual

9. Kim Y-S, Peacock DCP and Sanderson DJ: “Fault-DamageZones,” Journal of Structural Geology 26 (2004): 503–517.

10. Kim et al, reference 9.11. Kaufman PS, McAllister E and Smallshire R: “Collection

and Visualization of 3D Digital Geologic Data Sets: An Example from the Moab Fault Zone, UT,” presented at the American Association of Petroleum GeologistsAnnual Meeting, New Orleans, April 16–19, 2000.McAllister E, Smallshire R, Knipe R and Kaufman P: “Geometry of Fault-Damage Zones from High ResolutionMapping of the Moab Fault Zone, UT,” presented at theAmerican Association of Petroleum Geologists AnnualMeeting, New Orleans, April 16–19, 2000.

12. “Quantification of Fault-Related Diagenetic Variation of Reservoir Properties at Outcrop,” http://www.fault-analysis-group.ucd.ie/Projects/UTAH.html (accessedJanuary 15, 2005).

> The Bartlett Wash study area, Moab, Utah, USA. A photographic cross section along the Moab fault allows illuminating viewsof the complex fault-zone architecture within the detailed mapping area (top). An aerial view (bottom) from the footwall to thehanging wall shows another perspective of the sharp contact formed by the Moab fault.

Detailed mapping area

Entradasandstone

Footwall

Hangingwall

Entradasandstone

SW NE

Vantage point ofupper photo

Approximate

328

m0 100

0 ftNE

Moab,Utah USA

field trips, and to use the fault population distribution as an input to flow models (right).

Although the static geometry and fault-rockproperties are the principal controls on cross-fault flow in the subsurface, fault reactivation isanother phenomenon that influences the flowproperties along the fault. Changes in tectonicstress regimes over geological time, for example,may reactivate a fault, opening pathways thatdid not exist previously, and allowing hydrocar-bons to leak. On a reservoir-production timescale, changes in pore-pressure regimes as aresult of current production or injection in andaround fault systems can initiate fault reactiva-tion and cause loss of seal.

Local pressure increases near or within thefault plane resulting from injection decrease theeffective normal stress, which may cause thefault to reactivate.13 Also, pressure changes inthe rocks surrounding faults, for example fromdepleting a reservoir, alter the in-situ stressesacting on fault planes and, depending on faultalignment relative to the principal stresses, maylead to reactivation and subsequent seal failure.This behavior has been documented in suchareas as the North Sea, the Gulf of Mexico andthe Bight basin, Australia.14

These pressure changes have major implica-tions in production, enhanced oil recovery (EOR)and pressure maintenance, and in subsurface gasstorage, including carbon dioxide [CO2] storagefor the reduction of greenhouse-gas emissions.15

The reactivation of reservoir-bounding faults compromises fault-sealing mechanisms, shearswell casings, and causes compaction and subsi-dence. The integration of fault-rock strengthproperties, the fault geometry and in-situ stressconditions provides valuable input for modelingand assessing the reactivation risk.16 The in-situstress orientations are interpreted with boreholeimaging devices, like the FMI Fullbore FormationMicroImager or OBMI Oil-Base MicroImagertools, and from the acquisition of pore pressuredata, using sampling tools such as the MDT Modular Formation Dynamics Tester or the RFT Repeat Formation Tester devices.

The Roles of Pressure and Timing in Fault SealingAn important concept in estimating the sealingcapacity of faults relates to the threshold pressure (Pt). In water-wet rocks, Pt is the lowestcapillary pressure (Pc) at which hydrocarbonsform a continuous path through the largest inter-connected pore throats in the fault rock.17 Knowingthe Pt of different fault rocks, generated underdifferent conditions, allows geoscientists to calculate

the maximum petroleum-column height (Ht) orsealing capacity of the fault rock that preventshydrocarbon migration across the fault. The capil-lary pressure of hydrocarbons under hydrostaticconditions against a fault seal increases upwardfrom zero at the free-water level (FWL), which is at the base of the hydrocarbon column. A capillary or membrane seal prevents hydrocarbonmigration across the fault for a hydrocarbon column height measured from the FWL to wherePc equals Pt. Membrane sealing occurs because ofthe surface tension between water and hydrocar-bon, so the effective permeability to hydrocarbonis zero when Pc is less than Pt (next page, top).

A hydrocarbon column with Pc greater thanPt of the fault rock will migrate slowly across thefault. The flow is retarded by the hydraulic resis-tant sealing of the fault rock. Hydraulic resistantsealing occurs when the relative permeability tohydrocarbon is low because of the water-wetfault rock and low pressure potential across the

fault for small hydrocarbon columns. The hydro-carbons may migrate at a slow rate, but hydraulicresistant sealing provides an effective seal overgeological time. At the base of the hydraulic-resistance zone, Pc is equal to Pt. Relativepermeability to hydrocarbon at this elevation iszero, but increases above this point in a transitionzone from membrane sealing up to geologicallysignificant leakage because of an increase in therelative permeability. Geologists considerhydraulic-resistance seal failure significant once the leakage rate exceeds the hydrocarbon-charge rate, at which point hydrocarbons stop accumulating.

Water-pressure differences in the reservoiracross a fault or in fault fill influence the heightof the resulting hydrocarbon column. Higherwater pressure in the aquifer outside the trap,for example, leads to water flow into the reservoir if the hydrocarbon saturation in thefault zone is less than the irreducible water

44 Oilfield Review

> Mapping the Moab fault zone. More than 70,000 structural features were mapped at the BartlettWash site to populate an analog model of the fault-damage zone (top). The density of small faultswithin the damage zone of the main segment of the Moab fault decreases as distance from the main fault increases. Red fault traces are within the inner damage zone, while yellow features arewithin the outer damage zone (bottom). Powerful software tools allow geologists to use innovativetechniques like virtual field trips, capturing the knowledge and experiences of team members at the site.

NApproximate

82

m0 225

ft0

Winter 2004/2005 45

saturation, Swirr. These conditions improve thefault-seal potential and increase hydrocarbon-column height. Lower pressures in the aquiferoutside the trap and in the fault fill at irre-ducible water saturation will lead to decreasedhydrocarbon-column heights in the trap. Theseinterrelationships between fluids, pressures androck properties are important controls for predicting fault behavior and sealing capabilities.

Fault architecture, throw distributions,lithologies, fault-rock distributions and proper-ties all impact the flow properties of faults. Faulthistory, however, is equally important when considering the sealing potential of fault traps inexploration and production. The burial history,deformation timing and hydrocarbon-charge history influence fault-rock properties and theirimpact on fault-seal capacity.

Successful reservoir-development strategiesmust incorporate the faulting and burial historyto more accurately predict the fault-seal risk.For example, separate tectonic events createnew faults and reactivate existing faults. Fractures may propagate, potentially changingthe reservoir permeability characteristics. Fault-rock properties also change with burialand uplift. Permeability across faults and in surrounding rocks generally decreases withburial depth (below right). Increases in temper-ature boost the rate of quartz precipitation,which can significantly reduce the transmissibil-ity across a fault.

13. Hsieh PA and Bredehoeft JD: “A Reservoir Analysis ofthe Denver Earthquakes: A Case of Induced Seismicity,”Journal of Geophysical Research 86 (1981): 903–920.

14. Wiprut D and Zoback MD: “Fault Reactivation and FluidFlow Along a Previously Dormant Normal Fault in theNorthern North Sea,” Geology 28, no. 7 (2001): 595–598. Zoback MD and Zinke JC: “Production-Induced Normal Faulting in the Valhall and Ekofisk Oil Fields,”http://www.geomi.com/images/PDFs/MDZ-Zinke_PAG_2002.pdf (accessed January 15, 2005).“Wetland Subsidence, Fault Reactivation, and Hydrocarbon Production in the U.S. Gulf Coast Region,”USGS Fact Sheet FS-091-01, http://pubs.usgs.gov/fs/fs091-01/ (accessed January 15, 2005).Hillis RR and Reynolds SD: “In Situ Stress Field, FaultReactivation and Seal Integrity in the Bight Basin,”http://ftp.petroleum.pir.sa.gov.au/products/data/rb2003_2.pdf (accessed January 15, 2005).

15. Bennaceur K, Gupta N, Monea M, Ramakrishnan TS,Randen T, Sakurai S and Whittaker S: “CO2 Capture andStorage—A Solution Within,” Oilfield Review 16, no. 3(Autumn 2004): 44–61.Hawkes CD, McLellan PJ, Zimmer U and Bachu S:“Geomechanical Factors Affecting Geological Storage ofCO2 in Depleted Oil and Gas Reservoirs,” paper 2004–258,presented at the Canadian Petroleum Society 55thAnnual Technical Meeting, Calgary, June 8–10, 2004.

16. Jones RM and Hillis RR: “An Integrated, QuantitativeApproach to Assessing Fault-Seal Risk,” AmericanAssociation of Petroleum Geologists Bulletin 87, no. 3(March 2003): 507–524.

17. Brown A: “Capillary Effects on Fault-Fill Sealing,” American Association of Petroleum Geologists Bulletin87, no. 3 (March 2003): 381–395.

> Capillary pressure diagram. The pressure-depth plot (left) shows thecapillary pressure, Pc, as the difference between the pressures of water andhydrocarbon with depth. The hydrocarbon has a steeper pressure gradientthan the water, so the capillary pressure increases above the free-water level (FWL) where the capillary pressure is zero. The plot on theright shows a typical mercury-injection capillary pressure curve asmeasured in the laboratory. The entry pressure, Pe, is the pressure at whichthe hydrocarbons first enter the sample. A hydrocarbon-column height, Ht,can be trapped below the threshold capillary pressure, Pt, and seals bymembrane sealing. Trap geometries may allow hydrocarbon columns toexceed this height. The hydrocarbon flow across the seal above Ht ispossible at a rate dependent on the relative permeability of the seal.

Dept

h

Pc = Pt

Pc

Ht

Pressure

Free-water level

Hydrocarbonpressure

Hydrostaticpressure

Fractionalmercury [Hg] saturation

Log

capi

llary

pre

ssur

e

Membraneseal

1 0

Hydraulicresistantseal

Swirr

Pt

Pe

Hydraulicseal failure

> Permeability reduction in a cataclastic fault zone with increased burialdepth in three different basin examples. Permeability reductions occur incataclastic fault zones primarily because of mechanical grain crushing andincreased quartz cementation at greater depths. In basins having a highmean-effective-stress, strong cataclasis in fault zones (blue) would occur,resulting in lower permeabilities at greater burial depths. In basins havinglower mean-effective-stresses, moderate cataclasis in faults zones (red)would result, making permeability in those fault zones higher. In basinswhere quartz-cementation occurs in fault zones (green), fault-zonepermeabilities would be higher at shallower burial depths but become veryimpermeable below 3 km [9,840 ft] of depth because of increased quartzcementation. Other factors, such as geologic history and host-rocklithology, play a significant role in determining which processes dictatefault-zone permeability.

Max

imum

bur

ial d

epth

, km

1

2

3

4

50.0001 0.001 0.01 0.1

Average permeability, mD

Basin A - quartz-cementation influence at depth > 3 kmBasin B - moderate cataclasisBasin C - strong cataclasis

Fault-activity maps that color-code the geological timing of structural development helpasset teams quantify the risk of developing aprospect or of taking subsequent developmentsteps, such as initiating an EOR process. Knowledge of the geologic history and its impact is also important when predicting fault-sealing properties.

Fault-Seal Analysis MethodsSuccessful fault-seal analysis methods integratefundamental information on the fault-zone

architecture, fault-rock properties and pressuredata. An important tool for evaluating the flowpotential across a fault is a strike view, or map ofthe fault plane with the hanging wall and foot-wall intersections superimposed on the modeledfault surface.18 Allan diagrams use this tech-nique to show possible fluid-migration pathways,leak points or sealing areas across the fault, and have also helped explain the location ofhydrocarbon/water contacts in various fieldsworldwide. Allan diagrams typically use the seismically interpreted horizons to define the

hanging wall and footwall offset across the faultand lithology interpreted from well logs to identify the stratigraphic changes between theseismic horizons. Sophisticated mapping toolsallow the development of Allan diagrams as 3D models. (left) These models require signifi-cant amounts of data and can be time-consumingto develop, although new software tools, such asthe Petrel workflow tools application, havereduced the processing time significantly.

An alternative to the complicated evaluationof the distribution of the stratigraphy across thefault plane, as used in Allan diagrams, is a simplified juxtaposition triangle diagram, whichenables a quick initial examination and predic-tion of fault-seal capacity. This technique imagesthe hanging wall and juxtapositions for varyingthrows and allows an evaluation of the juxta-posed stratigraphic intervals for a given throw(below left). These diagrams simplify the analy-sis of juxtaposition for a single fault plane. Theeffects of multiple small-throw faults may alsobe quickly evaluated using these diagrams. Thejuxtaposition is simply evaluated at the smallerthrows for each fault.

In the initial analysis, triangle diagramsshow the juxtaposition of the stratigraphy acrossthe fault. Reservoirs juxtaposed against low-permeability rocks such as shales are expectedto seal, whereas reservoir-to-reservoir juxtaposi-tions across the fault are more likely to leak.Juxtaposition diagrams may also be used to eval-uate the fault rocks present and their associatedproperties that develop within the fault zone.For instance, the distribution of clay smearsfrom clay-rich layers in the fault zone can bedetermined and their effects on the seal quanti-fied. Also, critical throws can be assessed whenhigher permeability cataclastic faults may repre-sent a crossfault flow risk. This occurs wheretwo permeable siliciclastic reservoirs are juxta-posed across the fault—one in the hanging wall(HW) and one in the footwall (FW) (next page).

Several methods have been developed toestimate the distribution of fault rocks within afault zone. Two of the most commonly appliedmethods are shale-gouge ratio (SGR) and claysmear.19 Researchers at RDR have also recentlyintroduced a modified SGR, or effective shale-gouge ratio (ESGR), that permits a greatercontrol on the architecture and distribution ofthe fault rocks along the fault surface during the analysis.

The SGR method estimates the percentage ofclay from the host lithology mixed within thefault zone. The algorithm calculates the net claywithin the lithology that is displaced past each

46 Oilfield Review

> Allan diagram. Based on seismic data and wellbore information, Allandiagrams demonstrate the juxtaposition relationships across a fault plane.These diagrams are often used to identify potential petroleum leak points(red) along the fault strike.

Dept

h

Vertical projection plane

Potential leaka b

c d

a b

c d

> Analyzing juxtaposition. A juxtaposition triangle diagram (left) allows simple and quick evaluation of stratigraphic juxtaposition scenarios, for example when a reservoir, Sand A, is juxtaposed againstan impermeable shale that provides a seal. Another scenario might identify a sand-against-sandplacement that fails to provide an adequate seal for hydrocarbon trapping. Lithology is shown on the left; the horizontal axis shows the amount of throw; and the diagonal, dashed arrows show thejuxtaposition scenario at a specified throw and layer. The block diagram (right) is presented to show a 3D representation.

Throw, m0 50 100 150 200

Dept

h

Sand A

Sand ASand A

Triangle Diagram

BlockDiagram

Shale

Shale

SandA

Sand

Sand

Winter 2004/2005 47

point in the fault by taking the sum of the layerthickness times the clay percentage divided bythe fault throw. This calculation is derivedacross a modeled fault surface with a calculatedthrow distribution and clay-percentage estimatesfrom well logs. The ESGR uses a weighted SGRthat allows a nonuniform distribution of theclays within the section dragged past each pointon the fault surface to model a more complex fault-zone process.

Outcrop studies of fault zones have alsorevealed that clay smearing is a common fault-zone process in which clay is smeared along thefault zone from a local shale bed. The thicknessof the clay smear along the fault increases withthe thickness of the source shale bed anddecreases with distance from the source shale.

Multiple shale layers tend to combine to producea continuous smear, enhancing fault sealing.

The basic method for modeling the fault-rockdistributions involves calculating the throw distribution on a gridded fault surface from thehorizon intersections on the fault, infilling thedetailed stratigraphy with the estimated thick-nesses and clay contents, and contouring thederived fault-seal properties onto the fault surface. Contours of capillary pressure measuredalong the fault provide a calibration to the sealing capacity for the estimated fault-rockproperties. These pressure data are oftenacquired in open hole using formation-samplingtools, such as the MDT or RFT devices.

While the calculation of fault-seal potentialacross a fault seems straightforward, it may

be an oversimplification. From outcrop andexhumed fault studies, geoscientists find thatshale smears are not distributed evenly withinfault zones; they may be interrupted, creatingmultiple gaps, which reduce the sealing effectover geologic time scales. One study of the Calabacillas normal fault in New Mexico, USA,

18. Allan US: “Model for Hydrocarbon Migration and Entrapment Within Faulted Structures,” AmericanAssociation of Petroleum Geologists Bulletin 73, no. 7(July 1989): 803–811.

19. Yielding G, Freeman B and Needham DT: “QuantitativeFault Seal Prediction,” American Association ofPetroleum Geologists Bulletin 81, no. 6 (June 1997):897–917.Bretan P, Yielding G and Jones H: “Using CalibratedShale Gouge Ratio to Estimate Hydrocarbon ColumnHeights,” American Association of Petroleum GeologistsBulletin 87, no. 3 (March 2003): 397–413.Yielding et al, reference 2.

> Integration of fault-rock data with juxtaposition triangle diagrams. Incorporating stratigraphic information, log data and coredata (left) with juxtaposition information (right) provides fault-rock types and distribution for a range of fault throws. In thisnormal-fault example, the diagram identifies a smear from impure sandstone from Zone C that forms a seal across from thehigh-permeability sandstone reservoir in Zone E.

SandNet/Gross

mD

Nophyllosilicate

data

Reservoir againstolder stratigraphyin footwall

Reservoir againstyounger stratigraphyin hanging wall

The leaky zone has been sealedby a potential phyllosilicate smearof Reservoir Unit C

High-permeabilitysandstoneSandstoneImpure sandstonePhyllosilicate-rich(including shale)Cemented sandstone

Host-Rock Lithology

Cataclastic seals (sand on sand)Cataclastic seals (sand on sandpast other lithologies)Cataclastic seals (high-permeability sand onhigh-permeability sand past other lithologies)Cataclastic on phyllosilicate frameworksmear seals (impure sands in FW or HW)High potential for cataclastic on phyllosilicateframework smear (impure sand past other lithologies)Cement seal (cemented in FW or HW)High potential for cement seal (cementedsection past other lithologies)Phyllosilicate smear (phyllosilicate-rich units or shale in FW or HW)High potential for phyllosilicate smear(phyllosilicate-rich units past other lithologies)

Fault-Rock/Seal Type

PhyllosilicatePlug Permeability

Host-Rock

Lithology

3,60

0

Core

dep

th, m 3,

580

3,56

0A

3,54

03,

520

3,62

03,

640

3,66

03,

680

B

C

D

E

F

G

A

B

C

D

E

F

G

Hang

ing

wal

l (HW

) res

ervo

ir lit

holo

gy

0 20 40 60 80 0 10 100 1,000%

0.5

Zone

A

B

C

D

E

F

G

1007550250 125 150Fault throw, m

Plug Porosity, %0 10 20 30 40 50

Foot

wal

l (FW

) res

ervo

ir lit

holo

gy

0 1.0

found that clay smears tend to be continuous fora distance of two to six times the clay source-bedthickness, but then thin significantly away fromthe base of the clay source-bed on the footwall.20

Moreover, smears are frequently breached by small-throw faults. Consequently, smear-estimation and seal-calibration techniques canoverestimate fault-seal potential, especially nearthe base of a clay source-bed.

The contours of capillary pressure and fault-rock property estimates over a fault surface areundercalibrated using the methods described. Amore accurate analysis should include the calibration of fault-rock properties estimatedfrom core measurements. Measured thresholdpressure and permeability across small faults incore help predict the sealing capacity and flowproperties of the estimated fault-rock distribu-tion. Fault rocks in core also define the range offault-rock types, created by processes such ascataclasis or grain crushing, and allow evalua-tion of the impact of the geologic history andfault timing.

Fault-rock databases from specific basins arekey to the calibration of the fault-rock sealingpotential. Fault-rock data are a crucial input tosuccessful reservoir simulations, which also relyon field data, including seismic surveys, well logs,core logs and studies, and field pressure-data.These data are also important in reducing therisk in an exploration setting, where there maybe significantly less data available.

Increased Knowledge, Reduced UncertaintyFaults in core provide not only a calibration to fault-rock properties such as porosity, permeability and threshold pressures, but alsofault distribution and density at a scale belowthat of seismic resolution. Recent advances inseismic interpretation methods, such as automatic fault-picking and attribute-mappingsoftware, help geophysicists interpret large seis-mic volumes in less time and in greater detailthan manual methods. However, much of thefault detail still exists at a scale below the seis-mic resolution, so detection of these small faultsmust rely on high-resolution borehole-imagingtools and the detailed study of fullbore cores.

The highly compartmentalized Hibernia field in the Jeanne d’Arc basin offshore Newfoundland, Canada, demonstrates theimportance of detailed core examinations.21 TheHibernia field is situated in a sedimentary basinwithin the greater Jeanne d’Arc basin that hasundergone multiple rifting events associatedwith the breakup of the supercontinent Pangeaand the formation of the Atlantic Ocean fromthe late Triassic to the early Cretaceous period.

Since first production in 1997, geologists andengineers with the Hibernia Management andDevelopment Company knew that the two mainHibernia reservoirs were compartmentalized byfaults. An estimated 30 fault blocks were identi-fied from observed variations in fluid-contactheights and in pressures. As development of thefield continued, there were indications that thefield might be even more compartmentalizedthan originally thought.22 However, the assetteam was uncertain about the degree to whichthe faults were diminishing individual well production and injection performance.

Fullbore cores were taken from the lowerreservoir in the hanging-wall section in two wells,the B-16 2 in Block Q and the B-16 4 in Block R,to characterize the deformation and the fault-zone architecture (above). The cores wereexamined for geologic structures, and sampleswere collected for analysis of microstructural andpetrophysical properties. The fault rocks wereclassified according to clay content.

Fault rocks with less than 15% clay exhibitedboth disaggregation bands, which are localizedzones of particulate flow with little grain fracturing, and deformation bands with cata-clastic seams with variable amounts of grain-sizereduction due to mechanical crushing of thegrains. Despite the lack of clay, these fault rockshave an average permeability of 0.06 mD, whichis almost five orders of magnitude lower than

the host-rock permeability. Fault rocks contain-ing an intermediate amount of clay—15 to40%—are classified as phyllosilicate-frameworkfault rocks, and they exhibited even lower permeabilities than their low-clay counterparts.The high clay-content rocks, characterized bygreater than 40% clay, formed clay smears.These fault rocks typically have permeabilitiesof less than 0.001 mD, equivalent to the host-rock properties. The analysis of the fault rocksin core showed that the fault-rock types arecapable of significantly reducing the permeabil-ity across the faults in Hibernia field.

To evaluate the sealing potential of the faultscompartmentalizing the reservoir, the fault-rocktypes and properties are integrated with thefault-rock distribution estimates from the juxta-position diagrams. These diagrams show thatwhere the fault throw is less than the individuallayer thickness and the reservoir is juxtaposedagainst itself, the sealing properties are dictatedby the cataclastic fault-rock properties. Conversely, where the fault throw exceeds theindividual layer thickness, juxtaposition sealingof reservoir against nonreservoir rock is theprincipal seal.

A juxtaposition triangle diagram of theHibernia formation at Well B-16 2 demonstratesthe predicted fault-rock distributions and theirinterpreted effects on the fluid flow (next page).

48 Oilfield Review

Fault trendN-SW-dip

E-dip

NW-SENE-dip

SW-dip

Compartments

JI

H

E

K

C

FD

PM

L NA

B

QR

S

V

T

XY Z

G

FF

EE

AABB WCC

DD

O

Bonavistaplatform

Mur

re fa

ult

Nautilus fault

P

V

T

Q R

S

B-16 16

B-16 17

B-16 2B-16 4

1.2

km0 2

0 miles

N

Hibernia

St. John’s

Halifax

CANADA

USA> Hibernia structural map and the location of twocored wells, the B-16 2 well and the B-16 4 wellin Blocks Q and R, respectively.

Winter 2004/2005 49

20. Doughty PT: “Clay Smear Seals and Fault Sealing Potential of an Exhumed Growth Fault, Rio Grande Rift,New Mexico,” American Association of Petroleum Geologists Bulletin 87, no. 3 (March 2003): 427–444.

21. Porter JR, McAllister E, Fisher QJ, Knipe RJ, CondliffeDM, Kay MA, Stylianides G and Sinclair IK: “Impact ofFault-Damage Zones on Reservoir Performance in theHibernia Oilfield (Jeanne d’Arc Basin, Newfoundland):An Analysis of Structural, Petrophysical and DynamicWell Test Data,” special paper 43 in Hiscott R and Pulham A (eds): Petroleum Resources and Reservoirs ofthe Grand Banks, Eastern Canadian Margin. St. John’s,Newfoundland, Canada: Geological Association ofCanada (2004): 129–142.

22. Gormley JR, Andrews RJ, Baskin DK and Stokes R: “An Integrated Study of Reservoir Compartmentalizationin the Hibernia Formation, Hibernia Field,” Abstracts,Vol 26. Geological Association of Canada, MineralogicalAssociation of Canada Annual Meeting, St. John’s, Newfoundland (2001): 52–53.

> Juxtaposition (top) and fault-seal (bottom) diagrams for Hibernia Well B-16 2. The juxtaposition diagram identifies a sand-against-sand juxtaposition at a75-m [246-ft] throw. In this scenario, a Layer 2 sandstone in the hanging wall (HW) is dragged past other lithologies—impure sandstones and a medialshale—in the footwall (FW) and is juxtaposed against the basal sandstone in the upper Layer 3 interval. This represents a possible leak area. However,when clay smearing is taken into account, the predicted potential leak area is reduced significantly.

Layer 2

Layer 1

Medial shale

Uppersandstone

Basalsandstone

Shale

Shale

Uppe

rM

iddl

eLa

yer 3

Sandstone

Shale

Sandstone

Basa

l

Shale-rich against shale-rich pastother lithologies

Sand in FW and HWSand against sand past other lithologiesImpure sand in FW and HWImpure sand against impure sandpast other lithologiesShale-rich in FW and HW

Juxtaposition

Note: At 75-m throw, Layer 2sandstone juxtaposed againstUpper Layer 3 basal sandstone

0 100 200Fault throw, m

Lithofacies (volume-shale derived)Sandstone-dominatedMixed heterolithicShale/mudstone dominated

Sand in FW and HWSand against sand past other lithologiesImpure sand in FW and HWImpure sand against impure sandpast other lithologiesHigh potential for PFFR(shale-rich in FW and HW)Phyllosilicate-rich fault rocks (clay smears)High potential for phyllosilicate-rich(clay smears)

Fault-Seal Types (using clay-smear factor 3.0)

Cataclasites

Phyllosilicate-frameworkfault rocks

(PFFR)

Phyllosilicate-rich

Note: Reduction in windowsize due to smear potential ofintervening impure sandstonesand medial shale

Layer 2

Layer 1

Medial shale

Uppersandstone

Basalsandstone

Shale

Shale

Uppe

rM

iddl

eLa

yer 3

Sandstone

Shale

Sandstone

Basa

l

0 100 200Fault throw, m

This diagram shows that for throws less than 30 m [98 ft], fault rocks are predominantly cata-clasites or zones of grain crushing. On the otherhand, where throws are greater than 30 m, thelower permeability, clay-rich, phyllosilicate-richfault rocks are present. These results show thatfault rocks in the Hibernia field have the poten-tial to degrade the performance of bothproduction and injection wells.

When combined with production history-matching models, which yield nonuniquesolutions from many possible geologic scenarios,the fault-seal analysis calibrated to the faultdata from core bolstered the interpretation ofhow faults affect fluid flow in the field. This ledto the drilling of the injector well B-16 21, whichwas positioned to avoid dangerous fault-damagezones. The new injector well improved reservoirsweep and provided additional pressure supportfor nearby producing wells.

Fault-Seal Analysis Aids DrillingOpen, conductive fault systems may be as chal-lenging as sealing faults in field development,especially where they pose a serious drilling hazard. Since development drilling began in1970, the highly faulted Prudhoe Bay field,Alaska, USA, has produced more than 10 billionbarrels [1.6 billion m3] of oil. Throughout thefield’s history, lost-circulation problems havebeen commonplace and directly related to thenumber of faults crossed while drilling wells.With substantial remaining recoverablereserves, continued development by BP andConocoPhillips requires drilling into smaller

fault blocks and through more faults. As a result,lost-circulation problems have increased dramatically, even as total drilled footage hasdecreased in recent years (above).

Problems reached critical levels in 1998,when 66 out of 120 wells and sidetracks experi-enced lost-circulation problems, costing overUS$10 million. Trouble-time costs added 50% to100% to well costs. In some cases, loss ratesexceeded 1,000 bbl/hr [159 m3/hr], raising serious safety concerns and risking the loss ofwells. BP and ConocoPhillips, then Arco Alaska,considered several options to address the fault-related lost-circulation problems. The Prudhoe Bay asset team could choose not to drillrisky targets, reducing development options andrecoverable reserves, or could employ expensivedrilling contingencies that may have mitigatedthe problem, but at the expense of understandingits cause.

The Prudhoe Bay partner companies, alongwith RDR, decided to investigate the cause ofthese lost-circulation problems—faults that actas conduits for drilling mud. In Prudhoe Bay field,more than 5,400 faults have been interpreted byseismic surveys. The faults range in strike lengthfrom 500 to 15,000 ft [152 to 4,570 m] with throwsfrom 20 to 200 ft [6 to 60 m] (next page, top).First, the existing seismic data were reprocessedto improve the fault interpretation. The mappedfaults were then added to a database, whichincluded fault parameters such as orientationand length. Along with geologic data, drilling datafor all wells in the field were compiled, includinglost-circulation volumes and rates, and the

location of losses. Wellbore data and productionhistory-matching were also used to gain a morethorough understanding of fault, fluid and reser-voir behavior. Although this analysis helpedexplain 80% of the lost-circulation problems, it showed that a more detailed exploration offault-rock properties across the Prudhoe Bay field was warranted.

Analysis of the fault distributions and fault-rock properties from thousands of feet of corefrom 14 wells provided the necessary calibrationto evaluate fault behavior. Open vuggy fracturesin the core identified conductive zones thatcould pose potential drilling hazards. Full-fieldand local-stress modeling, integrated with the tectonic history, showed a preferential orientation of conductive faults parallel to themaximum in-situ stress direction. An integrateddatabase of fault styles and architecture, fault-rock properties, and lost-circulation datafacilitated the study of fault-damage zones andthe analysis of fault sealing.

The database properties calibrated to juxtaposition and fault-rock distributions fromclay-content of individual faults helped reducethe risk of drilling development wells in PrudhoeBay field. Predrill well planning now incorporatesthe data from the database to avoid hazardousdrilling areas (next page, bottom).

In the year following this integrated fault-characterization project, 65 wells and sidetrackswere drilled. The number of problematic wells,those losing more than 100 bbl [16 m3] ofdrilling fluid, dropped from 32 to 16% of the totalwells. Lost-circulation zones were anticipatedand accounted for, reducing trouble-time anddecreasing drilling costs by US$2 to 5 millionduring that year. Only two wells had significantproblems. A more thorough knowledge of faultsin Prudhoe Bay field reduced drilling risk,improved well planning and increased assetteam confidence in further development. Thesignificant reduction in drilling risk has openedup more drillable targets that were once deemed too risky, while potentially increasingrecoverable reserves.

Complex Problem, Simple AnswerFaults and their influences on fluid flow withinreservoirs are complex. Technological advance-ments have improved our ability to measurethese influences, both directly and indirectly.Well-testing techniques, production history-matching and the injection of tracers, forexample, help assess whether reservoir compart-ments exist, and if they do, whether they are incommunication or are isolated. Wellbore

50 Oilfield Review

> Increase in lost-circulation problems in Prudhoe Bay field. Even as totaldrilled footage (blue) declined during the last two decades, lost-circulationproblems (red) became more serious and costly. As smaller fault blockswere drilled using horizontal wells, more faults were crossed than duringearly development drilling. It was observed that the increased horizontalfootage drilled in the late 1990s directly correlated to mud losses, identifyingfaults as the predominant source of drilling problems.

Tota

l dril

led

foot

age,

ft

140,000

120,000

100,000

80,000

60,000

40,000

20,000

0

1,600,000

1,400,000

1,200,000

1,000,000

800,000

600,000

400,000

200,000

0

Tota

l los

t circ

ulat

ion,

bbl

1965 1970 1975 1980 1985 1990 1995 2000Year

Total drilled footageTotal lost circulation

Winter 2004/2005 51

measurements and sampling tools also are usedto evaluate reservoir rocks, fluids and pressuresto determine compartmentalization. Recently,engineers have successfully identified fluid compositional variances related to compartmen-talization using the Schlumberger MDT device.23

The evaluation, calibration and prediction of thefaults that compartmentalize reservoirs requirea systematic analysis that should include inte-grating datasets from properties measured inconventional core, to subsurface well and production data, seismic interpretation and outcrop and subsurface analogs.

Poorly resolved subsurface fault complexitiesmay be incorporated into reservoir fluid-flow simulators using the results from detailedoutcrop analog studies. In simulators, the effectsof faults are represented as effective trans-missibility factors across defined traverses.Fault-related transmissibility depends on thenumber of faults, the thickness of the associateddamage zones and the fault properties, such asfault-rock permeability and pressure thresholds.

Incorporating the fault-rock properties fromdatabases has improved history-matching andfluid-flow modeling across faults.24 These resultsstill contain risk and uncertainty. In fault-sealanalysis, there will always be uncertainties relating to the internal architecture of faults,the host-rock properties, the definition of stratigraphic units from seismic surveys, capillary pressure effects and how far to projectthe model given the limited amount of well data.Fault-rock property databases provide the rangeand magnitude of the uncertainty that can beincorporated into risk modeling using MonteCarlo techniques, for example.

In fault-seal analysis, the complexity of faultsmust be captured and modeled, but the answermust be simple enough to be used effectively inreservoir simulations to reduce uncertaintywhen exploring and developing enigmatic,faulted siliciclastic reservoirs. —MGG

23. Mullins OC, Hashem M, Elshahawi H, Fujisawa G, Dong C, Betancourt S and Terabayashi T: “HydrocarbonCompositional Analysis In Situ in Openhole Wireline Logging,” Transactions of the SPWLA 45th Annual Logging Symposium, Noordwijk, The Netherlands, June 6–9, 2004, paper FFF. Fujisawa G, Betancourt SS, Mullins OC, Torgerson T,O’Keefe M, Terabayashi T, Dong C and Eriksen KO:“Large Hydrocarbon Compositional Gradient Revealed byIn-Situ Optical Spectroscopy,” paper SPE 89704, presentedat the SPE Annual Technical Conference and Exhibition,Houston, September 26–29, 2004. Elshahawi H, Hashem M, Mullins OC, Fujisawa G, Dong C,Betancourt S and Hegeman P: “In-Situ Characterization

of Formation Fluid Samples: Case Studies,” paper SPE90932, presented at the SPE Annual Technical Conferenceand Exhibition, Houston, September 26–29, 2004. Betancourt S, Fujisawa G, Mullins O, Carnegie A, Dong C, Kurkjian A, Eriksen KO, Haggag M, Jaramillo ARand Terabayashi H: “Analyzing Hydrocarbons in the Borehole,” Oilfield Review 15, no. 3 (Autumn 2003): 54–61.

24. Knai TA and Knipe RJ: “The Impact of Faults on FluidFlow in the Heidrun Field,” in Jones G, Fisher QJ andKnipe RJ (eds): Faulting, Fault Sealing and Fluid Flow inHydrocarbon Reservoirs: Geological Society SpecialPublication 147. Bath, England: The Geological SocietyPublishing House (1998): 269–282.

> Fault map of Prudhoe Bay field. The structural complexity of Prudhoe Bay field is demonstrated bya fault map that shows extensive faulting throughout much of the field.

Prudhoe Bay Fault Map

Prudhoe Bay

Anchorage

AlaskaCANADAUSA

5

km0 8

0 miles

> Improved well planning. With improved knowledge of fault behavior inPrudhoe Bay field, high-loss areas (pink) can now be avoided or plannedfor while drilling to reach the target. Specific information, such as shale-gouge ratio along individual faults, allows the asset team to identify theoptimal points at which to cross faults to minimize mud losses and reducedrilling costs.

Kickoff point

Targ

et

Example ofprevious well plan

Plan forlosses

Crossfault at highshale-gouge ratio

Targ

et

Plan View