AUTHORS

Likuan Zhang Key Laboratory of Petro-leum Resources Research Institute of Geologyand Geophysics Chinese Academy of SciencesBeijing 100029 China zhanglikuan1979163com

Likuan Zhang is currently a postdoctoral re-searcher at the Institute of Geology and Geo-physics Chinese Academy of Sciences Hereceived a BA degree in petroleum geologyfrom Jilin University China an MS degree fromNorthwestern University China and a PhDfrom the Institute of Geology and GeophysicsChinese Academy of Sciences His researchinterests concern fluid flow in fault zonesnumerical basin modeling and hydrocarbonmigration in petroleum systems

GEOHORIZONS

Quantitative evaluation ofsynsedimentary fault openingand sealing properties usinghydrocarbon connectionprobability assessmentLikuan Zhang Xiaorong Luo Qianjin Liao Wan YangGuy Vasseur Changhua Yu Junqing Su Shuqin YuanDunqing Xiao and Zhaoming Wang

Xiaorong Luo Key Laboratory of Petro-leum Resources Research Institute of Geologyand Geophysics Chinese Academy of SciencesBeijing 100029 China luoxrmailigcasaccn

Xiaorong Luo is a research scientist in the Chi-nese Academy of Sciences and has BS andMS degrees in geology from NorthwesternUniversity China and a PhD in geophysics fromthe University of Montpellier France His re-search in the last 30 years has been in petro-leum geology currently focusing on numericalmodeling geopressuring and hydrocarbonmigration and accumulation

Qianjin Liao Petroleum Exploration andDevelopment Research Institute of DagangOilfield Company Ltd Tianjin 300280 China

Qianjin Liao received his BA degree in geologyfrom Southwestern Petroleum Institute and hisMS degree in geochemistry from the Institute ofExploration and Development PetroChina He iscurrently a manager of the Petroleum Explorationand Development Research Institute of DagangOilfield Company His research interests are re-lated to the evaluation of hydrocarbon resourcesand hydrocarbon systems

Wan Yang Key Laboratory of PetroleumResources Research Chinese Academy of Sci-ences IGG Beijing 100029 China present ad-

ABSTRACT

Hydraulic behaviors of faults in sedimentary basins have beenpaid close attention in studies of hydrocarbon migration andaccumulation because of their important functions in basinhydraulic circulations In previous studies however the func-tion of faults in hydrocarbon migration is characterized bythe sealing capacity of faults In fact sealing is only an impres-sive and time-dependent aspect of the hydraulic behavior offaults which may act as seals during some periods and as path-ways some time later Therefore in hydrocarbon migrationstudies sealing indices may successfully be used in some casesbut not in others In this article we introduce an empiricalmethod (termed the fault-connectivity probability method)for assessing the hydraulic connecting capacity of a fault forhydrocarbon migration over geological time scales The meth-od is based on the recognition that observable hydrocarbon inreservoirs should result from the opening and closing behaviorof the fault during the entire process of hydrocarbon migrationIn practice the cumulative petroleum migration through a seg-ment of the fault zone is identified by the presence (or not) ofhydrocarbon-bearing layers on both sides of the segmentData from the Chengbei step-fault zone (CSFZ) in the Qi-kou depression Bohai Bay Basin northeast China were used

dress Department of Geology Wichita StateUniversity Wichita Kansas 67260

Wan Yang obtained a PhD from the Universityof Texas at Austin in 1995 and is currently anassociate professor at Wichita State University

Copyright copy2010 The American Association of Petroleum Geologists All rights reserved

Manuscript received July 2 2009 provisional acceptance September 9 2009 revised manuscriptreceived November 25 2009 final acceptance December 14 2009DOI10130612140909115

AAPG Bulletin v 94 no 9 (September 2010) pp 1379ndash 1399 1379

teaching sedimentology and terrigenous clasticdepositional systems He has worked on depo-sitional systems analysis sequence and cyclo-stratigraphy paleoclimatology and reservoircharacterization of marine and nonmarine silici-clastic and carbonate rocks in China and theUnited States

Guy Vasseur SISYPHE UMR7619 Univer-sity Pierre et Marie Curie 75252 Paris CedexFrance

Guy Vasseur received an engineering degreefrom Ecole Polytechnique (Paris) in 1963 and aDoctorat drsquoEtat degree (PhD) from the Uni-versity of Paris in 1971 He then became a seniorscientist of Centre National de la RechercheScientifique and worked at the Universities ofMontpellier 2 and Paris 6 in various domainsof geophysics His main domain of competenceis the modeling of heat and mass transfer inthe Earthrsquos crust

Changhua Yu Petroleum Exploration andDevelopment Research Institute of DagangOilfield Company Ltd Tianjin 300280 China

Changhua Yu is an engineer at the PetroleumExploration and Development Research Instituteof Dagang Oilfield Company He received a BSdegree from Jianghan Petroleum Institute Chinain 1998 His works focus on the structural con-trols on fluid flow in petroleum systems

Junqing Su Petroleum Exploration andDevelopment Research Institute of DagangOilfield Company Ltd Tianjin 300280 China

Junqing Su got an MS degree in geology fromYangtze University China He did geological re-search for the Petroleum Exploration and De-velopment Research Institute of Dagang OilfieldCompany and has more than 20 years of ex-ploration experience His research interestsfocus on 3-D seismic interpretation and compre-hensive analysis of petroleum systems

Shuqin Yuan Key Laboratory of PetroleumResources Research Chinese Academy of Sci-ences IGG Beijing 100029 China present ad-dress Petroleum Exploration and DevelopmentResearch Institute of Dagang Oilfield CompanyLtd Tianjin 300280 China

Shuqin Yuan is a senior expert in the PetroleumExploration and Development Research Insti-tute of Dagang Oilfield Company She has anMS degree in geology from Yangtze University

1380 Geohorizons

to develop this method Fluid pressure in mudstones normalstress perpendicular to fault plane and shale gouge ratio areidentified as the key factors representing fault-seal capacityThey are combined to define a nondimensional fault openingindex (FOI) The values of FOI are calculated from the mea-sured values of the key factors and the relationship betweenFOI and fault-connectivity probability on any fault segmentis established through statistical analysis Based on the datafrom the CSFZ when the FOI is less than 075 the fault-connectivity probability is 0 when FOI ranges from 075 to325 the corresponding fault-connectivity probability in-creases from 0 to 1 following a quadratic polynomial relation-ship when FOI is greater than 325 the fault-connectivityprobability is 1 The values of fault-connectivity probabilitycan be contoured on a fault plane to characterize the variationsof hydraulic connective capacity on the fault plane The appli-cability of this concept for other oil fields (in particular thequantitative relationship between FOI and fault-connectivityprobability) has still to be ascertained

INTRODUCTION

Faults may act as either barriers to fluid flow or conduits (Smith1966 Karlsen and Skeie 2006) Fluid flow along faults causeschanges in temperature pressure and stresses within and sur-rounding fault zones (Sibson 1981Vasseur andDemongodin1995 Xie et al 2001Gratier et al 2002)Obviously the open-ing and closing behavior of a fault controls hydrocarbon mi-gration and accumulation and therefore the distribution of hy-drocarbon fields in petroliferous basins (Selley 1998 Hardingand Tuminas 1989)

Studies on faults serving as flow barriers or conduits (Knipe1989 Sorkhabi and Tsuji 2005) have been conducted throughoutcrop observations lab experiments and numerical simula-tions to characterize the geometry composition deformationand diagenesis of fault zones In fault zones several observedphysical and chemical properties have been proposed as proxydata to evaluate the timing and duration of fault opening andclosing (Smith 1966Weber et al 1978 Engelder 1979 Allen1989 Knipe 1992 1997 Antonelini and Aydin 1994 Gibson1994 Berg and Alana 1995 Sorkhabi and Tsuji 2005)

A variety of approaches have been used to characterizethe sealing ability of faults to study their impact on fluid flowPerkins (1961) emphasized that the juxtaposition of sand-stones may destroy the sealing capability of faults Howeverhe noticed that faulted sandstones may be sealed by mudstones

China Her research focuses on hydrocarbonsystems and the potential of hydrocarbon playsin petroliferous basins

Dunqing Xiao Petroleum Exploration andDevelopment Research Institute of DagangOilfield Company Ltd Tianjin 300280 China

Dunqing Xiao received a BA degree in geologyfrom the China University of Geoscience He isa principal geologist in the Petroleum Explora-tion and Development Research Institute ofDagang Oilfield Company with research inter-ests on geological conditions of oil and gas ac-cumulation and exploration of oil and gas plays

Zhaoming Wang Research Institute ofExploration and Development PetroChina Beijing100083 China

Zhaoming Wang received a BS degree in petro-leum geology from the China University of Pe-troleum Dongying and an MS degree and PhDfrom the Institute of Geology and GeophysicsChinese Academy of Sciences He has workedin the Shengli oil field Currently he is working inthe Institute of Exploration and DevelopmentPetroChina His interests are in the study of re-source assessment of sedimentary basins andhydrocarbon migration

ACKNOWLEDGEMENTS

This study was supported partly by the ChineseNational Major Fundamental Research Devel-oping Project (2006CB202305) and partly by theChinese National Natural Science Foundation(40772090) The study could not have beenpossible without support from Dagang Oil Com-pany Ltd We appreciate the permission fromDagang Oil Company to publish this work Wethank Beicip-Franlab for providing the Temis3Dsoftware used for the pressure modeling Weappreciate the comments and encouragementfrom A C Aplin and three reviewers StephenNaruk Barry J Katz and Alexander A Kitchkawhich have helped improve the manuscriptThe AAPG Editor thanks the following reviewersfor their work on this article Barry J KatzAlexander A Kitchka and Stephen J Naruk

squeezed into the sandstones during faultingAllen (1989) statedthat stratal discontinuity across a fault must be studied threedimensionally Smith (1966) introduced some theoretical con-siderations on sealing and nonsealing faults based on the cap-illary model of Hubbert (1953) To characterize the sealingheterogeneity along fault planes several parameters such asclay smear potential shale smear factor and shale gouge ratio(SGR) were proposed to estimate the degree of mudstonesmearing within fault zones and to evaluate the seal capacityof faults (Bouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) Using outcrop observations Hasegawa et al(2005) and Kachi et al (2005) designed several seal capacityparameters to describe the three-dimensional (3-D) sealingcharacteristics of faults and emphasize the heterogeneity offault-seal capacity These detailed studies have pushed theconcept of fault-seal capacity from qualitative toward quanti-tative (Sorkhabi and Tsuji 2005) Howevermost of these pre-vious studies are based on the present characteristics of a faultwhereas hydrocarbon accumulation involves the sealing char-acteristics of the faults during the entire pastmigration processIndeed this entire migration process may be very long be-cause of the limited hydrocarbon migration flux through faultzones (Byerlee 1993 Haney et al 2005) andor because ofthe very slow maturation of hydrocarbons in the source rocks

In reality active tectonics associated with faults are discon-tinuous processes (Sibson et al 1975 Hooper 1991) andfaults open and close episodically as a result of a variety of geo-logic processes (eg earthquake cycles Sibson 1981 Hooper1991) With respect to their hydrodynamic properties faultshave aubiquitous property theymay act either as seals or as pref-erential pathways for fluid flow (Yielding et al 1999) There-fore any comprehensive study of hydrocarbon migrationthrough faults should take this dual behavior into account as wellas its temporal evolution Indeed it is widely accepted that in afault zone fluid can flow in pulses from one compartment to an-other in relation with the earthquake rupture of existing seals(Byerlee 1993 Lockner and Byerlee 1995 Haney et al 2005)

Hydrocarbon flow occurring during fault opening shouldleave some indications that can be used to identify whetherandwhere the fault was open (Sorkhabi andTsuji 2005) Indi-cators such as veins related to faulting and diagenetic miner-alogy and fluid inclusions within and adjacent to fault zonesare equivocal in many cases and require extensive testing andanalyses many samples This is feasible for outcrop studies butis difficult for core analysis because of the scarcity of cores in faultzones Other types of proxy are needed to assess fault openingand closing and its seal capacity

Zhang et al 1381

The purpose of this study is to propose suchproxies which can be used to characterize the sealcapacity of faults in the Chengbei step-fault zone(CSFZ) Bohai Bay Basin China Our approach instudying fault-seal capacity is to consider the long-term hydrocarbon migration process as a cumula-tive effect of many small increments of migratinghydrocarbons through a segment of a fault zonesuch as the migration episodes that occur duringearthquakes The presence of hydrocarbon accumu-lations in footwall and hanging-wall reservoirs istherefore used as a calibration criterion indicatingcumulative fault opening and closing cycles for mi-gration over geological time A method of quanti-tative assessment of 3-D fault opening and closingat different locations on a fault plane is proposedand the opening and closing behavior on a faultplane during the entiremigration historymay semi-quantitatively be described using a concept of fault-connectivity probability

1382 Geohorizons

GEOLOGICAL SETTING OF THE CHENGBEISTEP-FAULT ZONE

Structure Stratigraphy and Petroleum System

The CSFZ is located on the slope between theChengning uplift and Qikou depression in theBohai Bay Basin northeast China (Figure 1) It cov-ers an area of approximately 550 km2 (212 mi2)with threefifths of the field offshore The struc-tural framework is a gentle slope cut by multiplestep-like normal faults in the Cenozoic strata (Wanget al 2003) The CSFZ as well as the entire BohaiBay Basin has experienced a rifting episode com-posed of two phases a Paleogene synrift phase anda Neogene to Quaternary postrift phase (Li et al1998) The tectonic evolution is related to strike-slip movements along the Tanlu fault and Qinlingstructural zone (Xu 1980 Yang andXu 2004) TheCenozoic strata are composed of continental clastic

Figure 1 Location map of the Chengbei step-fault zone as outlined by a square on the right panel on the southern slope of the Qikoudepression in Bohai Bay Basin eastern China

deposits with a total thickness of 2000ndash5000 m(6562ndash16404 ft) (Figure 2) The Paleogene synriftdeposits include the Shahejie and Dongying forma-tions The Neogene to Quaternary sequence con-sists of the Guantao Minghuazhen and Pingyuanformations A regional unconformity separatesPaleogene and Neogene strata (Yuan et al 2004)

Oilndashsource rock correlation shows that hydro-carbons in the CSFZ were derived from sourcerocks composed of dark-colored mudstones of theSha-1 and Sha-3 members of the Shahejie Forma-tion in the Qikou depression (Wang et al 2006)(Figure 2) The major reservoir rocks include fandeltaic and lacustrine littoral sandstones of theSha-2 and Sha-3 members of the Shahejie Forma-

tion delta-front and prodeltaic sandstones of theDong-1 andDong-3members of the Dongying For-mation and Neogene fluvial sandstones (Yuanet al 2004) (Figure 2) Thick mudstones in themiddle Sha-1 and Dong-2 members and lowerMinghuazhen Formation constitute regional sealrocks (Yu et al 2006) (Figure 2)

Petroleum exploration in the CSFZ began in1965 During the last 43 yr 11 oil and gas fieldshave beendiscovered (Figure 3)More than 140 ex-ploratory wells have been drilled and a 3-D seis-mic survey covers nearly all major structures Alarge quantity of exploration and developmentdata are available to study fault-seal capacity inthe CSFZ (Figure 3)

Figure 2 Summary diagram showing chrono- and lithostratigraphy tectonic stages and characteristics of the petroleum system in theChengbei step-fault zone (after Yuan et al 2004)

Zhang et al 1383

Movement of Faults andHydrocarbon Expulsion

The study area has experienced regional transten-sion since the end of theMesozoic (Li et al 1998)resulting in extensive faulting Major faults whichcontrolled the structural framework and deposi-tional history include the Qidong (QD) Zhang-donghai 4 (ZDH4) Zaobei (ZB) Yangerzhuang(YEZ) Yangerzhuangnan (YEZN) and Zhangbei(ZHB) faults (Yuan et al 2004 Yu et al 2006)(Figure 3) Seismic interpretation indicates that theyare synsedimentary faults (Figure 4) In general

1384 Geohorizons

these faults are parallel to each other striking in anortheast or nearly eastndashwest direction and dippingtoward the depression to the north The distribu-tion of discovered hydrocarbon accumulations isclosely related to the faults (Figure 3) These faultsappear to be important factors in hydrocarbon mi-gration and accumulation

In this study the incremental movements alonga given fault are assumed to be the result of multi-ple tectonic episodes in the CSFZ The fault growthindex defined as the ratio between the thicknessesof the same formation measured on both sides ofthe fault plane (Childs et al 2003) was used to

Figure 3 Faults oil fields key wells used in this study (black dots) other wells (black circles) and 3-D seismic data sets (dashed linerectangles) in the Chengbei step-fault zone Faults are mapped at the top of the Shahejie Formation Section AAprime is presented in Figure 4QD = Qidong YEZ = Yangerzhuang YEZN = Yangerzhuangnan ZB = Zaobei ZDH4 = Zhangdonghai 4 ZHB = Zhangbei

analyze the development and activity of growthfaults in the study area (Zhang et al 2007) Theyrevealed the occurrence of three major episodesof intensive faulting during the deposition of theShahejie Donying and upper Minghuazhen for-mations respectively The tectonic intensity hadbeen decreasing from the Paleogene to theNeogene

Basin modeling applied to the whole region(Figure 3) indicates that hydrocarbonwas expelledfrom source rocks during twomajor episodes (Wanget al 2006 Yu et al 2006) The first episode wasat the end of the Dongying (sim26Ma) when sourcerocks in the Sha-3member becamematureHydro-carbon expulsion ceased because of regional uplift-ing and erosion at the end of the Paleogene There-fore this early episode of hydrocarbon expulsionwas short and the amount of expelled hydrocar-bon was limited Major expulsion occurred duringthe second episode from the late Miocene to Qua-ternary (Wang et al 2006) The deposition of theNeogene Guantao and Minghuazhen formationscaused maturation and hydrocarbon generationin the Sha-3 and Sha-1 source rocks Hydrocarbonexpulsion peaked at the end of the Minghuazhen(sim2 Ma)

FAULT CONNECTIVITY ANDPROBABILISTIC INTERPRETATION

We characterized the fault opening or closing on astatistical basis and developed the concept offault-connectivity probability

Fault Opening and Closing

A fault is not a simple discontinuity but instead azone consisting of a fault core and a surroundingdam-age zone (Chester and Logan 1986 Smith et al1990 Forster and Evans 1991 Chester et al 1993Bruhn et al 1994 Caine et al 1996) (Figure 5)The fault core represents the part of a fault zonewhere most of the displacement is accommodatedand consists of slip surfaces fault gouge brecciasand cataclasites (Chester and Logan 1986 Chesteret al 1993 Caine et al 1996 Clausen et al2003) The damage zone may be composed of sub-sidiary faults faulted rocks and fractures (Chesterand Logan 1986 Chester et al 1993 Bruhn et al1994 Caine et al 1996 Jourde et al 2002) Thisdefinition of a fault zone requires the evaluationof its overall behavior and particularly that of its

Figure 4 Interpreted seismic section AAprime (see the location in Figure 3) showing major faults and stratigraphic intervals in this studyQD = Qidong ZDH4 = Zhangdonghai 4 YEZ = Yangerzhuang YEZN = Yangerzhuangnan

Zhang et al 1385

potential hydrocarbon-carrying capacities duringthe entire history of hydrocarbon migration

Furthermore consider that the commercialquantity of hydrocarbons having migrated throughfault zonesmust be a result of multiple events (suchas earthquakes) that occurred over a long-term geo-logic period The present distribution of hydrocar-bons along a fault is the consequence of these mul-tiple episodes of migration and accumulation Atany time any point on the fault has the potentialto act as a barrier or conduit However the behav-ior of this point during a single displacement eventcannot be evaluated because hydrocarbon migra-tion processes corresponding to this single faultingevent cannot be assessed Nevertheless the cumu-lative effects of these events can be identified solong as some of these events leave a preserved trace

Identifying Fault Connectivity

The presence or absence of hydrocarbon-filled res-ervoirs on both sides of a fault gives a quantitativeindication onwhether the fault has been open dur-

1386 Geohorizons

ing at least some of the events occurring as a func-tion of geological time For a given location on thefault an open and close indicator may be con-structed as explained in the following paragraphwhere it is applied to the CSFZ

This study concentrates on faults that havebeen drilled extensively on both their footwallsand hanging walls so that sufficient data are avail-able to better characterize the history of fault open-ing and closing Actually the CSFZ is particularlyconvenient for the use of hydrocarbons as an indi-cator of fault opening and closing This is becausehydrocarbons accumulatedprincipally during a rela-tively short period from the Miocene to Pliocene(Yu et al 2006) Since then the study area haskept subsiding without significant changes in tec-tonic style (Wang et al 2003)

As illustrated on Figure 3 the six main faultsof the CSFZ clearly connected with oil and gas dis-coveries are selected The precise relationship be-tween the fault geometry and the occurrence of hy-drocarbons in reservoirs may be assessed using datafrom exploration wells and 3-D seismic surveysFigure 6 presents an enlargement of the eastern part

Figure 5 Conceptual model of ele-ments in a fault zone (modified fromChester and Logan 1986 Smith et al1990 Bruhn et al 1994 Caine et al 1996)

Figure 6 Detailed configuration of the eastern part of the Zhangdonghai 4 (ZDH4) fault (a) A cross section showing the deep part of the fault zone and the occurrence of hydro-carbons The location is on panels b c and d (b c d) Maps showing the horizontal extent of the oil-bearing layer in Es2

2 Es21 and Es1 (see Figure 2) and the contour maps at the top of

these intervals The contour interval is 50 m

Zhangetal

1387

of fault zone ZDH4 A cross section perpendicularto the main fault (Figure 6a) displays the majorhorizons interpreted from 3-D seismic data togetherwith major reservoirs The lateral extent of the oil-bearing layer (as well as the oil-water contacts) andthe topography at the top of the Es2

2 Es21 and Es1

reservoirs (Figure 6bndashd) emphasize that in mostoccasions faults acted as a seal with respect to thereservoirs However hydrocarbons that accumu-lated in the footwall cannot be from the local sourcerock of Es1 or Es3 because it is poor and immaturein this area Oils must have been fed through open-ings of normal fault ZDH4 (Figure 6a) which ex-tend from the base of Es2

2 in the footwall to thetop of Es2

1 in the hanging wallOn a cross section of the fault plane we iden-

tify particular points (called grid nodes) as the topof each bed in both the hanging wall and the foot-wall (Figure 7a) Assuming that the fault is theonly possible pathway the opening and closingstate of such a node on the fault during the migra-tion period is determined by identifying the hy-drocarbon presence or absence in reservoirs above

1388 Geohorizons

that node on both sides of the fault plane Fourpossible scenarios may occur and are used to iden-tify whether the node had served as a migrationpath The four scenarios can be described as fol-lows (Figure 7b) (1) the reservoirs below containhydrocarbons and the ones above do not (the caseof point D on Figure 7b) (2) reservoirs both aboveand below the node contain hydrocarbons (thecase of point C on Figure 7b) (3) the reservoirsabove the node contain hydrocarbons and thosebelow the node do not (the case of points A andB on Figure 7b) and (4) reservoirs both aboveand below the node do not contain hydrocarbons(the cases of points E and F on Figure 7b) It is notdifficult to assess that in scenarios 2 and 3 the nodemust have been open at some time during migra-tion on the contrary in scenario 1 no hydrocarbonaccumulation above this node exists at present andthus the node was probably closed during migra-tion In the last scenario it is unknown whetherthe node was open or closed during migration

At the scale of the CFSZ many such cross sec-tions which include both structural information

Figure 7 Determination of fault connectivity on the basis of hydrocarbon occurrence (a) Schematic block diagram showing a numer-ical model of a normal fault and positions of several stratigraphic intervals at the footwall and hanging wall on the fault surface (b) Sketchdemonstrating the criteria used for identifying whether a fault was open From the occurrence of hydrocarbon nodes A B and C wereclearly open node D remains closed and the state of nodes E and F were not certain during migration Ni is defined as the fault-connectivity index For a given node Ni = 1 indicates the fault has opened for migration Ni = 0 indicates the fault kept closing and Ni = means the effect of the fault cannot be identified and the node will not be taken into account in statistical analysis

and indications of hydrocarbon occurrence maybe constructed A systematic study derived fromavailable seismic sections is therefore performedThe geometrical and structural data are first digi-tized for grid mapping of the main faults An ex-ample of such mapping is shown on Figure 8which presents a perspective view of the footwallof fault ZDH4 The traces of the main horizons(both of the hanging wall and of the footwall) aswell as the traces of seismic cross sections are thebasic elements of this grid Based on the mappingof the oil and gas shows assessing the connectiv-ity of the fault at various nodes of a grid is com-monly possible

Fault-Connectivity Indicator as aProbabilistic Criterion

Sixteen cross-well seismic sections perpendicularto six selected major faults are used to analyze thefault connectivity during migration At each inter-section point between the fault plane and a carrierbed (Figure 7b) a binary index termed the ldquofault-connectivity indexrdquo (Ni) is defined as 0 when evi-dence for migration is absent and 1 when evidencefor migration is present If the evidence is inconclu-sive the corresponding data point is discarded

About 120 data points are compiled each being as-signed with an openclose index

The relevance of this index to the actual open-ing and closing of a fault is debatable First it ispossible that part of the migration occurs in otherpathways instead of the main fault For instancesmall faults or cracks generally accompanying mas-ter faults in extensional basins could channel a sub-stantial fraction of the migrated hydrocarbonsThere also is an implicit assumption (Figure 7) thatat each data point hydrocarbon migrates in a 2-Dplane perpendicular to the fault In fact it is likelythat pathways along the fault are muchmore com-plex than being depicted here and their exact 3-Dgeometry is difficult to assess A statistical approachmay be effective in countering this complexity Ata given point on the fault it is convenient to con-sider the fault-connectivity index as a binary vari-able with a value of 0 or 1 The probability that theindex has a value of 1 depends on the position of thepoint on the fault and is a function of other variablesof the fault such as physical parameters character-izing the environment and history of the faultThe previously collected data allow the character-ization of this probability relationship as follows

For a specified value range of some physicalparameters associated with a fault (eg depth

Figure 8 Structural block diagram of the footwall of fault ZDH4 The positions of stratigraphic intervals are marked as solid lines on thefootwall and as dashed lines on the hanging wall

Zhang et al 1389

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

teaching sedimentology and terrigenous clasticdepositional systems He has worked on depo-sitional systems analysis sequence and cyclo-stratigraphy paleoclimatology and reservoircharacterization of marine and nonmarine silici-clastic and carbonate rocks in China and theUnited States

Guy Vasseur SISYPHE UMR7619 Univer-sity Pierre et Marie Curie 75252 Paris CedexFrance

Guy Vasseur received an engineering degreefrom Ecole Polytechnique (Paris) in 1963 and aDoctorat drsquoEtat degree (PhD) from the Uni-versity of Paris in 1971 He then became a seniorscientist of Centre National de la RechercheScientifique and worked at the Universities ofMontpellier 2 and Paris 6 in various domainsof geophysics His main domain of competenceis the modeling of heat and mass transfer inthe Earthrsquos crust

Changhua Yu Petroleum Exploration andDevelopment Research Institute of DagangOilfield Company Ltd Tianjin 300280 China

Changhua Yu is an engineer at the PetroleumExploration and Development Research Instituteof Dagang Oilfield Company He received a BSdegree from Jianghan Petroleum Institute Chinain 1998 His works focus on the structural con-trols on fluid flow in petroleum systems

Junqing Su Petroleum Exploration andDevelopment Research Institute of DagangOilfield Company Ltd Tianjin 300280 China

Junqing Su got an MS degree in geology fromYangtze University China He did geological re-search for the Petroleum Exploration and De-velopment Research Institute of Dagang OilfieldCompany and has more than 20 years of ex-ploration experience His research interestsfocus on 3-D seismic interpretation and compre-hensive analysis of petroleum systems

Shuqin Yuan Key Laboratory of PetroleumResources Research Chinese Academy of Sci-ences IGG Beijing 100029 China present ad-dress Petroleum Exploration and DevelopmentResearch Institute of Dagang Oilfield CompanyLtd Tianjin 300280 China

Shuqin Yuan is a senior expert in the PetroleumExploration and Development Research Insti-tute of Dagang Oilfield Company She has anMS degree in geology from Yangtze University

1380 Geohorizons

to develop this method Fluid pressure in mudstones normalstress perpendicular to fault plane and shale gouge ratio areidentified as the key factors representing fault-seal capacityThey are combined to define a nondimensional fault openingindex (FOI) The values of FOI are calculated from the mea-sured values of the key factors and the relationship betweenFOI and fault-connectivity probability on any fault segmentis established through statistical analysis Based on the datafrom the CSFZ when the FOI is less than 075 the fault-connectivity probability is 0 when FOI ranges from 075 to325 the corresponding fault-connectivity probability in-creases from 0 to 1 following a quadratic polynomial relation-ship when FOI is greater than 325 the fault-connectivityprobability is 1 The values of fault-connectivity probabilitycan be contoured on a fault plane to characterize the variationsof hydraulic connective capacity on the fault plane The appli-cability of this concept for other oil fields (in particular thequantitative relationship between FOI and fault-connectivityprobability) has still to be ascertained

INTRODUCTION

Faults may act as either barriers to fluid flow or conduits (Smith1966 Karlsen and Skeie 2006) Fluid flow along faults causeschanges in temperature pressure and stresses within and sur-rounding fault zones (Sibson 1981Vasseur andDemongodin1995 Xie et al 2001Gratier et al 2002)Obviously the open-ing and closing behavior of a fault controls hydrocarbon mi-gration and accumulation and therefore the distribution of hy-drocarbon fields in petroliferous basins (Selley 1998 Hardingand Tuminas 1989)

Studies on faults serving as flow barriers or conduits (Knipe1989 Sorkhabi and Tsuji 2005) have been conducted throughoutcrop observations lab experiments and numerical simula-tions to characterize the geometry composition deformationand diagenesis of fault zones In fault zones several observedphysical and chemical properties have been proposed as proxydata to evaluate the timing and duration of fault opening andclosing (Smith 1966Weber et al 1978 Engelder 1979 Allen1989 Knipe 1992 1997 Antonelini and Aydin 1994 Gibson1994 Berg and Alana 1995 Sorkhabi and Tsuji 2005)

A variety of approaches have been used to characterizethe sealing ability of faults to study their impact on fluid flowPerkins (1961) emphasized that the juxtaposition of sand-stones may destroy the sealing capability of faults Howeverhe noticed that faulted sandstones may be sealed by mudstones

China Her research focuses on hydrocarbonsystems and the potential of hydrocarbon playsin petroliferous basins

Dunqing Xiao Petroleum Exploration andDevelopment Research Institute of DagangOilfield Company Ltd Tianjin 300280 China

Dunqing Xiao received a BA degree in geologyfrom the China University of Geoscience He isa principal geologist in the Petroleum Explora-tion and Development Research Institute ofDagang Oilfield Company with research inter-ests on geological conditions of oil and gas ac-cumulation and exploration of oil and gas plays

Zhaoming Wang Research Institute ofExploration and Development PetroChina Beijing100083 China

Zhaoming Wang received a BS degree in petro-leum geology from the China University of Pe-troleum Dongying and an MS degree and PhDfrom the Institute of Geology and GeophysicsChinese Academy of Sciences He has workedin the Shengli oil field Currently he is working inthe Institute of Exploration and DevelopmentPetroChina His interests are in the study of re-source assessment of sedimentary basins andhydrocarbon migration

ACKNOWLEDGEMENTS

This study was supported partly by the ChineseNational Major Fundamental Research Devel-oping Project (2006CB202305) and partly by theChinese National Natural Science Foundation(40772090) The study could not have beenpossible without support from Dagang Oil Com-pany Ltd We appreciate the permission fromDagang Oil Company to publish this work Wethank Beicip-Franlab for providing the Temis3Dsoftware used for the pressure modeling Weappreciate the comments and encouragementfrom A C Aplin and three reviewers StephenNaruk Barry J Katz and Alexander A Kitchkawhich have helped improve the manuscriptThe AAPG Editor thanks the following reviewersfor their work on this article Barry J KatzAlexander A Kitchka and Stephen J Naruk

squeezed into the sandstones during faultingAllen (1989) statedthat stratal discontinuity across a fault must be studied threedimensionally Smith (1966) introduced some theoretical con-siderations on sealing and nonsealing faults based on the cap-illary model of Hubbert (1953) To characterize the sealingheterogeneity along fault planes several parameters such asclay smear potential shale smear factor and shale gouge ratio(SGR) were proposed to estimate the degree of mudstonesmearing within fault zones and to evaluate the seal capacityof faults (Bouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) Using outcrop observations Hasegawa et al(2005) and Kachi et al (2005) designed several seal capacityparameters to describe the three-dimensional (3-D) sealingcharacteristics of faults and emphasize the heterogeneity offault-seal capacity These detailed studies have pushed theconcept of fault-seal capacity from qualitative toward quanti-tative (Sorkhabi and Tsuji 2005) Howevermost of these pre-vious studies are based on the present characteristics of a faultwhereas hydrocarbon accumulation involves the sealing char-acteristics of the faults during the entire pastmigration processIndeed this entire migration process may be very long be-cause of the limited hydrocarbon migration flux through faultzones (Byerlee 1993 Haney et al 2005) andor because ofthe very slow maturation of hydrocarbons in the source rocks

In reality active tectonics associated with faults are discon-tinuous processes (Sibson et al 1975 Hooper 1991) andfaults open and close episodically as a result of a variety of geo-logic processes (eg earthquake cycles Sibson 1981 Hooper1991) With respect to their hydrodynamic properties faultshave aubiquitous property theymay act either as seals or as pref-erential pathways for fluid flow (Yielding et al 1999) There-fore any comprehensive study of hydrocarbon migrationthrough faults should take this dual behavior into account as wellas its temporal evolution Indeed it is widely accepted that in afault zone fluid can flow in pulses from one compartment to an-other in relation with the earthquake rupture of existing seals(Byerlee 1993 Lockner and Byerlee 1995 Haney et al 2005)

Hydrocarbon flow occurring during fault opening shouldleave some indications that can be used to identify whetherandwhere the fault was open (Sorkhabi andTsuji 2005) Indi-cators such as veins related to faulting and diagenetic miner-alogy and fluid inclusions within and adjacent to fault zonesare equivocal in many cases and require extensive testing andanalyses many samples This is feasible for outcrop studies butis difficult for core analysis because of the scarcity of cores in faultzones Other types of proxy are needed to assess fault openingand closing and its seal capacity

Zhang et al 1381

The purpose of this study is to propose suchproxies which can be used to characterize the sealcapacity of faults in the Chengbei step-fault zone(CSFZ) Bohai Bay Basin China Our approach instudying fault-seal capacity is to consider the long-term hydrocarbon migration process as a cumula-tive effect of many small increments of migratinghydrocarbons through a segment of a fault zonesuch as the migration episodes that occur duringearthquakes The presence of hydrocarbon accumu-lations in footwall and hanging-wall reservoirs istherefore used as a calibration criterion indicatingcumulative fault opening and closing cycles for mi-gration over geological time A method of quanti-tative assessment of 3-D fault opening and closingat different locations on a fault plane is proposedand the opening and closing behavior on a faultplane during the entiremigration historymay semi-quantitatively be described using a concept of fault-connectivity probability

1382 Geohorizons

GEOLOGICAL SETTING OF THE CHENGBEISTEP-FAULT ZONE

Structure Stratigraphy and Petroleum System

The CSFZ is located on the slope between theChengning uplift and Qikou depression in theBohai Bay Basin northeast China (Figure 1) It cov-ers an area of approximately 550 km2 (212 mi2)with threefifths of the field offshore The struc-tural framework is a gentle slope cut by multiplestep-like normal faults in the Cenozoic strata (Wanget al 2003) The CSFZ as well as the entire BohaiBay Basin has experienced a rifting episode com-posed of two phases a Paleogene synrift phase anda Neogene to Quaternary postrift phase (Li et al1998) The tectonic evolution is related to strike-slip movements along the Tanlu fault and Qinlingstructural zone (Xu 1980 Yang andXu 2004) TheCenozoic strata are composed of continental clastic

Figure 1 Location map of the Chengbei step-fault zone as outlined by a square on the right panel on the southern slope of the Qikoudepression in Bohai Bay Basin eastern China

deposits with a total thickness of 2000ndash5000 m(6562ndash16404 ft) (Figure 2) The Paleogene synriftdeposits include the Shahejie and Dongying forma-tions The Neogene to Quaternary sequence con-sists of the Guantao Minghuazhen and Pingyuanformations A regional unconformity separatesPaleogene and Neogene strata (Yuan et al 2004)

Oilndashsource rock correlation shows that hydro-carbons in the CSFZ were derived from sourcerocks composed of dark-colored mudstones of theSha-1 and Sha-3 members of the Shahejie Forma-tion in the Qikou depression (Wang et al 2006)(Figure 2) The major reservoir rocks include fandeltaic and lacustrine littoral sandstones of theSha-2 and Sha-3 members of the Shahejie Forma-

tion delta-front and prodeltaic sandstones of theDong-1 andDong-3members of the Dongying For-mation and Neogene fluvial sandstones (Yuanet al 2004) (Figure 2) Thick mudstones in themiddle Sha-1 and Dong-2 members and lowerMinghuazhen Formation constitute regional sealrocks (Yu et al 2006) (Figure 2)

Petroleum exploration in the CSFZ began in1965 During the last 43 yr 11 oil and gas fieldshave beendiscovered (Figure 3)More than 140 ex-ploratory wells have been drilled and a 3-D seis-mic survey covers nearly all major structures Alarge quantity of exploration and developmentdata are available to study fault-seal capacity inthe CSFZ (Figure 3)

Figure 2 Summary diagram showing chrono- and lithostratigraphy tectonic stages and characteristics of the petroleum system in theChengbei step-fault zone (after Yuan et al 2004)

Zhang et al 1383

Movement of Faults andHydrocarbon Expulsion

The study area has experienced regional transten-sion since the end of theMesozoic (Li et al 1998)resulting in extensive faulting Major faults whichcontrolled the structural framework and deposi-tional history include the Qidong (QD) Zhang-donghai 4 (ZDH4) Zaobei (ZB) Yangerzhuang(YEZ) Yangerzhuangnan (YEZN) and Zhangbei(ZHB) faults (Yuan et al 2004 Yu et al 2006)(Figure 3) Seismic interpretation indicates that theyare synsedimentary faults (Figure 4) In general

1384 Geohorizons

these faults are parallel to each other striking in anortheast or nearly eastndashwest direction and dippingtoward the depression to the north The distribu-tion of discovered hydrocarbon accumulations isclosely related to the faults (Figure 3) These faultsappear to be important factors in hydrocarbon mi-gration and accumulation

In this study the incremental movements alonga given fault are assumed to be the result of multi-ple tectonic episodes in the CSFZ The fault growthindex defined as the ratio between the thicknessesof the same formation measured on both sides ofthe fault plane (Childs et al 2003) was used to

Figure 3 Faults oil fields key wells used in this study (black dots) other wells (black circles) and 3-D seismic data sets (dashed linerectangles) in the Chengbei step-fault zone Faults are mapped at the top of the Shahejie Formation Section AAprime is presented in Figure 4QD = Qidong YEZ = Yangerzhuang YEZN = Yangerzhuangnan ZB = Zaobei ZDH4 = Zhangdonghai 4 ZHB = Zhangbei

analyze the development and activity of growthfaults in the study area (Zhang et al 2007) Theyrevealed the occurrence of three major episodesof intensive faulting during the deposition of theShahejie Donying and upper Minghuazhen for-mations respectively The tectonic intensity hadbeen decreasing from the Paleogene to theNeogene

Basin modeling applied to the whole region(Figure 3) indicates that hydrocarbonwas expelledfrom source rocks during twomajor episodes (Wanget al 2006 Yu et al 2006) The first episode wasat the end of the Dongying (sim26Ma) when sourcerocks in the Sha-3member becamematureHydro-carbon expulsion ceased because of regional uplift-ing and erosion at the end of the Paleogene There-fore this early episode of hydrocarbon expulsionwas short and the amount of expelled hydrocar-bon was limited Major expulsion occurred duringthe second episode from the late Miocene to Qua-ternary (Wang et al 2006) The deposition of theNeogene Guantao and Minghuazhen formationscaused maturation and hydrocarbon generationin the Sha-3 and Sha-1 source rocks Hydrocarbonexpulsion peaked at the end of the Minghuazhen(sim2 Ma)

FAULT CONNECTIVITY ANDPROBABILISTIC INTERPRETATION

We characterized the fault opening or closing on astatistical basis and developed the concept offault-connectivity probability

Fault Opening and Closing

A fault is not a simple discontinuity but instead azone consisting of a fault core and a surroundingdam-age zone (Chester and Logan 1986 Smith et al1990 Forster and Evans 1991 Chester et al 1993Bruhn et al 1994 Caine et al 1996) (Figure 5)The fault core represents the part of a fault zonewhere most of the displacement is accommodatedand consists of slip surfaces fault gouge brecciasand cataclasites (Chester and Logan 1986 Chesteret al 1993 Caine et al 1996 Clausen et al2003) The damage zone may be composed of sub-sidiary faults faulted rocks and fractures (Chesterand Logan 1986 Chester et al 1993 Bruhn et al1994 Caine et al 1996 Jourde et al 2002) Thisdefinition of a fault zone requires the evaluationof its overall behavior and particularly that of its

Figure 4 Interpreted seismic section AAprime (see the location in Figure 3) showing major faults and stratigraphic intervals in this studyQD = Qidong ZDH4 = Zhangdonghai 4 YEZ = Yangerzhuang YEZN = Yangerzhuangnan

Zhang et al 1385

potential hydrocarbon-carrying capacities duringthe entire history of hydrocarbon migration

Furthermore consider that the commercialquantity of hydrocarbons having migrated throughfault zonesmust be a result of multiple events (suchas earthquakes) that occurred over a long-term geo-logic period The present distribution of hydrocar-bons along a fault is the consequence of these mul-tiple episodes of migration and accumulation Atany time any point on the fault has the potentialto act as a barrier or conduit However the behav-ior of this point during a single displacement eventcannot be evaluated because hydrocarbon migra-tion processes corresponding to this single faultingevent cannot be assessed Nevertheless the cumu-lative effects of these events can be identified solong as some of these events leave a preserved trace

Identifying Fault Connectivity

The presence or absence of hydrocarbon-filled res-ervoirs on both sides of a fault gives a quantitativeindication onwhether the fault has been open dur-

1386 Geohorizons

ing at least some of the events occurring as a func-tion of geological time For a given location on thefault an open and close indicator may be con-structed as explained in the following paragraphwhere it is applied to the CSFZ

This study concentrates on faults that havebeen drilled extensively on both their footwallsand hanging walls so that sufficient data are avail-able to better characterize the history of fault open-ing and closing Actually the CSFZ is particularlyconvenient for the use of hydrocarbons as an indi-cator of fault opening and closing This is becausehydrocarbons accumulatedprincipally during a rela-tively short period from the Miocene to Pliocene(Yu et al 2006) Since then the study area haskept subsiding without significant changes in tec-tonic style (Wang et al 2003)

As illustrated on Figure 3 the six main faultsof the CSFZ clearly connected with oil and gas dis-coveries are selected The precise relationship be-tween the fault geometry and the occurrence of hy-drocarbons in reservoirs may be assessed using datafrom exploration wells and 3-D seismic surveysFigure 6 presents an enlargement of the eastern part

Figure 5 Conceptual model of ele-ments in a fault zone (modified fromChester and Logan 1986 Smith et al1990 Bruhn et al 1994 Caine et al 1996)

Figure 6 Detailed configuration of the eastern part of the Zhangdonghai 4 (ZDH4) fault (a) A cross section showing the deep part of the fault zone and the occurrence of hydro-carbons The location is on panels b c and d (b c d) Maps showing the horizontal extent of the oil-bearing layer in Es2

2 Es21 and Es1 (see Figure 2) and the contour maps at the top of

these intervals The contour interval is 50 m

Zhangetal

1387

of fault zone ZDH4 A cross section perpendicularto the main fault (Figure 6a) displays the majorhorizons interpreted from 3-D seismic data togetherwith major reservoirs The lateral extent of the oil-bearing layer (as well as the oil-water contacts) andthe topography at the top of the Es2

2 Es21 and Es1

reservoirs (Figure 6bndashd) emphasize that in mostoccasions faults acted as a seal with respect to thereservoirs However hydrocarbons that accumu-lated in the footwall cannot be from the local sourcerock of Es1 or Es3 because it is poor and immaturein this area Oils must have been fed through open-ings of normal fault ZDH4 (Figure 6a) which ex-tend from the base of Es2

2 in the footwall to thetop of Es2

1 in the hanging wallOn a cross section of the fault plane we iden-

tify particular points (called grid nodes) as the topof each bed in both the hanging wall and the foot-wall (Figure 7a) Assuming that the fault is theonly possible pathway the opening and closingstate of such a node on the fault during the migra-tion period is determined by identifying the hy-drocarbon presence or absence in reservoirs above

1388 Geohorizons

that node on both sides of the fault plane Fourpossible scenarios may occur and are used to iden-tify whether the node had served as a migrationpath The four scenarios can be described as fol-lows (Figure 7b) (1) the reservoirs below containhydrocarbons and the ones above do not (the caseof point D on Figure 7b) (2) reservoirs both aboveand below the node contain hydrocarbons (thecase of point C on Figure 7b) (3) the reservoirsabove the node contain hydrocarbons and thosebelow the node do not (the case of points A andB on Figure 7b) and (4) reservoirs both aboveand below the node do not contain hydrocarbons(the cases of points E and F on Figure 7b) It is notdifficult to assess that in scenarios 2 and 3 the nodemust have been open at some time during migra-tion on the contrary in scenario 1 no hydrocarbonaccumulation above this node exists at present andthus the node was probably closed during migra-tion In the last scenario it is unknown whetherthe node was open or closed during migration

At the scale of the CFSZ many such cross sec-tions which include both structural information

Figure 7 Determination of fault connectivity on the basis of hydrocarbon occurrence (a) Schematic block diagram showing a numer-ical model of a normal fault and positions of several stratigraphic intervals at the footwall and hanging wall on the fault surface (b) Sketchdemonstrating the criteria used for identifying whether a fault was open From the occurrence of hydrocarbon nodes A B and C wereclearly open node D remains closed and the state of nodes E and F were not certain during migration Ni is defined as the fault-connectivity index For a given node Ni = 1 indicates the fault has opened for migration Ni = 0 indicates the fault kept closing and Ni = means the effect of the fault cannot be identified and the node will not be taken into account in statistical analysis

and indications of hydrocarbon occurrence maybe constructed A systematic study derived fromavailable seismic sections is therefore performedThe geometrical and structural data are first digi-tized for grid mapping of the main faults An ex-ample of such mapping is shown on Figure 8which presents a perspective view of the footwallof fault ZDH4 The traces of the main horizons(both of the hanging wall and of the footwall) aswell as the traces of seismic cross sections are thebasic elements of this grid Based on the mappingof the oil and gas shows assessing the connectiv-ity of the fault at various nodes of a grid is com-monly possible

Fault-Connectivity Indicator as aProbabilistic Criterion

Sixteen cross-well seismic sections perpendicularto six selected major faults are used to analyze thefault connectivity during migration At each inter-section point between the fault plane and a carrierbed (Figure 7b) a binary index termed the ldquofault-connectivity indexrdquo (Ni) is defined as 0 when evi-dence for migration is absent and 1 when evidencefor migration is present If the evidence is inconclu-sive the corresponding data point is discarded

About 120 data points are compiled each being as-signed with an openclose index

The relevance of this index to the actual open-ing and closing of a fault is debatable First it ispossible that part of the migration occurs in otherpathways instead of the main fault For instancesmall faults or cracks generally accompanying mas-ter faults in extensional basins could channel a sub-stantial fraction of the migrated hydrocarbonsThere also is an implicit assumption (Figure 7) thatat each data point hydrocarbon migrates in a 2-Dplane perpendicular to the fault In fact it is likelythat pathways along the fault are muchmore com-plex than being depicted here and their exact 3-Dgeometry is difficult to assess A statistical approachmay be effective in countering this complexity Ata given point on the fault it is convenient to con-sider the fault-connectivity index as a binary vari-able with a value of 0 or 1 The probability that theindex has a value of 1 depends on the position of thepoint on the fault and is a function of other variablesof the fault such as physical parameters character-izing the environment and history of the faultThe previously collected data allow the character-ization of this probability relationship as follows

For a specified value range of some physicalparameters associated with a fault (eg depth

Figure 8 Structural block diagram of the footwall of fault ZDH4 The positions of stratigraphic intervals are marked as solid lines on thefootwall and as dashed lines on the hanging wall

Zhang et al 1389

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

China Her research focuses on hydrocarbonsystems and the potential of hydrocarbon playsin petroliferous basins

Dunqing Xiao Petroleum Exploration andDevelopment Research Institute of DagangOilfield Company Ltd Tianjin 300280 China

Dunqing Xiao received a BA degree in geologyfrom the China University of Geoscience He isa principal geologist in the Petroleum Explora-tion and Development Research Institute ofDagang Oilfield Company with research inter-ests on geological conditions of oil and gas ac-cumulation and exploration of oil and gas plays

Zhaoming Wang Research Institute ofExploration and Development PetroChina Beijing100083 China

Zhaoming Wang received a BS degree in petro-leum geology from the China University of Pe-troleum Dongying and an MS degree and PhDfrom the Institute of Geology and GeophysicsChinese Academy of Sciences He has workedin the Shengli oil field Currently he is working inthe Institute of Exploration and DevelopmentPetroChina His interests are in the study of re-source assessment of sedimentary basins andhydrocarbon migration

ACKNOWLEDGEMENTS

This study was supported partly by the ChineseNational Major Fundamental Research Devel-oping Project (2006CB202305) and partly by theChinese National Natural Science Foundation(40772090) The study could not have beenpossible without support from Dagang Oil Com-pany Ltd We appreciate the permission fromDagang Oil Company to publish this work Wethank Beicip-Franlab for providing the Temis3Dsoftware used for the pressure modeling Weappreciate the comments and encouragementfrom A C Aplin and three reviewers StephenNaruk Barry J Katz and Alexander A Kitchkawhich have helped improve the manuscriptThe AAPG Editor thanks the following reviewersfor their work on this article Barry J KatzAlexander A Kitchka and Stephen J Naruk

squeezed into the sandstones during faultingAllen (1989) statedthat stratal discontinuity across a fault must be studied threedimensionally Smith (1966) introduced some theoretical con-siderations on sealing and nonsealing faults based on the cap-illary model of Hubbert (1953) To characterize the sealingheterogeneity along fault planes several parameters such asclay smear potential shale smear factor and shale gouge ratio(SGR) were proposed to estimate the degree of mudstonesmearing within fault zones and to evaluate the seal capacityof faults (Bouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) Using outcrop observations Hasegawa et al(2005) and Kachi et al (2005) designed several seal capacityparameters to describe the three-dimensional (3-D) sealingcharacteristics of faults and emphasize the heterogeneity offault-seal capacity These detailed studies have pushed theconcept of fault-seal capacity from qualitative toward quanti-tative (Sorkhabi and Tsuji 2005) Howevermost of these pre-vious studies are based on the present characteristics of a faultwhereas hydrocarbon accumulation involves the sealing char-acteristics of the faults during the entire pastmigration processIndeed this entire migration process may be very long be-cause of the limited hydrocarbon migration flux through faultzones (Byerlee 1993 Haney et al 2005) andor because ofthe very slow maturation of hydrocarbons in the source rocks

In reality active tectonics associated with faults are discon-tinuous processes (Sibson et al 1975 Hooper 1991) andfaults open and close episodically as a result of a variety of geo-logic processes (eg earthquake cycles Sibson 1981 Hooper1991) With respect to their hydrodynamic properties faultshave aubiquitous property theymay act either as seals or as pref-erential pathways for fluid flow (Yielding et al 1999) There-fore any comprehensive study of hydrocarbon migrationthrough faults should take this dual behavior into account as wellas its temporal evolution Indeed it is widely accepted that in afault zone fluid can flow in pulses from one compartment to an-other in relation with the earthquake rupture of existing seals(Byerlee 1993 Lockner and Byerlee 1995 Haney et al 2005)

Hydrocarbon flow occurring during fault opening shouldleave some indications that can be used to identify whetherandwhere the fault was open (Sorkhabi andTsuji 2005) Indi-cators such as veins related to faulting and diagenetic miner-alogy and fluid inclusions within and adjacent to fault zonesare equivocal in many cases and require extensive testing andanalyses many samples This is feasible for outcrop studies butis difficult for core analysis because of the scarcity of cores in faultzones Other types of proxy are needed to assess fault openingand closing and its seal capacity

Zhang et al 1381

The purpose of this study is to propose suchproxies which can be used to characterize the sealcapacity of faults in the Chengbei step-fault zone(CSFZ) Bohai Bay Basin China Our approach instudying fault-seal capacity is to consider the long-term hydrocarbon migration process as a cumula-tive effect of many small increments of migratinghydrocarbons through a segment of a fault zonesuch as the migration episodes that occur duringearthquakes The presence of hydrocarbon accumu-lations in footwall and hanging-wall reservoirs istherefore used as a calibration criterion indicatingcumulative fault opening and closing cycles for mi-gration over geological time A method of quanti-tative assessment of 3-D fault opening and closingat different locations on a fault plane is proposedand the opening and closing behavior on a faultplane during the entiremigration historymay semi-quantitatively be described using a concept of fault-connectivity probability

1382 Geohorizons

GEOLOGICAL SETTING OF THE CHENGBEISTEP-FAULT ZONE

Structure Stratigraphy and Petroleum System

The CSFZ is located on the slope between theChengning uplift and Qikou depression in theBohai Bay Basin northeast China (Figure 1) It cov-ers an area of approximately 550 km2 (212 mi2)with threefifths of the field offshore The struc-tural framework is a gentle slope cut by multiplestep-like normal faults in the Cenozoic strata (Wanget al 2003) The CSFZ as well as the entire BohaiBay Basin has experienced a rifting episode com-posed of two phases a Paleogene synrift phase anda Neogene to Quaternary postrift phase (Li et al1998) The tectonic evolution is related to strike-slip movements along the Tanlu fault and Qinlingstructural zone (Xu 1980 Yang andXu 2004) TheCenozoic strata are composed of continental clastic

Figure 1 Location map of the Chengbei step-fault zone as outlined by a square on the right panel on the southern slope of the Qikoudepression in Bohai Bay Basin eastern China

deposits with a total thickness of 2000ndash5000 m(6562ndash16404 ft) (Figure 2) The Paleogene synriftdeposits include the Shahejie and Dongying forma-tions The Neogene to Quaternary sequence con-sists of the Guantao Minghuazhen and Pingyuanformations A regional unconformity separatesPaleogene and Neogene strata (Yuan et al 2004)

Oilndashsource rock correlation shows that hydro-carbons in the CSFZ were derived from sourcerocks composed of dark-colored mudstones of theSha-1 and Sha-3 members of the Shahejie Forma-tion in the Qikou depression (Wang et al 2006)(Figure 2) The major reservoir rocks include fandeltaic and lacustrine littoral sandstones of theSha-2 and Sha-3 members of the Shahejie Forma-

tion delta-front and prodeltaic sandstones of theDong-1 andDong-3members of the Dongying For-mation and Neogene fluvial sandstones (Yuanet al 2004) (Figure 2) Thick mudstones in themiddle Sha-1 and Dong-2 members and lowerMinghuazhen Formation constitute regional sealrocks (Yu et al 2006) (Figure 2)

Petroleum exploration in the CSFZ began in1965 During the last 43 yr 11 oil and gas fieldshave beendiscovered (Figure 3)More than 140 ex-ploratory wells have been drilled and a 3-D seis-mic survey covers nearly all major structures Alarge quantity of exploration and developmentdata are available to study fault-seal capacity inthe CSFZ (Figure 3)

Figure 2 Summary diagram showing chrono- and lithostratigraphy tectonic stages and characteristics of the petroleum system in theChengbei step-fault zone (after Yuan et al 2004)

Zhang et al 1383

Movement of Faults andHydrocarbon Expulsion

The study area has experienced regional transten-sion since the end of theMesozoic (Li et al 1998)resulting in extensive faulting Major faults whichcontrolled the structural framework and deposi-tional history include the Qidong (QD) Zhang-donghai 4 (ZDH4) Zaobei (ZB) Yangerzhuang(YEZ) Yangerzhuangnan (YEZN) and Zhangbei(ZHB) faults (Yuan et al 2004 Yu et al 2006)(Figure 3) Seismic interpretation indicates that theyare synsedimentary faults (Figure 4) In general

1384 Geohorizons

these faults are parallel to each other striking in anortheast or nearly eastndashwest direction and dippingtoward the depression to the north The distribu-tion of discovered hydrocarbon accumulations isclosely related to the faults (Figure 3) These faultsappear to be important factors in hydrocarbon mi-gration and accumulation

In this study the incremental movements alonga given fault are assumed to be the result of multi-ple tectonic episodes in the CSFZ The fault growthindex defined as the ratio between the thicknessesof the same formation measured on both sides ofthe fault plane (Childs et al 2003) was used to

Figure 3 Faults oil fields key wells used in this study (black dots) other wells (black circles) and 3-D seismic data sets (dashed linerectangles) in the Chengbei step-fault zone Faults are mapped at the top of the Shahejie Formation Section AAprime is presented in Figure 4QD = Qidong YEZ = Yangerzhuang YEZN = Yangerzhuangnan ZB = Zaobei ZDH4 = Zhangdonghai 4 ZHB = Zhangbei

analyze the development and activity of growthfaults in the study area (Zhang et al 2007) Theyrevealed the occurrence of three major episodesof intensive faulting during the deposition of theShahejie Donying and upper Minghuazhen for-mations respectively The tectonic intensity hadbeen decreasing from the Paleogene to theNeogene

Basin modeling applied to the whole region(Figure 3) indicates that hydrocarbonwas expelledfrom source rocks during twomajor episodes (Wanget al 2006 Yu et al 2006) The first episode wasat the end of the Dongying (sim26Ma) when sourcerocks in the Sha-3member becamematureHydro-carbon expulsion ceased because of regional uplift-ing and erosion at the end of the Paleogene There-fore this early episode of hydrocarbon expulsionwas short and the amount of expelled hydrocar-bon was limited Major expulsion occurred duringthe second episode from the late Miocene to Qua-ternary (Wang et al 2006) The deposition of theNeogene Guantao and Minghuazhen formationscaused maturation and hydrocarbon generationin the Sha-3 and Sha-1 source rocks Hydrocarbonexpulsion peaked at the end of the Minghuazhen(sim2 Ma)

FAULT CONNECTIVITY ANDPROBABILISTIC INTERPRETATION

We characterized the fault opening or closing on astatistical basis and developed the concept offault-connectivity probability

Fault Opening and Closing

A fault is not a simple discontinuity but instead azone consisting of a fault core and a surroundingdam-age zone (Chester and Logan 1986 Smith et al1990 Forster and Evans 1991 Chester et al 1993Bruhn et al 1994 Caine et al 1996) (Figure 5)The fault core represents the part of a fault zonewhere most of the displacement is accommodatedand consists of slip surfaces fault gouge brecciasand cataclasites (Chester and Logan 1986 Chesteret al 1993 Caine et al 1996 Clausen et al2003) The damage zone may be composed of sub-sidiary faults faulted rocks and fractures (Chesterand Logan 1986 Chester et al 1993 Bruhn et al1994 Caine et al 1996 Jourde et al 2002) Thisdefinition of a fault zone requires the evaluationof its overall behavior and particularly that of its

Figure 4 Interpreted seismic section AAprime (see the location in Figure 3) showing major faults and stratigraphic intervals in this studyQD = Qidong ZDH4 = Zhangdonghai 4 YEZ = Yangerzhuang YEZN = Yangerzhuangnan

Zhang et al 1385

potential hydrocarbon-carrying capacities duringthe entire history of hydrocarbon migration

Furthermore consider that the commercialquantity of hydrocarbons having migrated throughfault zonesmust be a result of multiple events (suchas earthquakes) that occurred over a long-term geo-logic period The present distribution of hydrocar-bons along a fault is the consequence of these mul-tiple episodes of migration and accumulation Atany time any point on the fault has the potentialto act as a barrier or conduit However the behav-ior of this point during a single displacement eventcannot be evaluated because hydrocarbon migra-tion processes corresponding to this single faultingevent cannot be assessed Nevertheless the cumu-lative effects of these events can be identified solong as some of these events leave a preserved trace

Identifying Fault Connectivity

The presence or absence of hydrocarbon-filled res-ervoirs on both sides of a fault gives a quantitativeindication onwhether the fault has been open dur-

1386 Geohorizons

ing at least some of the events occurring as a func-tion of geological time For a given location on thefault an open and close indicator may be con-structed as explained in the following paragraphwhere it is applied to the CSFZ

This study concentrates on faults that havebeen drilled extensively on both their footwallsand hanging walls so that sufficient data are avail-able to better characterize the history of fault open-ing and closing Actually the CSFZ is particularlyconvenient for the use of hydrocarbons as an indi-cator of fault opening and closing This is becausehydrocarbons accumulatedprincipally during a rela-tively short period from the Miocene to Pliocene(Yu et al 2006) Since then the study area haskept subsiding without significant changes in tec-tonic style (Wang et al 2003)

As illustrated on Figure 3 the six main faultsof the CSFZ clearly connected with oil and gas dis-coveries are selected The precise relationship be-tween the fault geometry and the occurrence of hy-drocarbons in reservoirs may be assessed using datafrom exploration wells and 3-D seismic surveysFigure 6 presents an enlargement of the eastern part

Figure 5 Conceptual model of ele-ments in a fault zone (modified fromChester and Logan 1986 Smith et al1990 Bruhn et al 1994 Caine et al 1996)

Figure 6 Detailed configuration of the eastern part of the Zhangdonghai 4 (ZDH4) fault (a) A cross section showing the deep part of the fault zone and the occurrence of hydro-carbons The location is on panels b c and d (b c d) Maps showing the horizontal extent of the oil-bearing layer in Es2

2 Es21 and Es1 (see Figure 2) and the contour maps at the top of

these intervals The contour interval is 50 m

Zhangetal

1387

of fault zone ZDH4 A cross section perpendicularto the main fault (Figure 6a) displays the majorhorizons interpreted from 3-D seismic data togetherwith major reservoirs The lateral extent of the oil-bearing layer (as well as the oil-water contacts) andthe topography at the top of the Es2

2 Es21 and Es1

reservoirs (Figure 6bndashd) emphasize that in mostoccasions faults acted as a seal with respect to thereservoirs However hydrocarbons that accumu-lated in the footwall cannot be from the local sourcerock of Es1 or Es3 because it is poor and immaturein this area Oils must have been fed through open-ings of normal fault ZDH4 (Figure 6a) which ex-tend from the base of Es2

2 in the footwall to thetop of Es2

1 in the hanging wallOn a cross section of the fault plane we iden-

tify particular points (called grid nodes) as the topof each bed in both the hanging wall and the foot-wall (Figure 7a) Assuming that the fault is theonly possible pathway the opening and closingstate of such a node on the fault during the migra-tion period is determined by identifying the hy-drocarbon presence or absence in reservoirs above

1388 Geohorizons

that node on both sides of the fault plane Fourpossible scenarios may occur and are used to iden-tify whether the node had served as a migrationpath The four scenarios can be described as fol-lows (Figure 7b) (1) the reservoirs below containhydrocarbons and the ones above do not (the caseof point D on Figure 7b) (2) reservoirs both aboveand below the node contain hydrocarbons (thecase of point C on Figure 7b) (3) the reservoirsabove the node contain hydrocarbons and thosebelow the node do not (the case of points A andB on Figure 7b) and (4) reservoirs both aboveand below the node do not contain hydrocarbons(the cases of points E and F on Figure 7b) It is notdifficult to assess that in scenarios 2 and 3 the nodemust have been open at some time during migra-tion on the contrary in scenario 1 no hydrocarbonaccumulation above this node exists at present andthus the node was probably closed during migra-tion In the last scenario it is unknown whetherthe node was open or closed during migration

At the scale of the CFSZ many such cross sec-tions which include both structural information

Figure 7 Determination of fault connectivity on the basis of hydrocarbon occurrence (a) Schematic block diagram showing a numer-ical model of a normal fault and positions of several stratigraphic intervals at the footwall and hanging wall on the fault surface (b) Sketchdemonstrating the criteria used for identifying whether a fault was open From the occurrence of hydrocarbon nodes A B and C wereclearly open node D remains closed and the state of nodes E and F were not certain during migration Ni is defined as the fault-connectivity index For a given node Ni = 1 indicates the fault has opened for migration Ni = 0 indicates the fault kept closing and Ni = means the effect of the fault cannot be identified and the node will not be taken into account in statistical analysis

and indications of hydrocarbon occurrence maybe constructed A systematic study derived fromavailable seismic sections is therefore performedThe geometrical and structural data are first digi-tized for grid mapping of the main faults An ex-ample of such mapping is shown on Figure 8which presents a perspective view of the footwallof fault ZDH4 The traces of the main horizons(both of the hanging wall and of the footwall) aswell as the traces of seismic cross sections are thebasic elements of this grid Based on the mappingof the oil and gas shows assessing the connectiv-ity of the fault at various nodes of a grid is com-monly possible

Fault-Connectivity Indicator as aProbabilistic Criterion

Sixteen cross-well seismic sections perpendicularto six selected major faults are used to analyze thefault connectivity during migration At each inter-section point between the fault plane and a carrierbed (Figure 7b) a binary index termed the ldquofault-connectivity indexrdquo (Ni) is defined as 0 when evi-dence for migration is absent and 1 when evidencefor migration is present If the evidence is inconclu-sive the corresponding data point is discarded

About 120 data points are compiled each being as-signed with an openclose index

The relevance of this index to the actual open-ing and closing of a fault is debatable First it ispossible that part of the migration occurs in otherpathways instead of the main fault For instancesmall faults or cracks generally accompanying mas-ter faults in extensional basins could channel a sub-stantial fraction of the migrated hydrocarbonsThere also is an implicit assumption (Figure 7) thatat each data point hydrocarbon migrates in a 2-Dplane perpendicular to the fault In fact it is likelythat pathways along the fault are muchmore com-plex than being depicted here and their exact 3-Dgeometry is difficult to assess A statistical approachmay be effective in countering this complexity Ata given point on the fault it is convenient to con-sider the fault-connectivity index as a binary vari-able with a value of 0 or 1 The probability that theindex has a value of 1 depends on the position of thepoint on the fault and is a function of other variablesof the fault such as physical parameters character-izing the environment and history of the faultThe previously collected data allow the character-ization of this probability relationship as follows

For a specified value range of some physicalparameters associated with a fault (eg depth

Figure 8 Structural block diagram of the footwall of fault ZDH4 The positions of stratigraphic intervals are marked as solid lines on thefootwall and as dashed lines on the hanging wall

Zhang et al 1389

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

The purpose of this study is to propose suchproxies which can be used to characterize the sealcapacity of faults in the Chengbei step-fault zone(CSFZ) Bohai Bay Basin China Our approach instudying fault-seal capacity is to consider the long-term hydrocarbon migration process as a cumula-tive effect of many small increments of migratinghydrocarbons through a segment of a fault zonesuch as the migration episodes that occur duringearthquakes The presence of hydrocarbon accumu-lations in footwall and hanging-wall reservoirs istherefore used as a calibration criterion indicatingcumulative fault opening and closing cycles for mi-gration over geological time A method of quanti-tative assessment of 3-D fault opening and closingat different locations on a fault plane is proposedand the opening and closing behavior on a faultplane during the entiremigration historymay semi-quantitatively be described using a concept of fault-connectivity probability

1382 Geohorizons

GEOLOGICAL SETTING OF THE CHENGBEISTEP-FAULT ZONE

Structure Stratigraphy and Petroleum System

The CSFZ is located on the slope between theChengning uplift and Qikou depression in theBohai Bay Basin northeast China (Figure 1) It cov-ers an area of approximately 550 km2 (212 mi2)with threefifths of the field offshore The struc-tural framework is a gentle slope cut by multiplestep-like normal faults in the Cenozoic strata (Wanget al 2003) The CSFZ as well as the entire BohaiBay Basin has experienced a rifting episode com-posed of two phases a Paleogene synrift phase anda Neogene to Quaternary postrift phase (Li et al1998) The tectonic evolution is related to strike-slip movements along the Tanlu fault and Qinlingstructural zone (Xu 1980 Yang andXu 2004) TheCenozoic strata are composed of continental clastic

Figure 1 Location map of the Chengbei step-fault zone as outlined by a square on the right panel on the southern slope of the Qikoudepression in Bohai Bay Basin eastern China

deposits with a total thickness of 2000ndash5000 m(6562ndash16404 ft) (Figure 2) The Paleogene synriftdeposits include the Shahejie and Dongying forma-tions The Neogene to Quaternary sequence con-sists of the Guantao Minghuazhen and Pingyuanformations A regional unconformity separatesPaleogene and Neogene strata (Yuan et al 2004)

Oilndashsource rock correlation shows that hydro-carbons in the CSFZ were derived from sourcerocks composed of dark-colored mudstones of theSha-1 and Sha-3 members of the Shahejie Forma-tion in the Qikou depression (Wang et al 2006)(Figure 2) The major reservoir rocks include fandeltaic and lacustrine littoral sandstones of theSha-2 and Sha-3 members of the Shahejie Forma-

tion delta-front and prodeltaic sandstones of theDong-1 andDong-3members of the Dongying For-mation and Neogene fluvial sandstones (Yuanet al 2004) (Figure 2) Thick mudstones in themiddle Sha-1 and Dong-2 members and lowerMinghuazhen Formation constitute regional sealrocks (Yu et al 2006) (Figure 2)

Petroleum exploration in the CSFZ began in1965 During the last 43 yr 11 oil and gas fieldshave beendiscovered (Figure 3)More than 140 ex-ploratory wells have been drilled and a 3-D seis-mic survey covers nearly all major structures Alarge quantity of exploration and developmentdata are available to study fault-seal capacity inthe CSFZ (Figure 3)

Figure 2 Summary diagram showing chrono- and lithostratigraphy tectonic stages and characteristics of the petroleum system in theChengbei step-fault zone (after Yuan et al 2004)

Zhang et al 1383

Movement of Faults andHydrocarbon Expulsion

The study area has experienced regional transten-sion since the end of theMesozoic (Li et al 1998)resulting in extensive faulting Major faults whichcontrolled the structural framework and deposi-tional history include the Qidong (QD) Zhang-donghai 4 (ZDH4) Zaobei (ZB) Yangerzhuang(YEZ) Yangerzhuangnan (YEZN) and Zhangbei(ZHB) faults (Yuan et al 2004 Yu et al 2006)(Figure 3) Seismic interpretation indicates that theyare synsedimentary faults (Figure 4) In general

1384 Geohorizons

these faults are parallel to each other striking in anortheast or nearly eastndashwest direction and dippingtoward the depression to the north The distribu-tion of discovered hydrocarbon accumulations isclosely related to the faults (Figure 3) These faultsappear to be important factors in hydrocarbon mi-gration and accumulation

In this study the incremental movements alonga given fault are assumed to be the result of multi-ple tectonic episodes in the CSFZ The fault growthindex defined as the ratio between the thicknessesof the same formation measured on both sides ofthe fault plane (Childs et al 2003) was used to

Figure 3 Faults oil fields key wells used in this study (black dots) other wells (black circles) and 3-D seismic data sets (dashed linerectangles) in the Chengbei step-fault zone Faults are mapped at the top of the Shahejie Formation Section AAprime is presented in Figure 4QD = Qidong YEZ = Yangerzhuang YEZN = Yangerzhuangnan ZB = Zaobei ZDH4 = Zhangdonghai 4 ZHB = Zhangbei

analyze the development and activity of growthfaults in the study area (Zhang et al 2007) Theyrevealed the occurrence of three major episodesof intensive faulting during the deposition of theShahejie Donying and upper Minghuazhen for-mations respectively The tectonic intensity hadbeen decreasing from the Paleogene to theNeogene

Basin modeling applied to the whole region(Figure 3) indicates that hydrocarbonwas expelledfrom source rocks during twomajor episodes (Wanget al 2006 Yu et al 2006) The first episode wasat the end of the Dongying (sim26Ma) when sourcerocks in the Sha-3member becamematureHydro-carbon expulsion ceased because of regional uplift-ing and erosion at the end of the Paleogene There-fore this early episode of hydrocarbon expulsionwas short and the amount of expelled hydrocar-bon was limited Major expulsion occurred duringthe second episode from the late Miocene to Qua-ternary (Wang et al 2006) The deposition of theNeogene Guantao and Minghuazhen formationscaused maturation and hydrocarbon generationin the Sha-3 and Sha-1 source rocks Hydrocarbonexpulsion peaked at the end of the Minghuazhen(sim2 Ma)

FAULT CONNECTIVITY ANDPROBABILISTIC INTERPRETATION

We characterized the fault opening or closing on astatistical basis and developed the concept offault-connectivity probability

Fault Opening and Closing

A fault is not a simple discontinuity but instead azone consisting of a fault core and a surroundingdam-age zone (Chester and Logan 1986 Smith et al1990 Forster and Evans 1991 Chester et al 1993Bruhn et al 1994 Caine et al 1996) (Figure 5)The fault core represents the part of a fault zonewhere most of the displacement is accommodatedand consists of slip surfaces fault gouge brecciasand cataclasites (Chester and Logan 1986 Chesteret al 1993 Caine et al 1996 Clausen et al2003) The damage zone may be composed of sub-sidiary faults faulted rocks and fractures (Chesterand Logan 1986 Chester et al 1993 Bruhn et al1994 Caine et al 1996 Jourde et al 2002) Thisdefinition of a fault zone requires the evaluationof its overall behavior and particularly that of its

Figure 4 Interpreted seismic section AAprime (see the location in Figure 3) showing major faults and stratigraphic intervals in this studyQD = Qidong ZDH4 = Zhangdonghai 4 YEZ = Yangerzhuang YEZN = Yangerzhuangnan

Zhang et al 1385

potential hydrocarbon-carrying capacities duringthe entire history of hydrocarbon migration

Furthermore consider that the commercialquantity of hydrocarbons having migrated throughfault zonesmust be a result of multiple events (suchas earthquakes) that occurred over a long-term geo-logic period The present distribution of hydrocar-bons along a fault is the consequence of these mul-tiple episodes of migration and accumulation Atany time any point on the fault has the potentialto act as a barrier or conduit However the behav-ior of this point during a single displacement eventcannot be evaluated because hydrocarbon migra-tion processes corresponding to this single faultingevent cannot be assessed Nevertheless the cumu-lative effects of these events can be identified solong as some of these events leave a preserved trace

Identifying Fault Connectivity

The presence or absence of hydrocarbon-filled res-ervoirs on both sides of a fault gives a quantitativeindication onwhether the fault has been open dur-

1386 Geohorizons

ing at least some of the events occurring as a func-tion of geological time For a given location on thefault an open and close indicator may be con-structed as explained in the following paragraphwhere it is applied to the CSFZ

This study concentrates on faults that havebeen drilled extensively on both their footwallsand hanging walls so that sufficient data are avail-able to better characterize the history of fault open-ing and closing Actually the CSFZ is particularlyconvenient for the use of hydrocarbons as an indi-cator of fault opening and closing This is becausehydrocarbons accumulatedprincipally during a rela-tively short period from the Miocene to Pliocene(Yu et al 2006) Since then the study area haskept subsiding without significant changes in tec-tonic style (Wang et al 2003)

As illustrated on Figure 3 the six main faultsof the CSFZ clearly connected with oil and gas dis-coveries are selected The precise relationship be-tween the fault geometry and the occurrence of hy-drocarbons in reservoirs may be assessed using datafrom exploration wells and 3-D seismic surveysFigure 6 presents an enlargement of the eastern part

Figure 5 Conceptual model of ele-ments in a fault zone (modified fromChester and Logan 1986 Smith et al1990 Bruhn et al 1994 Caine et al 1996)

Figure 6 Detailed configuration of the eastern part of the Zhangdonghai 4 (ZDH4) fault (a) A cross section showing the deep part of the fault zone and the occurrence of hydro-carbons The location is on panels b c and d (b c d) Maps showing the horizontal extent of the oil-bearing layer in Es2

2 Es21 and Es1 (see Figure 2) and the contour maps at the top of

these intervals The contour interval is 50 m

Zhangetal

1387

of fault zone ZDH4 A cross section perpendicularto the main fault (Figure 6a) displays the majorhorizons interpreted from 3-D seismic data togetherwith major reservoirs The lateral extent of the oil-bearing layer (as well as the oil-water contacts) andthe topography at the top of the Es2

2 Es21 and Es1

reservoirs (Figure 6bndashd) emphasize that in mostoccasions faults acted as a seal with respect to thereservoirs However hydrocarbons that accumu-lated in the footwall cannot be from the local sourcerock of Es1 or Es3 because it is poor and immaturein this area Oils must have been fed through open-ings of normal fault ZDH4 (Figure 6a) which ex-tend from the base of Es2

2 in the footwall to thetop of Es2

1 in the hanging wallOn a cross section of the fault plane we iden-

tify particular points (called grid nodes) as the topof each bed in both the hanging wall and the foot-wall (Figure 7a) Assuming that the fault is theonly possible pathway the opening and closingstate of such a node on the fault during the migra-tion period is determined by identifying the hy-drocarbon presence or absence in reservoirs above

1388 Geohorizons

that node on both sides of the fault plane Fourpossible scenarios may occur and are used to iden-tify whether the node had served as a migrationpath The four scenarios can be described as fol-lows (Figure 7b) (1) the reservoirs below containhydrocarbons and the ones above do not (the caseof point D on Figure 7b) (2) reservoirs both aboveand below the node contain hydrocarbons (thecase of point C on Figure 7b) (3) the reservoirsabove the node contain hydrocarbons and thosebelow the node do not (the case of points A andB on Figure 7b) and (4) reservoirs both aboveand below the node do not contain hydrocarbons(the cases of points E and F on Figure 7b) It is notdifficult to assess that in scenarios 2 and 3 the nodemust have been open at some time during migra-tion on the contrary in scenario 1 no hydrocarbonaccumulation above this node exists at present andthus the node was probably closed during migra-tion In the last scenario it is unknown whetherthe node was open or closed during migration

At the scale of the CFSZ many such cross sec-tions which include both structural information

Figure 7 Determination of fault connectivity on the basis of hydrocarbon occurrence (a) Schematic block diagram showing a numer-ical model of a normal fault and positions of several stratigraphic intervals at the footwall and hanging wall on the fault surface (b) Sketchdemonstrating the criteria used for identifying whether a fault was open From the occurrence of hydrocarbon nodes A B and C wereclearly open node D remains closed and the state of nodes E and F were not certain during migration Ni is defined as the fault-connectivity index For a given node Ni = 1 indicates the fault has opened for migration Ni = 0 indicates the fault kept closing and Ni = means the effect of the fault cannot be identified and the node will not be taken into account in statistical analysis

and indications of hydrocarbon occurrence maybe constructed A systematic study derived fromavailable seismic sections is therefore performedThe geometrical and structural data are first digi-tized for grid mapping of the main faults An ex-ample of such mapping is shown on Figure 8which presents a perspective view of the footwallof fault ZDH4 The traces of the main horizons(both of the hanging wall and of the footwall) aswell as the traces of seismic cross sections are thebasic elements of this grid Based on the mappingof the oil and gas shows assessing the connectiv-ity of the fault at various nodes of a grid is com-monly possible

Fault-Connectivity Indicator as aProbabilistic Criterion

Sixteen cross-well seismic sections perpendicularto six selected major faults are used to analyze thefault connectivity during migration At each inter-section point between the fault plane and a carrierbed (Figure 7b) a binary index termed the ldquofault-connectivity indexrdquo (Ni) is defined as 0 when evi-dence for migration is absent and 1 when evidencefor migration is present If the evidence is inconclu-sive the corresponding data point is discarded

About 120 data points are compiled each being as-signed with an openclose index

The relevance of this index to the actual open-ing and closing of a fault is debatable First it ispossible that part of the migration occurs in otherpathways instead of the main fault For instancesmall faults or cracks generally accompanying mas-ter faults in extensional basins could channel a sub-stantial fraction of the migrated hydrocarbonsThere also is an implicit assumption (Figure 7) thatat each data point hydrocarbon migrates in a 2-Dplane perpendicular to the fault In fact it is likelythat pathways along the fault are muchmore com-plex than being depicted here and their exact 3-Dgeometry is difficult to assess A statistical approachmay be effective in countering this complexity Ata given point on the fault it is convenient to con-sider the fault-connectivity index as a binary vari-able with a value of 0 or 1 The probability that theindex has a value of 1 depends on the position of thepoint on the fault and is a function of other variablesof the fault such as physical parameters character-izing the environment and history of the faultThe previously collected data allow the character-ization of this probability relationship as follows

For a specified value range of some physicalparameters associated with a fault (eg depth

Figure 8 Structural block diagram of the footwall of fault ZDH4 The positions of stratigraphic intervals are marked as solid lines on thefootwall and as dashed lines on the hanging wall

Zhang et al 1389

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

deposits with a total thickness of 2000ndash5000 m(6562ndash16404 ft) (Figure 2) The Paleogene synriftdeposits include the Shahejie and Dongying forma-tions The Neogene to Quaternary sequence con-sists of the Guantao Minghuazhen and Pingyuanformations A regional unconformity separatesPaleogene and Neogene strata (Yuan et al 2004)

Oilndashsource rock correlation shows that hydro-carbons in the CSFZ were derived from sourcerocks composed of dark-colored mudstones of theSha-1 and Sha-3 members of the Shahejie Forma-tion in the Qikou depression (Wang et al 2006)(Figure 2) The major reservoir rocks include fandeltaic and lacustrine littoral sandstones of theSha-2 and Sha-3 members of the Shahejie Forma-

tion delta-front and prodeltaic sandstones of theDong-1 andDong-3members of the Dongying For-mation and Neogene fluvial sandstones (Yuanet al 2004) (Figure 2) Thick mudstones in themiddle Sha-1 and Dong-2 members and lowerMinghuazhen Formation constitute regional sealrocks (Yu et al 2006) (Figure 2)

Petroleum exploration in the CSFZ began in1965 During the last 43 yr 11 oil and gas fieldshave beendiscovered (Figure 3)More than 140 ex-ploratory wells have been drilled and a 3-D seis-mic survey covers nearly all major structures Alarge quantity of exploration and developmentdata are available to study fault-seal capacity inthe CSFZ (Figure 3)

Figure 2 Summary diagram showing chrono- and lithostratigraphy tectonic stages and characteristics of the petroleum system in theChengbei step-fault zone (after Yuan et al 2004)

Zhang et al 1383

Movement of Faults andHydrocarbon Expulsion

The study area has experienced regional transten-sion since the end of theMesozoic (Li et al 1998)resulting in extensive faulting Major faults whichcontrolled the structural framework and deposi-tional history include the Qidong (QD) Zhang-donghai 4 (ZDH4) Zaobei (ZB) Yangerzhuang(YEZ) Yangerzhuangnan (YEZN) and Zhangbei(ZHB) faults (Yuan et al 2004 Yu et al 2006)(Figure 3) Seismic interpretation indicates that theyare synsedimentary faults (Figure 4) In general

1384 Geohorizons

these faults are parallel to each other striking in anortheast or nearly eastndashwest direction and dippingtoward the depression to the north The distribu-tion of discovered hydrocarbon accumulations isclosely related to the faults (Figure 3) These faultsappear to be important factors in hydrocarbon mi-gration and accumulation

In this study the incremental movements alonga given fault are assumed to be the result of multi-ple tectonic episodes in the CSFZ The fault growthindex defined as the ratio between the thicknessesof the same formation measured on both sides ofthe fault plane (Childs et al 2003) was used to

Figure 3 Faults oil fields key wells used in this study (black dots) other wells (black circles) and 3-D seismic data sets (dashed linerectangles) in the Chengbei step-fault zone Faults are mapped at the top of the Shahejie Formation Section AAprime is presented in Figure 4QD = Qidong YEZ = Yangerzhuang YEZN = Yangerzhuangnan ZB = Zaobei ZDH4 = Zhangdonghai 4 ZHB = Zhangbei

analyze the development and activity of growthfaults in the study area (Zhang et al 2007) Theyrevealed the occurrence of three major episodesof intensive faulting during the deposition of theShahejie Donying and upper Minghuazhen for-mations respectively The tectonic intensity hadbeen decreasing from the Paleogene to theNeogene

Basin modeling applied to the whole region(Figure 3) indicates that hydrocarbonwas expelledfrom source rocks during twomajor episodes (Wanget al 2006 Yu et al 2006) The first episode wasat the end of the Dongying (sim26Ma) when sourcerocks in the Sha-3member becamematureHydro-carbon expulsion ceased because of regional uplift-ing and erosion at the end of the Paleogene There-fore this early episode of hydrocarbon expulsionwas short and the amount of expelled hydrocar-bon was limited Major expulsion occurred duringthe second episode from the late Miocene to Qua-ternary (Wang et al 2006) The deposition of theNeogene Guantao and Minghuazhen formationscaused maturation and hydrocarbon generationin the Sha-3 and Sha-1 source rocks Hydrocarbonexpulsion peaked at the end of the Minghuazhen(sim2 Ma)

FAULT CONNECTIVITY ANDPROBABILISTIC INTERPRETATION

We characterized the fault opening or closing on astatistical basis and developed the concept offault-connectivity probability

Fault Opening and Closing

A fault is not a simple discontinuity but instead azone consisting of a fault core and a surroundingdam-age zone (Chester and Logan 1986 Smith et al1990 Forster and Evans 1991 Chester et al 1993Bruhn et al 1994 Caine et al 1996) (Figure 5)The fault core represents the part of a fault zonewhere most of the displacement is accommodatedand consists of slip surfaces fault gouge brecciasand cataclasites (Chester and Logan 1986 Chesteret al 1993 Caine et al 1996 Clausen et al2003) The damage zone may be composed of sub-sidiary faults faulted rocks and fractures (Chesterand Logan 1986 Chester et al 1993 Bruhn et al1994 Caine et al 1996 Jourde et al 2002) Thisdefinition of a fault zone requires the evaluationof its overall behavior and particularly that of its

Figure 4 Interpreted seismic section AAprime (see the location in Figure 3) showing major faults and stratigraphic intervals in this studyQD = Qidong ZDH4 = Zhangdonghai 4 YEZ = Yangerzhuang YEZN = Yangerzhuangnan

Zhang et al 1385

potential hydrocarbon-carrying capacities duringthe entire history of hydrocarbon migration

Furthermore consider that the commercialquantity of hydrocarbons having migrated throughfault zonesmust be a result of multiple events (suchas earthquakes) that occurred over a long-term geo-logic period The present distribution of hydrocar-bons along a fault is the consequence of these mul-tiple episodes of migration and accumulation Atany time any point on the fault has the potentialto act as a barrier or conduit However the behav-ior of this point during a single displacement eventcannot be evaluated because hydrocarbon migra-tion processes corresponding to this single faultingevent cannot be assessed Nevertheless the cumu-lative effects of these events can be identified solong as some of these events leave a preserved trace

Identifying Fault Connectivity

The presence or absence of hydrocarbon-filled res-ervoirs on both sides of a fault gives a quantitativeindication onwhether the fault has been open dur-

1386 Geohorizons

ing at least some of the events occurring as a func-tion of geological time For a given location on thefault an open and close indicator may be con-structed as explained in the following paragraphwhere it is applied to the CSFZ

This study concentrates on faults that havebeen drilled extensively on both their footwallsand hanging walls so that sufficient data are avail-able to better characterize the history of fault open-ing and closing Actually the CSFZ is particularlyconvenient for the use of hydrocarbons as an indi-cator of fault opening and closing This is becausehydrocarbons accumulatedprincipally during a rela-tively short period from the Miocene to Pliocene(Yu et al 2006) Since then the study area haskept subsiding without significant changes in tec-tonic style (Wang et al 2003)

As illustrated on Figure 3 the six main faultsof the CSFZ clearly connected with oil and gas dis-coveries are selected The precise relationship be-tween the fault geometry and the occurrence of hy-drocarbons in reservoirs may be assessed using datafrom exploration wells and 3-D seismic surveysFigure 6 presents an enlargement of the eastern part

Figure 5 Conceptual model of ele-ments in a fault zone (modified fromChester and Logan 1986 Smith et al1990 Bruhn et al 1994 Caine et al 1996)

Figure 6 Detailed configuration of the eastern part of the Zhangdonghai 4 (ZDH4) fault (a) A cross section showing the deep part of the fault zone and the occurrence of hydro-carbons The location is on panels b c and d (b c d) Maps showing the horizontal extent of the oil-bearing layer in Es2

2 Es21 and Es1 (see Figure 2) and the contour maps at the top of

these intervals The contour interval is 50 m

Zhangetal

1387

of fault zone ZDH4 A cross section perpendicularto the main fault (Figure 6a) displays the majorhorizons interpreted from 3-D seismic data togetherwith major reservoirs The lateral extent of the oil-bearing layer (as well as the oil-water contacts) andthe topography at the top of the Es2

2 Es21 and Es1

reservoirs (Figure 6bndashd) emphasize that in mostoccasions faults acted as a seal with respect to thereservoirs However hydrocarbons that accumu-lated in the footwall cannot be from the local sourcerock of Es1 or Es3 because it is poor and immaturein this area Oils must have been fed through open-ings of normal fault ZDH4 (Figure 6a) which ex-tend from the base of Es2

2 in the footwall to thetop of Es2

1 in the hanging wallOn a cross section of the fault plane we iden-

tify particular points (called grid nodes) as the topof each bed in both the hanging wall and the foot-wall (Figure 7a) Assuming that the fault is theonly possible pathway the opening and closingstate of such a node on the fault during the migra-tion period is determined by identifying the hy-drocarbon presence or absence in reservoirs above

1388 Geohorizons

that node on both sides of the fault plane Fourpossible scenarios may occur and are used to iden-tify whether the node had served as a migrationpath The four scenarios can be described as fol-lows (Figure 7b) (1) the reservoirs below containhydrocarbons and the ones above do not (the caseof point D on Figure 7b) (2) reservoirs both aboveand below the node contain hydrocarbons (thecase of point C on Figure 7b) (3) the reservoirsabove the node contain hydrocarbons and thosebelow the node do not (the case of points A andB on Figure 7b) and (4) reservoirs both aboveand below the node do not contain hydrocarbons(the cases of points E and F on Figure 7b) It is notdifficult to assess that in scenarios 2 and 3 the nodemust have been open at some time during migra-tion on the contrary in scenario 1 no hydrocarbonaccumulation above this node exists at present andthus the node was probably closed during migra-tion In the last scenario it is unknown whetherthe node was open or closed during migration

At the scale of the CFSZ many such cross sec-tions which include both structural information

Figure 7 Determination of fault connectivity on the basis of hydrocarbon occurrence (a) Schematic block diagram showing a numer-ical model of a normal fault and positions of several stratigraphic intervals at the footwall and hanging wall on the fault surface (b) Sketchdemonstrating the criteria used for identifying whether a fault was open From the occurrence of hydrocarbon nodes A B and C wereclearly open node D remains closed and the state of nodes E and F were not certain during migration Ni is defined as the fault-connectivity index For a given node Ni = 1 indicates the fault has opened for migration Ni = 0 indicates the fault kept closing and Ni = means the effect of the fault cannot be identified and the node will not be taken into account in statistical analysis

and indications of hydrocarbon occurrence maybe constructed A systematic study derived fromavailable seismic sections is therefore performedThe geometrical and structural data are first digi-tized for grid mapping of the main faults An ex-ample of such mapping is shown on Figure 8which presents a perspective view of the footwallof fault ZDH4 The traces of the main horizons(both of the hanging wall and of the footwall) aswell as the traces of seismic cross sections are thebasic elements of this grid Based on the mappingof the oil and gas shows assessing the connectiv-ity of the fault at various nodes of a grid is com-monly possible

Fault-Connectivity Indicator as aProbabilistic Criterion

Sixteen cross-well seismic sections perpendicularto six selected major faults are used to analyze thefault connectivity during migration At each inter-section point between the fault plane and a carrierbed (Figure 7b) a binary index termed the ldquofault-connectivity indexrdquo (Ni) is defined as 0 when evi-dence for migration is absent and 1 when evidencefor migration is present If the evidence is inconclu-sive the corresponding data point is discarded

About 120 data points are compiled each being as-signed with an openclose index

The relevance of this index to the actual open-ing and closing of a fault is debatable First it ispossible that part of the migration occurs in otherpathways instead of the main fault For instancesmall faults or cracks generally accompanying mas-ter faults in extensional basins could channel a sub-stantial fraction of the migrated hydrocarbonsThere also is an implicit assumption (Figure 7) thatat each data point hydrocarbon migrates in a 2-Dplane perpendicular to the fault In fact it is likelythat pathways along the fault are muchmore com-plex than being depicted here and their exact 3-Dgeometry is difficult to assess A statistical approachmay be effective in countering this complexity Ata given point on the fault it is convenient to con-sider the fault-connectivity index as a binary vari-able with a value of 0 or 1 The probability that theindex has a value of 1 depends on the position of thepoint on the fault and is a function of other variablesof the fault such as physical parameters character-izing the environment and history of the faultThe previously collected data allow the character-ization of this probability relationship as follows

For a specified value range of some physicalparameters associated with a fault (eg depth

Figure 8 Structural block diagram of the footwall of fault ZDH4 The positions of stratigraphic intervals are marked as solid lines on thefootwall and as dashed lines on the hanging wall

Zhang et al 1389

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

Movement of Faults andHydrocarbon Expulsion

The study area has experienced regional transten-sion since the end of theMesozoic (Li et al 1998)resulting in extensive faulting Major faults whichcontrolled the structural framework and deposi-tional history include the Qidong (QD) Zhang-donghai 4 (ZDH4) Zaobei (ZB) Yangerzhuang(YEZ) Yangerzhuangnan (YEZN) and Zhangbei(ZHB) faults (Yuan et al 2004 Yu et al 2006)(Figure 3) Seismic interpretation indicates that theyare synsedimentary faults (Figure 4) In general

1384 Geohorizons

these faults are parallel to each other striking in anortheast or nearly eastndashwest direction and dippingtoward the depression to the north The distribu-tion of discovered hydrocarbon accumulations isclosely related to the faults (Figure 3) These faultsappear to be important factors in hydrocarbon mi-gration and accumulation

In this study the incremental movements alonga given fault are assumed to be the result of multi-ple tectonic episodes in the CSFZ The fault growthindex defined as the ratio between the thicknessesof the same formation measured on both sides ofthe fault plane (Childs et al 2003) was used to

Figure 3 Faults oil fields key wells used in this study (black dots) other wells (black circles) and 3-D seismic data sets (dashed linerectangles) in the Chengbei step-fault zone Faults are mapped at the top of the Shahejie Formation Section AAprime is presented in Figure 4QD = Qidong YEZ = Yangerzhuang YEZN = Yangerzhuangnan ZB = Zaobei ZDH4 = Zhangdonghai 4 ZHB = Zhangbei

analyze the development and activity of growthfaults in the study area (Zhang et al 2007) Theyrevealed the occurrence of three major episodesof intensive faulting during the deposition of theShahejie Donying and upper Minghuazhen for-mations respectively The tectonic intensity hadbeen decreasing from the Paleogene to theNeogene

Basin modeling applied to the whole region(Figure 3) indicates that hydrocarbonwas expelledfrom source rocks during twomajor episodes (Wanget al 2006 Yu et al 2006) The first episode wasat the end of the Dongying (sim26Ma) when sourcerocks in the Sha-3member becamematureHydro-carbon expulsion ceased because of regional uplift-ing and erosion at the end of the Paleogene There-fore this early episode of hydrocarbon expulsionwas short and the amount of expelled hydrocar-bon was limited Major expulsion occurred duringthe second episode from the late Miocene to Qua-ternary (Wang et al 2006) The deposition of theNeogene Guantao and Minghuazhen formationscaused maturation and hydrocarbon generationin the Sha-3 and Sha-1 source rocks Hydrocarbonexpulsion peaked at the end of the Minghuazhen(sim2 Ma)

FAULT CONNECTIVITY ANDPROBABILISTIC INTERPRETATION

We characterized the fault opening or closing on astatistical basis and developed the concept offault-connectivity probability

Fault Opening and Closing

A fault is not a simple discontinuity but instead azone consisting of a fault core and a surroundingdam-age zone (Chester and Logan 1986 Smith et al1990 Forster and Evans 1991 Chester et al 1993Bruhn et al 1994 Caine et al 1996) (Figure 5)The fault core represents the part of a fault zonewhere most of the displacement is accommodatedand consists of slip surfaces fault gouge brecciasand cataclasites (Chester and Logan 1986 Chesteret al 1993 Caine et al 1996 Clausen et al2003) The damage zone may be composed of sub-sidiary faults faulted rocks and fractures (Chesterand Logan 1986 Chester et al 1993 Bruhn et al1994 Caine et al 1996 Jourde et al 2002) Thisdefinition of a fault zone requires the evaluationof its overall behavior and particularly that of its

Figure 4 Interpreted seismic section AAprime (see the location in Figure 3) showing major faults and stratigraphic intervals in this studyQD = Qidong ZDH4 = Zhangdonghai 4 YEZ = Yangerzhuang YEZN = Yangerzhuangnan

Zhang et al 1385

potential hydrocarbon-carrying capacities duringthe entire history of hydrocarbon migration

Furthermore consider that the commercialquantity of hydrocarbons having migrated throughfault zonesmust be a result of multiple events (suchas earthquakes) that occurred over a long-term geo-logic period The present distribution of hydrocar-bons along a fault is the consequence of these mul-tiple episodes of migration and accumulation Atany time any point on the fault has the potentialto act as a barrier or conduit However the behav-ior of this point during a single displacement eventcannot be evaluated because hydrocarbon migra-tion processes corresponding to this single faultingevent cannot be assessed Nevertheless the cumu-lative effects of these events can be identified solong as some of these events leave a preserved trace

Identifying Fault Connectivity

The presence or absence of hydrocarbon-filled res-ervoirs on both sides of a fault gives a quantitativeindication onwhether the fault has been open dur-

1386 Geohorizons

ing at least some of the events occurring as a func-tion of geological time For a given location on thefault an open and close indicator may be con-structed as explained in the following paragraphwhere it is applied to the CSFZ

This study concentrates on faults that havebeen drilled extensively on both their footwallsand hanging walls so that sufficient data are avail-able to better characterize the history of fault open-ing and closing Actually the CSFZ is particularlyconvenient for the use of hydrocarbons as an indi-cator of fault opening and closing This is becausehydrocarbons accumulatedprincipally during a rela-tively short period from the Miocene to Pliocene(Yu et al 2006) Since then the study area haskept subsiding without significant changes in tec-tonic style (Wang et al 2003)

As illustrated on Figure 3 the six main faultsof the CSFZ clearly connected with oil and gas dis-coveries are selected The precise relationship be-tween the fault geometry and the occurrence of hy-drocarbons in reservoirs may be assessed using datafrom exploration wells and 3-D seismic surveysFigure 6 presents an enlargement of the eastern part

Figure 5 Conceptual model of ele-ments in a fault zone (modified fromChester and Logan 1986 Smith et al1990 Bruhn et al 1994 Caine et al 1996)

Figure 6 Detailed configuration of the eastern part of the Zhangdonghai 4 (ZDH4) fault (a) A cross section showing the deep part of the fault zone and the occurrence of hydro-carbons The location is on panels b c and d (b c d) Maps showing the horizontal extent of the oil-bearing layer in Es2

2 Es21 and Es1 (see Figure 2) and the contour maps at the top of

these intervals The contour interval is 50 m

Zhangetal

1387

of fault zone ZDH4 A cross section perpendicularto the main fault (Figure 6a) displays the majorhorizons interpreted from 3-D seismic data togetherwith major reservoirs The lateral extent of the oil-bearing layer (as well as the oil-water contacts) andthe topography at the top of the Es2

2 Es21 and Es1

reservoirs (Figure 6bndashd) emphasize that in mostoccasions faults acted as a seal with respect to thereservoirs However hydrocarbons that accumu-lated in the footwall cannot be from the local sourcerock of Es1 or Es3 because it is poor and immaturein this area Oils must have been fed through open-ings of normal fault ZDH4 (Figure 6a) which ex-tend from the base of Es2

2 in the footwall to thetop of Es2

1 in the hanging wallOn a cross section of the fault plane we iden-

tify particular points (called grid nodes) as the topof each bed in both the hanging wall and the foot-wall (Figure 7a) Assuming that the fault is theonly possible pathway the opening and closingstate of such a node on the fault during the migra-tion period is determined by identifying the hy-drocarbon presence or absence in reservoirs above

1388 Geohorizons

that node on both sides of the fault plane Fourpossible scenarios may occur and are used to iden-tify whether the node had served as a migrationpath The four scenarios can be described as fol-lows (Figure 7b) (1) the reservoirs below containhydrocarbons and the ones above do not (the caseof point D on Figure 7b) (2) reservoirs both aboveand below the node contain hydrocarbons (thecase of point C on Figure 7b) (3) the reservoirsabove the node contain hydrocarbons and thosebelow the node do not (the case of points A andB on Figure 7b) and (4) reservoirs both aboveand below the node do not contain hydrocarbons(the cases of points E and F on Figure 7b) It is notdifficult to assess that in scenarios 2 and 3 the nodemust have been open at some time during migra-tion on the contrary in scenario 1 no hydrocarbonaccumulation above this node exists at present andthus the node was probably closed during migra-tion In the last scenario it is unknown whetherthe node was open or closed during migration

At the scale of the CFSZ many such cross sec-tions which include both structural information

Figure 7 Determination of fault connectivity on the basis of hydrocarbon occurrence (a) Schematic block diagram showing a numer-ical model of a normal fault and positions of several stratigraphic intervals at the footwall and hanging wall on the fault surface (b) Sketchdemonstrating the criteria used for identifying whether a fault was open From the occurrence of hydrocarbon nodes A B and C wereclearly open node D remains closed and the state of nodes E and F were not certain during migration Ni is defined as the fault-connectivity index For a given node Ni = 1 indicates the fault has opened for migration Ni = 0 indicates the fault kept closing and Ni = means the effect of the fault cannot be identified and the node will not be taken into account in statistical analysis

and indications of hydrocarbon occurrence maybe constructed A systematic study derived fromavailable seismic sections is therefore performedThe geometrical and structural data are first digi-tized for grid mapping of the main faults An ex-ample of such mapping is shown on Figure 8which presents a perspective view of the footwallof fault ZDH4 The traces of the main horizons(both of the hanging wall and of the footwall) aswell as the traces of seismic cross sections are thebasic elements of this grid Based on the mappingof the oil and gas shows assessing the connectiv-ity of the fault at various nodes of a grid is com-monly possible

Fault-Connectivity Indicator as aProbabilistic Criterion

Sixteen cross-well seismic sections perpendicularto six selected major faults are used to analyze thefault connectivity during migration At each inter-section point between the fault plane and a carrierbed (Figure 7b) a binary index termed the ldquofault-connectivity indexrdquo (Ni) is defined as 0 when evi-dence for migration is absent and 1 when evidencefor migration is present If the evidence is inconclu-sive the corresponding data point is discarded

About 120 data points are compiled each being as-signed with an openclose index

The relevance of this index to the actual open-ing and closing of a fault is debatable First it ispossible that part of the migration occurs in otherpathways instead of the main fault For instancesmall faults or cracks generally accompanying mas-ter faults in extensional basins could channel a sub-stantial fraction of the migrated hydrocarbonsThere also is an implicit assumption (Figure 7) thatat each data point hydrocarbon migrates in a 2-Dplane perpendicular to the fault In fact it is likelythat pathways along the fault are muchmore com-plex than being depicted here and their exact 3-Dgeometry is difficult to assess A statistical approachmay be effective in countering this complexity Ata given point on the fault it is convenient to con-sider the fault-connectivity index as a binary vari-able with a value of 0 or 1 The probability that theindex has a value of 1 depends on the position of thepoint on the fault and is a function of other variablesof the fault such as physical parameters character-izing the environment and history of the faultThe previously collected data allow the character-ization of this probability relationship as follows

For a specified value range of some physicalparameters associated with a fault (eg depth

Figure 8 Structural block diagram of the footwall of fault ZDH4 The positions of stratigraphic intervals are marked as solid lines on thefootwall and as dashed lines on the hanging wall

Zhang et al 1389

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

analyze the development and activity of growthfaults in the study area (Zhang et al 2007) Theyrevealed the occurrence of three major episodesof intensive faulting during the deposition of theShahejie Donying and upper Minghuazhen for-mations respectively The tectonic intensity hadbeen decreasing from the Paleogene to theNeogene

Basin modeling applied to the whole region(Figure 3) indicates that hydrocarbonwas expelledfrom source rocks during twomajor episodes (Wanget al 2006 Yu et al 2006) The first episode wasat the end of the Dongying (sim26Ma) when sourcerocks in the Sha-3member becamematureHydro-carbon expulsion ceased because of regional uplift-ing and erosion at the end of the Paleogene There-fore this early episode of hydrocarbon expulsionwas short and the amount of expelled hydrocar-bon was limited Major expulsion occurred duringthe second episode from the late Miocene to Qua-ternary (Wang et al 2006) The deposition of theNeogene Guantao and Minghuazhen formationscaused maturation and hydrocarbon generationin the Sha-3 and Sha-1 source rocks Hydrocarbonexpulsion peaked at the end of the Minghuazhen(sim2 Ma)

FAULT CONNECTIVITY ANDPROBABILISTIC INTERPRETATION

We characterized the fault opening or closing on astatistical basis and developed the concept offault-connectivity probability

Fault Opening and Closing

A fault is not a simple discontinuity but instead azone consisting of a fault core and a surroundingdam-age zone (Chester and Logan 1986 Smith et al1990 Forster and Evans 1991 Chester et al 1993Bruhn et al 1994 Caine et al 1996) (Figure 5)The fault core represents the part of a fault zonewhere most of the displacement is accommodatedand consists of slip surfaces fault gouge brecciasand cataclasites (Chester and Logan 1986 Chesteret al 1993 Caine et al 1996 Clausen et al2003) The damage zone may be composed of sub-sidiary faults faulted rocks and fractures (Chesterand Logan 1986 Chester et al 1993 Bruhn et al1994 Caine et al 1996 Jourde et al 2002) Thisdefinition of a fault zone requires the evaluationof its overall behavior and particularly that of its

Figure 4 Interpreted seismic section AAprime (see the location in Figure 3) showing major faults and stratigraphic intervals in this studyQD = Qidong ZDH4 = Zhangdonghai 4 YEZ = Yangerzhuang YEZN = Yangerzhuangnan

Zhang et al 1385

potential hydrocarbon-carrying capacities duringthe entire history of hydrocarbon migration

Furthermore consider that the commercialquantity of hydrocarbons having migrated throughfault zonesmust be a result of multiple events (suchas earthquakes) that occurred over a long-term geo-logic period The present distribution of hydrocar-bons along a fault is the consequence of these mul-tiple episodes of migration and accumulation Atany time any point on the fault has the potentialto act as a barrier or conduit However the behav-ior of this point during a single displacement eventcannot be evaluated because hydrocarbon migra-tion processes corresponding to this single faultingevent cannot be assessed Nevertheless the cumu-lative effects of these events can be identified solong as some of these events leave a preserved trace

Identifying Fault Connectivity

The presence or absence of hydrocarbon-filled res-ervoirs on both sides of a fault gives a quantitativeindication onwhether the fault has been open dur-

1386 Geohorizons

ing at least some of the events occurring as a func-tion of geological time For a given location on thefault an open and close indicator may be con-structed as explained in the following paragraphwhere it is applied to the CSFZ

This study concentrates on faults that havebeen drilled extensively on both their footwallsand hanging walls so that sufficient data are avail-able to better characterize the history of fault open-ing and closing Actually the CSFZ is particularlyconvenient for the use of hydrocarbons as an indi-cator of fault opening and closing This is becausehydrocarbons accumulatedprincipally during a rela-tively short period from the Miocene to Pliocene(Yu et al 2006) Since then the study area haskept subsiding without significant changes in tec-tonic style (Wang et al 2003)

As illustrated on Figure 3 the six main faultsof the CSFZ clearly connected with oil and gas dis-coveries are selected The precise relationship be-tween the fault geometry and the occurrence of hy-drocarbons in reservoirs may be assessed using datafrom exploration wells and 3-D seismic surveysFigure 6 presents an enlargement of the eastern part

Figure 5 Conceptual model of ele-ments in a fault zone (modified fromChester and Logan 1986 Smith et al1990 Bruhn et al 1994 Caine et al 1996)

Figure 6 Detailed configuration of the eastern part of the Zhangdonghai 4 (ZDH4) fault (a) A cross section showing the deep part of the fault zone and the occurrence of hydro-carbons The location is on panels b c and d (b c d) Maps showing the horizontal extent of the oil-bearing layer in Es2

2 Es21 and Es1 (see Figure 2) and the contour maps at the top of

these intervals The contour interval is 50 m

Zhangetal

1387

of fault zone ZDH4 A cross section perpendicularto the main fault (Figure 6a) displays the majorhorizons interpreted from 3-D seismic data togetherwith major reservoirs The lateral extent of the oil-bearing layer (as well as the oil-water contacts) andthe topography at the top of the Es2

2 Es21 and Es1

reservoirs (Figure 6bndashd) emphasize that in mostoccasions faults acted as a seal with respect to thereservoirs However hydrocarbons that accumu-lated in the footwall cannot be from the local sourcerock of Es1 or Es3 because it is poor and immaturein this area Oils must have been fed through open-ings of normal fault ZDH4 (Figure 6a) which ex-tend from the base of Es2

2 in the footwall to thetop of Es2

1 in the hanging wallOn a cross section of the fault plane we iden-

tify particular points (called grid nodes) as the topof each bed in both the hanging wall and the foot-wall (Figure 7a) Assuming that the fault is theonly possible pathway the opening and closingstate of such a node on the fault during the migra-tion period is determined by identifying the hy-drocarbon presence or absence in reservoirs above

1388 Geohorizons

that node on both sides of the fault plane Fourpossible scenarios may occur and are used to iden-tify whether the node had served as a migrationpath The four scenarios can be described as fol-lows (Figure 7b) (1) the reservoirs below containhydrocarbons and the ones above do not (the caseof point D on Figure 7b) (2) reservoirs both aboveand below the node contain hydrocarbons (thecase of point C on Figure 7b) (3) the reservoirsabove the node contain hydrocarbons and thosebelow the node do not (the case of points A andB on Figure 7b) and (4) reservoirs both aboveand below the node do not contain hydrocarbons(the cases of points E and F on Figure 7b) It is notdifficult to assess that in scenarios 2 and 3 the nodemust have been open at some time during migra-tion on the contrary in scenario 1 no hydrocarbonaccumulation above this node exists at present andthus the node was probably closed during migra-tion In the last scenario it is unknown whetherthe node was open or closed during migration

At the scale of the CFSZ many such cross sec-tions which include both structural information

Figure 7 Determination of fault connectivity on the basis of hydrocarbon occurrence (a) Schematic block diagram showing a numer-ical model of a normal fault and positions of several stratigraphic intervals at the footwall and hanging wall on the fault surface (b) Sketchdemonstrating the criteria used for identifying whether a fault was open From the occurrence of hydrocarbon nodes A B and C wereclearly open node D remains closed and the state of nodes E and F were not certain during migration Ni is defined as the fault-connectivity index For a given node Ni = 1 indicates the fault has opened for migration Ni = 0 indicates the fault kept closing and Ni = means the effect of the fault cannot be identified and the node will not be taken into account in statistical analysis

and indications of hydrocarbon occurrence maybe constructed A systematic study derived fromavailable seismic sections is therefore performedThe geometrical and structural data are first digi-tized for grid mapping of the main faults An ex-ample of such mapping is shown on Figure 8which presents a perspective view of the footwallof fault ZDH4 The traces of the main horizons(both of the hanging wall and of the footwall) aswell as the traces of seismic cross sections are thebasic elements of this grid Based on the mappingof the oil and gas shows assessing the connectiv-ity of the fault at various nodes of a grid is com-monly possible

Fault-Connectivity Indicator as aProbabilistic Criterion

Sixteen cross-well seismic sections perpendicularto six selected major faults are used to analyze thefault connectivity during migration At each inter-section point between the fault plane and a carrierbed (Figure 7b) a binary index termed the ldquofault-connectivity indexrdquo (Ni) is defined as 0 when evi-dence for migration is absent and 1 when evidencefor migration is present If the evidence is inconclu-sive the corresponding data point is discarded

About 120 data points are compiled each being as-signed with an openclose index

The relevance of this index to the actual open-ing and closing of a fault is debatable First it ispossible that part of the migration occurs in otherpathways instead of the main fault For instancesmall faults or cracks generally accompanying mas-ter faults in extensional basins could channel a sub-stantial fraction of the migrated hydrocarbonsThere also is an implicit assumption (Figure 7) thatat each data point hydrocarbon migrates in a 2-Dplane perpendicular to the fault In fact it is likelythat pathways along the fault are muchmore com-plex than being depicted here and their exact 3-Dgeometry is difficult to assess A statistical approachmay be effective in countering this complexity Ata given point on the fault it is convenient to con-sider the fault-connectivity index as a binary vari-able with a value of 0 or 1 The probability that theindex has a value of 1 depends on the position of thepoint on the fault and is a function of other variablesof the fault such as physical parameters character-izing the environment and history of the faultThe previously collected data allow the character-ization of this probability relationship as follows

For a specified value range of some physicalparameters associated with a fault (eg depth

Figure 8 Structural block diagram of the footwall of fault ZDH4 The positions of stratigraphic intervals are marked as solid lines on thefootwall and as dashed lines on the hanging wall

Zhang et al 1389

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

potential hydrocarbon-carrying capacities duringthe entire history of hydrocarbon migration

Furthermore consider that the commercialquantity of hydrocarbons having migrated throughfault zonesmust be a result of multiple events (suchas earthquakes) that occurred over a long-term geo-logic period The present distribution of hydrocar-bons along a fault is the consequence of these mul-tiple episodes of migration and accumulation Atany time any point on the fault has the potentialto act as a barrier or conduit However the behav-ior of this point during a single displacement eventcannot be evaluated because hydrocarbon migra-tion processes corresponding to this single faultingevent cannot be assessed Nevertheless the cumu-lative effects of these events can be identified solong as some of these events leave a preserved trace

Identifying Fault Connectivity

The presence or absence of hydrocarbon-filled res-ervoirs on both sides of a fault gives a quantitativeindication onwhether the fault has been open dur-

1386 Geohorizons

ing at least some of the events occurring as a func-tion of geological time For a given location on thefault an open and close indicator may be con-structed as explained in the following paragraphwhere it is applied to the CSFZ

This study concentrates on faults that havebeen drilled extensively on both their footwallsand hanging walls so that sufficient data are avail-able to better characterize the history of fault open-ing and closing Actually the CSFZ is particularlyconvenient for the use of hydrocarbons as an indi-cator of fault opening and closing This is becausehydrocarbons accumulatedprincipally during a rela-tively short period from the Miocene to Pliocene(Yu et al 2006) Since then the study area haskept subsiding without significant changes in tec-tonic style (Wang et al 2003)

As illustrated on Figure 3 the six main faultsof the CSFZ clearly connected with oil and gas dis-coveries are selected The precise relationship be-tween the fault geometry and the occurrence of hy-drocarbons in reservoirs may be assessed using datafrom exploration wells and 3-D seismic surveysFigure 6 presents an enlargement of the eastern part

Figure 5 Conceptual model of ele-ments in a fault zone (modified fromChester and Logan 1986 Smith et al1990 Bruhn et al 1994 Caine et al 1996)

Figure 6 Detailed configuration of the eastern part of the Zhangdonghai 4 (ZDH4) fault (a) A cross section showing the deep part of the fault zone and the occurrence of hydro-carbons The location is on panels b c and d (b c d) Maps showing the horizontal extent of the oil-bearing layer in Es2

2 Es21 and Es1 (see Figure 2) and the contour maps at the top of

these intervals The contour interval is 50 m

Zhangetal

1387

of fault zone ZDH4 A cross section perpendicularto the main fault (Figure 6a) displays the majorhorizons interpreted from 3-D seismic data togetherwith major reservoirs The lateral extent of the oil-bearing layer (as well as the oil-water contacts) andthe topography at the top of the Es2

2 Es21 and Es1

reservoirs (Figure 6bndashd) emphasize that in mostoccasions faults acted as a seal with respect to thereservoirs However hydrocarbons that accumu-lated in the footwall cannot be from the local sourcerock of Es1 or Es3 because it is poor and immaturein this area Oils must have been fed through open-ings of normal fault ZDH4 (Figure 6a) which ex-tend from the base of Es2

2 in the footwall to thetop of Es2

1 in the hanging wallOn a cross section of the fault plane we iden-

tify particular points (called grid nodes) as the topof each bed in both the hanging wall and the foot-wall (Figure 7a) Assuming that the fault is theonly possible pathway the opening and closingstate of such a node on the fault during the migra-tion period is determined by identifying the hy-drocarbon presence or absence in reservoirs above

1388 Geohorizons

that node on both sides of the fault plane Fourpossible scenarios may occur and are used to iden-tify whether the node had served as a migrationpath The four scenarios can be described as fol-lows (Figure 7b) (1) the reservoirs below containhydrocarbons and the ones above do not (the caseof point D on Figure 7b) (2) reservoirs both aboveand below the node contain hydrocarbons (thecase of point C on Figure 7b) (3) the reservoirsabove the node contain hydrocarbons and thosebelow the node do not (the case of points A andB on Figure 7b) and (4) reservoirs both aboveand below the node do not contain hydrocarbons(the cases of points E and F on Figure 7b) It is notdifficult to assess that in scenarios 2 and 3 the nodemust have been open at some time during migra-tion on the contrary in scenario 1 no hydrocarbonaccumulation above this node exists at present andthus the node was probably closed during migra-tion In the last scenario it is unknown whetherthe node was open or closed during migration

At the scale of the CFSZ many such cross sec-tions which include both structural information

Figure 7 Determination of fault connectivity on the basis of hydrocarbon occurrence (a) Schematic block diagram showing a numer-ical model of a normal fault and positions of several stratigraphic intervals at the footwall and hanging wall on the fault surface (b) Sketchdemonstrating the criteria used for identifying whether a fault was open From the occurrence of hydrocarbon nodes A B and C wereclearly open node D remains closed and the state of nodes E and F were not certain during migration Ni is defined as the fault-connectivity index For a given node Ni = 1 indicates the fault has opened for migration Ni = 0 indicates the fault kept closing and Ni = means the effect of the fault cannot be identified and the node will not be taken into account in statistical analysis

and indications of hydrocarbon occurrence maybe constructed A systematic study derived fromavailable seismic sections is therefore performedThe geometrical and structural data are first digi-tized for grid mapping of the main faults An ex-ample of such mapping is shown on Figure 8which presents a perspective view of the footwallof fault ZDH4 The traces of the main horizons(both of the hanging wall and of the footwall) aswell as the traces of seismic cross sections are thebasic elements of this grid Based on the mappingof the oil and gas shows assessing the connectiv-ity of the fault at various nodes of a grid is com-monly possible

Fault-Connectivity Indicator as aProbabilistic Criterion

Sixteen cross-well seismic sections perpendicularto six selected major faults are used to analyze thefault connectivity during migration At each inter-section point between the fault plane and a carrierbed (Figure 7b) a binary index termed the ldquofault-connectivity indexrdquo (Ni) is defined as 0 when evi-dence for migration is absent and 1 when evidencefor migration is present If the evidence is inconclu-sive the corresponding data point is discarded

About 120 data points are compiled each being as-signed with an openclose index

The relevance of this index to the actual open-ing and closing of a fault is debatable First it ispossible that part of the migration occurs in otherpathways instead of the main fault For instancesmall faults or cracks generally accompanying mas-ter faults in extensional basins could channel a sub-stantial fraction of the migrated hydrocarbonsThere also is an implicit assumption (Figure 7) thatat each data point hydrocarbon migrates in a 2-Dplane perpendicular to the fault In fact it is likelythat pathways along the fault are muchmore com-plex than being depicted here and their exact 3-Dgeometry is difficult to assess A statistical approachmay be effective in countering this complexity Ata given point on the fault it is convenient to con-sider the fault-connectivity index as a binary vari-able with a value of 0 or 1 The probability that theindex has a value of 1 depends on the position of thepoint on the fault and is a function of other variablesof the fault such as physical parameters character-izing the environment and history of the faultThe previously collected data allow the character-ization of this probability relationship as follows

For a specified value range of some physicalparameters associated with a fault (eg depth

Figure 8 Structural block diagram of the footwall of fault ZDH4 The positions of stratigraphic intervals are marked as solid lines on thefootwall and as dashed lines on the hanging wall

Zhang et al 1389

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

Figure 6 Detailed configuration of the eastern part of the Zhangdonghai 4 (ZDH4) fault (a) A cross section showing the deep part of the fault zone and the occurrence of hydro-carbons The location is on panels b c and d (b c d) Maps showing the horizontal extent of the oil-bearing layer in Es2

2 Es21 and Es1 (see Figure 2) and the contour maps at the top of

these intervals The contour interval is 50 m

Zhangetal

1387

of fault zone ZDH4 A cross section perpendicularto the main fault (Figure 6a) displays the majorhorizons interpreted from 3-D seismic data togetherwith major reservoirs The lateral extent of the oil-bearing layer (as well as the oil-water contacts) andthe topography at the top of the Es2

2 Es21 and Es1

reservoirs (Figure 6bndashd) emphasize that in mostoccasions faults acted as a seal with respect to thereservoirs However hydrocarbons that accumu-lated in the footwall cannot be from the local sourcerock of Es1 or Es3 because it is poor and immaturein this area Oils must have been fed through open-ings of normal fault ZDH4 (Figure 6a) which ex-tend from the base of Es2

2 in the footwall to thetop of Es2

1 in the hanging wallOn a cross section of the fault plane we iden-

tify particular points (called grid nodes) as the topof each bed in both the hanging wall and the foot-wall (Figure 7a) Assuming that the fault is theonly possible pathway the opening and closingstate of such a node on the fault during the migra-tion period is determined by identifying the hy-drocarbon presence or absence in reservoirs above

1388 Geohorizons

that node on both sides of the fault plane Fourpossible scenarios may occur and are used to iden-tify whether the node had served as a migrationpath The four scenarios can be described as fol-lows (Figure 7b) (1) the reservoirs below containhydrocarbons and the ones above do not (the caseof point D on Figure 7b) (2) reservoirs both aboveand below the node contain hydrocarbons (thecase of point C on Figure 7b) (3) the reservoirsabove the node contain hydrocarbons and thosebelow the node do not (the case of points A andB on Figure 7b) and (4) reservoirs both aboveand below the node do not contain hydrocarbons(the cases of points E and F on Figure 7b) It is notdifficult to assess that in scenarios 2 and 3 the nodemust have been open at some time during migra-tion on the contrary in scenario 1 no hydrocarbonaccumulation above this node exists at present andthus the node was probably closed during migra-tion In the last scenario it is unknown whetherthe node was open or closed during migration

At the scale of the CFSZ many such cross sec-tions which include both structural information

Figure 7 Determination of fault connectivity on the basis of hydrocarbon occurrence (a) Schematic block diagram showing a numer-ical model of a normal fault and positions of several stratigraphic intervals at the footwall and hanging wall on the fault surface (b) Sketchdemonstrating the criteria used for identifying whether a fault was open From the occurrence of hydrocarbon nodes A B and C wereclearly open node D remains closed and the state of nodes E and F were not certain during migration Ni is defined as the fault-connectivity index For a given node Ni = 1 indicates the fault has opened for migration Ni = 0 indicates the fault kept closing and Ni = means the effect of the fault cannot be identified and the node will not be taken into account in statistical analysis

and indications of hydrocarbon occurrence maybe constructed A systematic study derived fromavailable seismic sections is therefore performedThe geometrical and structural data are first digi-tized for grid mapping of the main faults An ex-ample of such mapping is shown on Figure 8which presents a perspective view of the footwallof fault ZDH4 The traces of the main horizons(both of the hanging wall and of the footwall) aswell as the traces of seismic cross sections are thebasic elements of this grid Based on the mappingof the oil and gas shows assessing the connectiv-ity of the fault at various nodes of a grid is com-monly possible

Fault-Connectivity Indicator as aProbabilistic Criterion

Sixteen cross-well seismic sections perpendicularto six selected major faults are used to analyze thefault connectivity during migration At each inter-section point between the fault plane and a carrierbed (Figure 7b) a binary index termed the ldquofault-connectivity indexrdquo (Ni) is defined as 0 when evi-dence for migration is absent and 1 when evidencefor migration is present If the evidence is inconclu-sive the corresponding data point is discarded

About 120 data points are compiled each being as-signed with an openclose index

The relevance of this index to the actual open-ing and closing of a fault is debatable First it ispossible that part of the migration occurs in otherpathways instead of the main fault For instancesmall faults or cracks generally accompanying mas-ter faults in extensional basins could channel a sub-stantial fraction of the migrated hydrocarbonsThere also is an implicit assumption (Figure 7) thatat each data point hydrocarbon migrates in a 2-Dplane perpendicular to the fault In fact it is likelythat pathways along the fault are muchmore com-plex than being depicted here and their exact 3-Dgeometry is difficult to assess A statistical approachmay be effective in countering this complexity Ata given point on the fault it is convenient to con-sider the fault-connectivity index as a binary vari-able with a value of 0 or 1 The probability that theindex has a value of 1 depends on the position of thepoint on the fault and is a function of other variablesof the fault such as physical parameters character-izing the environment and history of the faultThe previously collected data allow the character-ization of this probability relationship as follows

For a specified value range of some physicalparameters associated with a fault (eg depth

Figure 8 Structural block diagram of the footwall of fault ZDH4 The positions of stratigraphic intervals are marked as solid lines on thefootwall and as dashed lines on the hanging wall

Zhang et al 1389

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

of fault zone ZDH4 A cross section perpendicularto the main fault (Figure 6a) displays the majorhorizons interpreted from 3-D seismic data togetherwith major reservoirs The lateral extent of the oil-bearing layer (as well as the oil-water contacts) andthe topography at the top of the Es2

2 Es21 and Es1

reservoirs (Figure 6bndashd) emphasize that in mostoccasions faults acted as a seal with respect to thereservoirs However hydrocarbons that accumu-lated in the footwall cannot be from the local sourcerock of Es1 or Es3 because it is poor and immaturein this area Oils must have been fed through open-ings of normal fault ZDH4 (Figure 6a) which ex-tend from the base of Es2

2 in the footwall to thetop of Es2

1 in the hanging wallOn a cross section of the fault plane we iden-

tify particular points (called grid nodes) as the topof each bed in both the hanging wall and the foot-wall (Figure 7a) Assuming that the fault is theonly possible pathway the opening and closingstate of such a node on the fault during the migra-tion period is determined by identifying the hy-drocarbon presence or absence in reservoirs above

1388 Geohorizons

that node on both sides of the fault plane Fourpossible scenarios may occur and are used to iden-tify whether the node had served as a migrationpath The four scenarios can be described as fol-lows (Figure 7b) (1) the reservoirs below containhydrocarbons and the ones above do not (the caseof point D on Figure 7b) (2) reservoirs both aboveand below the node contain hydrocarbons (thecase of point C on Figure 7b) (3) the reservoirsabove the node contain hydrocarbons and thosebelow the node do not (the case of points A andB on Figure 7b) and (4) reservoirs both aboveand below the node do not contain hydrocarbons(the cases of points E and F on Figure 7b) It is notdifficult to assess that in scenarios 2 and 3 the nodemust have been open at some time during migra-tion on the contrary in scenario 1 no hydrocarbonaccumulation above this node exists at present andthus the node was probably closed during migra-tion In the last scenario it is unknown whetherthe node was open or closed during migration

At the scale of the CFSZ many such cross sec-tions which include both structural information

Figure 7 Determination of fault connectivity on the basis of hydrocarbon occurrence (a) Schematic block diagram showing a numer-ical model of a normal fault and positions of several stratigraphic intervals at the footwall and hanging wall on the fault surface (b) Sketchdemonstrating the criteria used for identifying whether a fault was open From the occurrence of hydrocarbon nodes A B and C wereclearly open node D remains closed and the state of nodes E and F were not certain during migration Ni is defined as the fault-connectivity index For a given node Ni = 1 indicates the fault has opened for migration Ni = 0 indicates the fault kept closing and Ni = means the effect of the fault cannot be identified and the node will not be taken into account in statistical analysis

and indications of hydrocarbon occurrence maybe constructed A systematic study derived fromavailable seismic sections is therefore performedThe geometrical and structural data are first digi-tized for grid mapping of the main faults An ex-ample of such mapping is shown on Figure 8which presents a perspective view of the footwallof fault ZDH4 The traces of the main horizons(both of the hanging wall and of the footwall) aswell as the traces of seismic cross sections are thebasic elements of this grid Based on the mappingof the oil and gas shows assessing the connectiv-ity of the fault at various nodes of a grid is com-monly possible

Fault-Connectivity Indicator as aProbabilistic Criterion

Sixteen cross-well seismic sections perpendicularto six selected major faults are used to analyze thefault connectivity during migration At each inter-section point between the fault plane and a carrierbed (Figure 7b) a binary index termed the ldquofault-connectivity indexrdquo (Ni) is defined as 0 when evi-dence for migration is absent and 1 when evidencefor migration is present If the evidence is inconclu-sive the corresponding data point is discarded

About 120 data points are compiled each being as-signed with an openclose index

The relevance of this index to the actual open-ing and closing of a fault is debatable First it ispossible that part of the migration occurs in otherpathways instead of the main fault For instancesmall faults or cracks generally accompanying mas-ter faults in extensional basins could channel a sub-stantial fraction of the migrated hydrocarbonsThere also is an implicit assumption (Figure 7) thatat each data point hydrocarbon migrates in a 2-Dplane perpendicular to the fault In fact it is likelythat pathways along the fault are muchmore com-plex than being depicted here and their exact 3-Dgeometry is difficult to assess A statistical approachmay be effective in countering this complexity Ata given point on the fault it is convenient to con-sider the fault-connectivity index as a binary vari-able with a value of 0 or 1 The probability that theindex has a value of 1 depends on the position of thepoint on the fault and is a function of other variablesof the fault such as physical parameters character-izing the environment and history of the faultThe previously collected data allow the character-ization of this probability relationship as follows

For a specified value range of some physicalparameters associated with a fault (eg depth

Figure 8 Structural block diagram of the footwall of fault ZDH4 The positions of stratigraphic intervals are marked as solid lines on thefootwall and as dashed lines on the hanging wall

Zhang et al 1389

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

and indications of hydrocarbon occurrence maybe constructed A systematic study derived fromavailable seismic sections is therefore performedThe geometrical and structural data are first digi-tized for grid mapping of the main faults An ex-ample of such mapping is shown on Figure 8which presents a perspective view of the footwallof fault ZDH4 The traces of the main horizons(both of the hanging wall and of the footwall) aswell as the traces of seismic cross sections are thebasic elements of this grid Based on the mappingof the oil and gas shows assessing the connectiv-ity of the fault at various nodes of a grid is com-monly possible

Fault-Connectivity Indicator as aProbabilistic Criterion

Sixteen cross-well seismic sections perpendicularto six selected major faults are used to analyze thefault connectivity during migration At each inter-section point between the fault plane and a carrierbed (Figure 7b) a binary index termed the ldquofault-connectivity indexrdquo (Ni) is defined as 0 when evi-dence for migration is absent and 1 when evidencefor migration is present If the evidence is inconclu-sive the corresponding data point is discarded

About 120 data points are compiled each being as-signed with an openclose index

The relevance of this index to the actual open-ing and closing of a fault is debatable First it ispossible that part of the migration occurs in otherpathways instead of the main fault For instancesmall faults or cracks generally accompanying mas-ter faults in extensional basins could channel a sub-stantial fraction of the migrated hydrocarbonsThere also is an implicit assumption (Figure 7) thatat each data point hydrocarbon migrates in a 2-Dplane perpendicular to the fault In fact it is likelythat pathways along the fault are muchmore com-plex than being depicted here and their exact 3-Dgeometry is difficult to assess A statistical approachmay be effective in countering this complexity Ata given point on the fault it is convenient to con-sider the fault-connectivity index as a binary vari-able with a value of 0 or 1 The probability that theindex has a value of 1 depends on the position of thepoint on the fault and is a function of other variablesof the fault such as physical parameters character-izing the environment and history of the faultThe previously collected data allow the character-ization of this probability relationship as follows

For a specified value range of some physicalparameters associated with a fault (eg depth

Figure 8 Structural block diagram of the footwall of fault ZDH4 The positions of stratigraphic intervals are marked as solid lines on thefootwall and as dashed lines on the hanging wall

Zhang et al 1389

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

dip and strike of the fault throw of the fault fluidpressure stress shale smear index etc) the prob-ability that the fault would be open is defined as

Np frac14 n=N eth1THORN

where n is the number of nodes where the fault-connectivity index is equal to 1 andN is the num-ber of nodes where the migration history can beassessed unequivocally (Figure 7)

ASSESSMENT OF PHYSICAL PARAMETERS

Many geologic factors and processes controlwhetherfaults serve as barriers or conduits for hydrocarbonmigration such as fault-plane attitude displace-ment lithology and mechanical properties of off-set strata burial depth juxtaposition of lithologyacross the fault plane fault gouges stress directionand magnitude fluid properties temperature andpressure conditions etc (Bouvier et al 1989Knott1993 Lindsay et al 1993) However a good pre-dictive method in assessing fault connectivity forhydrocarbon migration should use data that areroutinely available (Yielding et al 1997)

In this study burial depth dip and strike of thefault throw of the fault fluid pressure stress andshale smear defined by the SGR are used to char-acterize fault opening and sealing Among theseparameters normal stress and fluid pressure areexpected to have direct impact on the open andclosed behavior of a fault An analysis of a set ofavailable physical data shows that fluid pressureand normal stress are well correlated with mostother parameters except SGR The SGR representsthe sealing properties of a fault which also influ-ence the fluid flow regardless of whether the faultis open or closed (Yielding et al 1997) Thus fluidpressure stress and SGR can be regarded as repre-sentative parameters to characterize the effects offault behavior on hydrocarbon migration

Fluid Pressure in Mudstones

Fault opening during active faulting is expected tobe positively correlated with fluid pressure in mud-

1390 Geohorizons

stones (P) adjacent to the fault This is because themechanical strength of rock strata decreases withincreasing pore-fluid pressure (Hubbert and Rubey1959 Jaeger and Cook 1979) In an interbeddedshale-sandstone package overpressure may serveas a trigger of fault opening (Byerlee 1993 Luo2004) especially when the fluid pressure is muchlarger than the hydrostatic pressure The couplingof faulting and overpressure starts with the accumu-lation of fluid pressure resulting in a decrease of ef-fective stress andmechanical strength of rock stratafracturing and draining fluids out of the sandstones

Once a fault opens the rate of pressure releasewithin mudstones is much smaller than that with-in permeable sandstones (Luo 1999) thereforefaults can easily seal before any significant pressuredecrease occurs in mudstones (Luo et al 2003)During such a short-term opening of a fault the ab-normal pressure in permeable sandstones is re-leased whereas the pressurewithinmudstones re-mains mostly unchanged (Luo 1999 Luo et al2000)

Fluid pressure in mudstones may be estimatedfrom acoustic logs which are available to this studyin only a limited number of boreholes To esti-mate the pressure in shale along the whole faultsystem for a given geological period we used a3-D numerical basin model calibrated against theavailable data The Temis3D software package ofUngerer et al (1990) and Schneider et al (2000)was used Sediment compaction is assumed to bethe main fluid-pressure generation mechanismPresent pressures deduced from acoustic logs in60 wells are reduced following the method ofMagara (1978) They serve as calibration points forthe numerical basin model which allow one toderive the field of fluid pressure throughout thebasin The resulting fluid pressure on the ZDH4fault surface during the deposition of the upper-most Minghuazhen Formation (sim2Ma) is mappedin Figure 9a

Stress Normal to the Fault Plane

The tightness of small fractures within fault zonesmostly depends on the normal stress (d) applied tothe fault plane High normal stress would cause

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

plastic deformation during faulting and closure offractures (Harding and Tuminas 1989) On thecontrary for low normal stress fractures may beopen and serve as conduits to fluid flow (Zhouet al 2000 Lu and Ma 2002)

The normal stress component on a fault planecan be calculated from the position and orienta-tion of the fault plane and the knowledge of thesubsurface stress field (Jaeger and Cook 1979)The magnitude of the effective stress componentnormal to the fault plane is determined by the dipangle of the fault plane burial depth and the di-rection andmagnitude of tectonic stress For listricgrowth faults the dip angle decreases with depthcausing the gravity component of the overburdenperpendicular to the fault plane to increase pro-moting fault closing

The current stress field in this area was mea-sured by Qu et al (1993) and Xu et al (1996)Xu et al (1996) estimated the stress using hydro-fracturing in 50 wells in this area The maximumstress is vertical in the area and the larger horizon-tal stress is in the direction of 55ndash80deg (Qu et al1993) The maximum stress s1 was evaluated tobe equal to the weight of overburden The valuesof the horizontal principal stresses s2 and s3 fol-lowing the burial depth were estimated from themeasurements of Xu et al (1996) The fault-dipangle can be obtained from seismic interpretationThe normal component of stress was mapped onthe surface of fault ZDH4 (Figure 9b)

Shale Gouge Ratio

Clay or shale smeared into the fault zone has beenconsidered as an effective mechanism to seal afault (Weber et al 1978 Bouvier et al 1989Lindsay et al 1993 Gibson 1994 Yielding et al1997 Alexander and Handschy 1998) The poros-ity and permeability of a fault zone decrease with anincreasing amount of smear corresponding to mud-stones incorporated into the fault zone Variousparameters were proposed to estimate the amountof muddy fault smears on the basis of observed dis-tributions of mudstones within fault zones (egBouvier et al 1989 Lindsay et al 1993 Yieldinget al 1997) The SGR of Yielding et al (1997) is

commonly used for heterogeneous siliciclasticstrata (Sorkhabi and Tsuji 2005) It is defined asthe ratio between the throw of the fault and thetotal thickness of mudstones within the throw

SGR frac14Pnifrac141

hi

L 100 eth2THORN

where SGR is in percent L is the throw in metershi is the thickness of the ith mudstone bed and n isthe number of mudstone layers within the throwThe SGR includes therefore two factors that cor-relate with fault seal the lithologies of faultedstrata and the amount of displacement Thus SGRshould be an efficient parameter in evaluating thetendency of the fault to be smeared

Mudstone thickness data come from gamma-log analysis near the fault zones The depth differ-ence between the footwall and hanging wall foreach horizon defines the fault throw for that hori-zon which is used to interpolate the throw at thefault grid points The SGR at every grid node canbe calculated using equation 2 The SGR valuesmay be mapped on the fault surface (Figure 9c)

FAULT-CONNECTIVITY PROBABILITY AS AFUNCTION OF PHYSICAL PARAMETERS

As explained above it is expected at least on a sta-tistical basis that some relationship between fault-connectivity probability can be deduced from hy-drocarbon occurrence on both sides of a fault andthe selected parameters can be deduced indepen-dently from well and seismic data concerning thephysical properties of the fault with respect to itsability to open or close during some events On se-lected fault planes values of physical parameters ateach grid node are estimated and the status of thefault open or closed is evaluated qualitatively ac-cording to the procedure discussed above After re-moving the equivocal data a set of 117 data nodesare finalized

The SGR values are separated into 10 inter-vals For each interval the corresponding fault-connectivity probability (Np in equation 1) is

Zhang et al 1391

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

Figure 9 Distribution of various parameters on the surface of fault ZDH4 projected onto a vertical plane (a) Fluid pressure inmudstones (P) (b) stress normal to fault plane (d) (c) shale gouge ratio (SGR) and (d) fault-connectivity probability on the samefault plane

1392 Geohorizons

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

Figure 9 Continued

Zhang et al 1393

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

computed Figure 10a illustrates a clear negativetrend of Np vs SGR indicating that a smearedfault segment would have less chance to open

In Figure 10b seven intervals of normal stressvalues are used to estimate the relative proportionsof connected occurrence This proportion is againinterpreted as the probability that a fault could beopen for this range of normal stress A clear nega-tive trend of Np vs d is observed indicating thata fault subject to great normal stress would haveless chance to open

In Figure 10c seven intervals of fluid-pressurevalues are used to plot the relative proportions of

1394 Geohorizons

connected occurrence A clear positive trend ofNp vs P is observed indicating that a fault withhigh overpressure has a better chance to be open

As a result of the comparisons presented inFigure 10 the fault-connectivity probability ap-pears to depend significantly on the three inde-pendent physical parameters SGR d and P ona statistical basis The next step is to proceed toa multiparameter approach to improve the assess-ment of the fault-connectivity probability

Fault Opening Index

To combine these three parameters an FOI is de-fined as a dimensionless coefficient

FOI frac14 PdSGR

eth3THORN

The FOI is the ratio between the parameter favor-ing fault opening represented by the fluid pressurein mudstones near the fault and the parameters pro-moting fault closing such as the mudstone smearfactor and the normal stress on the fault plane Thisfunctional dependence is justified by the observedrelationship among individual parameters and bythe fact that the FOI is dimensionless

Fault-Connectivity Probability as a Functionof the Fault Opening Index

The relationship between FOI and fault connec-tivity for fluid flow is empirical through the studyof discovered oil fields The FOI values were cal-culated using data of SGR mudstone fluid pres-sure and normal stress and the correspondingfault-connectivity indices were similarly identi-fied by the statistical method mentioned above(Figure 7)

The 117 FOI values (on independent locationsof faults) are divided into 8 intervals with an incre-ment of 05 The number of total nodes (N) andthe number of nodes that had served as migra-tion pathways (n) are counted for every interval(Figure 11a) Presented as a percentage the valuesof Np = nN within a statistical range define the

Figure 10 Crossplot illustrating the relationship between fault-connectivity probability and (a) shale gouge ratio (SGR) (b) stressnormal to fault plane and (c) fluid pressure in mudstones

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

fault-connectivity probability Our compilation(Figure 11b) indicates that fault-connectivity prob-ability is 0 for FOI values smaller than 075 aquadratic polynomial relationship between FOIand Np is present where FOI ranges from 075 to325 and the fault-connectivity probability is equalto 1 for FOI values larger than 325 The value ofNp = 1 indicates that the node behaved as a migra-tion pathway

The interpolation of these statistical resultsleads to the following relationship

Np frac14001197 ethFOITHORN2 thorn 08931 ethFOITHORN 065641

8lt

FOI 075075 lt FOI lt 325

FOI 325 eth4THORN

Where Np is the fault-connectivity probabilityand FOI is the fault opening index

Characterization of Fault Opening and Closing

Using equation 4Np can be estimated for any gridnode in particular those where no hydrocarbonhas been discovered yet

Taking fault ZDH4 as an example Figure 9dillustrates the distribution of estimated Np on afault plane On this fault connectivity probabilityNp varies over a wide range For example the areawest of theH12well where the Sha-3 (Es3) Sha-2(Es2

2 and Es21) and Sha-1 (Es1) members are pre-

sent has a probability generally greater than 06except in between the XH1 well and XH6 wellwhere the probability is low (lt04) indicating thelow potential of a fault opening The probability isvery low (lt01) in the area of the Sha-1 member(Es1) and Dongying Formation (Ed) on the foot-wall where the fault was likely sealed As a resulthydrocarbons in the carrier beds of the ShahejieFormation (including Es3 Es2

2 Es21 and Es1) in

the hanging wall would not be able to migrateupward into the Neogene reservoirs on the foot-wall along the fault Instead hydrocarbons wouldhavemigrated into reservoirs within the Es2

2 Es21

and Es3 The lack of discoveries in the Neogenereservoirs in this area supports this interpretationIn the area east of the H12 well the fault planeagainst Es3 Es2

2 Es21 and Es1 in the hanging wall

and Es3 to the Guantao Formation(Ng) in the foot-wall both have high (gt05) connectivity probabil-ity indicating fault opening The fault plane mayhave been closed in the area against the lowerMin-ghuazhen Member (Nm2) Hydrocarbon accumu-lations have been discovered in Ng and Nm2 sup-porting the interpretation from fault-connectivityprobability

DISCUSSIONS

Faultmovement is mostly episodic (Hooper 1991Byerlee 1993 Xie et al 2001) stress accumulat-ing over a period would create new faults andorcause themovement of preexisting faults resultingin a sudden release of stress and fluids Faultingwould cease once stress and strain reach equilib-rium and another cycle of stress accumulation

Figure 11 (a) Distribution of the number of total nodes (N)and the number of nodes that served as migration pathways (n)in each statistical range of the fault opening index (FOI) value(b) Relationship between fault-connectivity probability and FOI

Zhang et al 1395

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

would follow (Roberts and Nunn 1995) In gener-al faults are open during faulting and can serve asvertical and lateral hydrocarbon migration path-ways (Hooper 1991 Anderson et al 1994)whereas faults are generally closed during the pe-riod of quiescence and would serve as barriers tohydrocarbon migration (Fowler 1970 Bouvieret al 1989) In fact a fault zone is not entirely openduring faulting instead it is open in some segmentsor places and closed in others (Haney et al 2005)Fault closure during faulting may be caused by sev-eral processes for example ductile strata matchingacross the fault zone sealingby fault gouge smearingofmudstones or enclosure of a segment of an openfault zone by low-permeability strata across thezone and closed fault zones above and below re-sulting in an increase of the length and complexityof pathways for fluid flow (Byerlee 1993 Locknerand Byerlee 1995) When faulting ceases some ofthe open fractures close due to a change in stressfield whereas other open fractures are cementedor welded by precipitates from trapped fluids Thecementation is very fast and can be regarded as in-stantaneous once faulting ceases at a geologic timescale (Ortoleva 1994)

The generation or activation of a fault and thedisplacement of strata across the fault are the cu-mulative results of many tectonic episodes eachof which corresponds likely to some enhancementof earthquake activity In addition the generationand expulsion of hydrocarbons in and from thesource rocks occur over a long geologic period Thecurrently observed hydrocarbon accumulationsthat weremigrated along fault zones are the cumu-lative products of many episodes of fault openingand associated hydrocarbon migration Thus thestudy of fault opening and closing with regard tohydrocarbon migration on the basis of observedor discovered distribution of hydrocarbon accumu-lation should be conducted on the basis of a timescale comparable to the cumulative processes notwith a time scale of individual events of fault open-ing and closing

In addition a fault is in fact a fault zone whichis composed of a series of smaller subsidiary faultsand fractures (Chester and Logan 1986 Chesteret al 1993 Bruhn et al 1994 Caine et al 1996

1396 Geohorizons

Jourde et al 2002) Not all the subsidiary faultsand fractures are open during an episode of faultmovement instead they close or open from timeto time and episode to episode affecting each otherin a complex way As a result fluid flow occursthree dimensionally along the fault zone and itsexact flow path changes from episode to episode(Lockner and Byerlee 1995) The observed hydro-carbon accumulation is the final integrated resultof all episodes of fluid flows along all fault planesfractures pore spaces and permeable rock bodieswithin a fault zone Hence the evaluation of faultopening and closing at a basin scale should be con-ducted for the fault zone as a whole and a faultzone should not be classified as absolutely openor closed regardless of its scale type and othercharacteristics (Luo et al 2003)

In establishing the statistical model of faultopening and closing only the presence and ab-sence of hydrocarbons in reservoirs above and be-low a given grid node along the same verticalplane were considered Fluid flow is in realitythree dimensional and fluid may flow horizon-tally between grid nodes Moreover some openmicrocracks in flexure zones of extensional basinscould divert part of the migrating flow This couldintroduce erroneous interpretation of reservoirleakage or accumulation Another bias could beintroduced by the fact that stresses and fluid pres-sures are obtained from present measurementswhereas migration occurs over long periods dur-ing which these physical parameters could varyTherefore the various inaccurate interpretationscould introduce unpredictable errors to the pro-posed analysis and justify a probabilistic approach

In areas with limited drilling the availabilityof accurate values of various parameters requiredfor calculating FOI and Np is a challenge For ex-ample mudstone content in footwalls and hangingwalls can only be estimated from interpreted de-positional systems However the uncertainties inestimating these parameters would certainly lowerthe accuracy for evaluating FOI and Np

Finally the concept and model of fault-connectivity probability has been developed fora given geological context on the basis of charac-teristics of extensional faults in the sedimentary

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

basins of eastern China The relevance of this mod-el in basins with similar geological settings seemslikely although the parameters of the FOINp rela-tion should be adapted Its applicability for othersettings such as reverse or strike-slip faults wouldrequire further tests

CONCLUSIONS

Opening and closing of fault zones to hydrocarbonmigration are controlled by complex factors andcannot be accurately evaluated on the basis of asingle factor A parameter that incorporates sev-eral factors is needed to better characterize faultopening and closing The cumulative opening orclosing behavior of a point on a fault plane to hy-drocarbonmigration can be established by the pres-ence or absence of hydrocarbons in the reservoirsin the footwall and hanging wall The concept andmodel of fault-connectivity probability proposedin this study incorporate fluid pressure stress nor-mal to the fault plane and SGR and can be used toquantitatively characterize fault opening and closingThemethod is effective in evaluating the history offault opening and closing to hydrocarbonmigrationin the CSFZ of Dagang field Bohai Bay Basin Itimproves the characterization of fault zones as po-tential hydrocarbon migration pathways and is astep further in establishing the complete frame-work of hydrocarbon carrier systems in petrolifer-ous basins

REFERENCES CITED

Alexander L L and J W Handschy 1998 Fluid flow in afaulted reservoir system Fault trap analysis for the Block330 field in Eugene Island south addition offshoreLouisiana AAPG Bulletin v 82 p 387ndash411

Allen U S 1989 Model for hydrocarbon migration andentrapment within faulted structures AAPG Bulletinv 73 p 803ndash811

Anderson R P Flemings S Losh J Austin and RWoodhams1994 Gulf of Mexico growth fault drilled seen as oil gasmigration pathway Oil amp Gas Journal v 92 (June 61994) p 97ndash103

Antonelini M and A Aydin 1994 Effect of faulting onfluid flow in porous sandstones Petro-physical proper-ties AAPG Bulletin v 78 p 355ndash377

Berg R R and A H Alana 1995 Sealing properties ofTertiary growth faults Texas Gulf Coast AAPG Bulle-tin v 79 p 375ndash393

Bouvier J D C H Kaars-Sijpesteijn D F Kluesner C COnyejekwe andRCVanderPal 1989Three-dimensionalseismic interpretation and fault sealing investigationsNun River field Nigeria AAPG Bulletin v 73 p 1397ndash1414

Bruhn R L W T Parry W A Yonkee and T Thompson1994 Fracturing and hydrothermal alteration in normalfault zones Pure and Applied Geophysics v 142 p 609ndash644 doi101007BF00876057

Byerlee J D 1993 Model for episodic flow of high pressurewater in fault zones before earthquakes Geology v 21p 303ndash306 doi1011300091-7613(1993)021lt0303MFEFOHgt23CO2

Caine J S J P Evans and C B Forster 1996 Fault zonearchitecture and permeability structure Geology v 24p 1025ndash1028 doi1011300091-7613(1996)024lt1025FZAAPSgt23CO2

Chester F M and J M Logan 1986 Implications for me-chanical properties of brittle faults from observationsof the Punchbowl fault zone California Pure and Ap-plied Geophysics v 124 p 79ndash106 doi101007BF00875720

Chester F M J P Evans and R L Biegel 1993 Internalstructure and weakeningmechanisms of the SanAndreasfault Journal of Geophysical Research v 98 p 771ndash786 doi10102992JB01866

Childs C A Nicola J J Walsha and J Wattersonc 2003The growth and propagation of synsedimentary faultsJournal of Structural Geology v 25 p 633ndash648 doi101016S0191-8141(02)00054-8

Clausen J A RHGabrielsen E Johnsen and J A Korstgaringrd2003 Fault architecture and clay smear distribution Ex-amples from field studies and drained ring-shear experi-ments Norwegian Journal of Geology v 83 p 131ndash146

Engelder J T 1979 Cataclasis and the generation of faultgouge Geological Society of American Bulletin v 85p 1515ndash1522 doi1011300016-7606(1974)85lt1515CATGOFgt20CO2

Forster C B and J P Evans 1991 Hydrogeology of thrustfaults and crystalline thrust sheets Results of combinedfield and modeling studies Geophysical Research Let-ters v 18 p 979ndash982 doi10102991GL00950

FowlerWA J 1970 Pressures hydrocarbon accumulationand salinitiesmdashChocolate Bayou field Brazoria CountyTexas Journal of Petroleum Technology v 22 p 411ndash423

Gibson R G 1994 Fault-zone seals in siliclastic strata of theColumbus Basin offshore Trinidad AAPG Bulletinv 78 p 1372ndash1385

Gratier J P P Favreau F Renard and E Pili 2002 Fluidpressure evolution during the earthquake cycle controlledby fluid flow and pressure solution crack sealing EarthPlanets Space v 54 p 1139ndash1146

Haney M H R Snieder J Sheiman and S Losh 2005 Amoving fluid pulse in a fault zone Nature v 437 p 46doi101038437046a

Zhang et al 1397

Harding T P and A C Tuminas 1989 Structural interpre-tation of hydrocarbon traps sealed by basement normalfault blocks at stable flank of fore-deep basins and at riftbasins AAPG Bulletin v 7 p 812ndash840

Hasegawa S R Sorkhabi S Iwanaga N Sakuyama andO A Mahmud 2005 Fault-seal analysis in the Temanafield offshore Sarawak and Malaysia in R Sorkhabi andY Tsuji eds Faults fluid flow and petroleum trapsAAPG Memoir 85 p 43ndash58

Hooper E C D 1991 Fluid migration along growth faultsin compacting sediments Journal of Petroleum Geol-ogy v 14 p 161ndash180 doi101111j1747-54571991tb00360x

Hubbert M K 1953 Entrapment of petroleum under hydro-dynamic conditions AAPG Bulletin v 37 p 1954ndash2026

Hubbert M K and W W Rubey 1959 Role of fluid pres-sure in mechanics of overthrust faulting Bulletin of theGeological Society of America v 70 p 115ndash205 doi1011300016-7606(1959)70[115ROFPIM]20CO2

Jaeger J C and G Cook 1979 Fundamentals of rock me-chanics London Chapman and Hall 593 p

Jourde H E A Flodin A Aydin L J Durlofski and X HWen 2002 Computing permeability of fault zones ineolian sandstone from outcrop measurements AAPGBulletin v 86 p 1187ndash1200

Kachi T H Yamada K Yasuhara M Fujimoto S HasegawaS Iwanaga and R Sorkhabi 2005 Fault-seal analysisapplied to the Erawan gas-condensate field in the Gulfof Thailand in R Sorkhabi and Y Tsuji eds Faults fluidflow and petroleum traps AAPG Memoir 85 p 59ndash78

Karlsen D A and J E Skeie 2006 Petroleum migrationfaults and overpressure Calibrating basin modelingusing petroleum in trapsmdashA review Journal of PetroleumGeology v 29 p 227ndash256 doi101111j1747-5457200600227x

Knipe R J 1989 Deformation mechanismsmdashRecognitionfrom natural tectonites Journal of Structural Geologyv 11 p 127ndash146 doi1010160191-8141(89)90039-4

Knipe R J 1992 Faulting processes and fault seal inR M Larsen H Brekke B T Larsen and E Talleraseds Structural and tectonic modeling and its applica-tion to petroleum geology Amsterdam Elsevier p 325ndash342

Knipe R J 1997 Juxtaposition and seal diagrams to helpanalyze fault seals in hydrocarbon reservoirs AAPG Bul-letin v 81 p 187ndash195 doi1013063B05B376-172A-11D7-8645000102C1865D

Knott S D 1993 Fault seal analysis in theNorth Sea AAPGBulletin v 77 p 778ndash792

Li S T F X Lu and C S Lin 1998 Evolution of Mesozoicand Cenozoic basins in east China and their geodynamicbackground (in Chinese) Wuhan China University ofGeosciences Press 238 p

Lindsay N G F C Murphy J J Walsh and J Watterson1993 Outcrop studies of shale smears on fault surfacesInternational Association of Sedimentologists SpecialPublication 15 p 113ndash123

Lockner D and J Byerlee 1995 An earthquake instabilitymodel based on faults containing high fluid-pressure

1398 Geohorizons

compartments Pure and Applied Geophysics v 145p 718ndash745

Lu Y F and F J Ma 2002 Controlling factors and classi-fication of fault seal (in Chinese) Journal of Jilin Uni-versity (Earth Science Edition) v 33 p 163ndash167

Luo X R 1999 Mathematical modeling of temperature-pressure transient variation in opening fractures and sed-imentary formations (in Chinese) Oil and Gas Geologyv 20 p 1ndash6

Luo X R 2004 Quantitative analysis on overpressuringmechanism resulted from tectonic stress (in Chinese)Chinese Journal of Geophysics v 47 p 484ndash493

Luo X R J H Yang and Z F Wang 2000 The over-pressuring mechanisms in aquifers and pressure pre-diction in basins (in Chinese) Geological Reviewv 46 p 6ndash10

Luo X R W L Dong J H Yang and W Yang 2003Overpressuring mechanisms in the Yinggehai BasinSouth China Sea AAPG Bulletin v 87 p 629ndash645doi10130610170201045

Magara K 1978 Compaction and fluid migration practicalpetroleum geology Amsterdam Elsevier 319 p

Ortoleva P J 1994 Basin compartments and seals AAPGMemoir 61 p 1ndash477

Perkins H 1961 Fault closure-type fields southeast Louisi-ana Gulf Coast Association Geological Society Transac-tions v 11 p 203ndash212

Qu G S H Zhang and H Ye 1993 The principal stressorientations by the borehole breakouts in Huanghua de-pression (in Chinese) North China Earthquake Sciencesv 11 p 9ndash18

Roberts S J and J A Nunn 1995 Episodic fluid expulsionfrom geopressured sediments Marine and PetroleumGeology v 12 p 195ndash202 doi1010160264-8172(95)92839-O

Schneider F S Wolf I Faille and D Pot 2000 A 3-D basinmodel for hydrocarbon potential evaluation Applicationto Congo offshore Oil and Gas Science and Technology-Review Institut Francais du Peacutetrole v 55 p 3ndash13doi102516ogst2000001

Selley R C 1998 Elements of petroleum geology NewYork Academic Press 470 p

Sibson R H 1981 Fluid flow accompanying faulting Fieldevidence andmodels inDW Simpson and PG Richardseds Earthquake predictionmdashAn international reviewAmerican Geophysical Union Monograph MauriceEwing Series 4 p 593ndash603

Sibson R H J Mc M Moore and A H Rankin 1975 Seis-mic pumpingmdashA hydrothermal fluid transport mecha-nism Journal of the Geological Society v 131 p 653ndash659 doi101144gsjgs13160653

Smith D A 1966 Theoretical considerations of sealing andnon-sealing faults AAPG Bulletin v 50 p 363ndash374

Smith L C B Forster and J P Evans 1990 Interaction offault zones fluid flow and heat transfer at the basinscale in S P Neuman and I Neretnieks eds Hydro-geology of low permeability environments Hydrogeol-ogy Selected Papers 2 p 41ndash67

Sorkhabi R and Y Tsuji 2005 The place of faults in pe-troleum traps in R Sorkhabi and Y Tsuji eds Faults

fluid flow and petroleum traps AAPG Memoir 85p 1ndash31

Ungerer P J Burrus B Doligez Y Chenet and F Bessis1990 Basin evaluation by integrated two-dimensionalmodeling of heat transfer fluid flow hydrocarbon gen-eration and migration AAPG Bulletin v 74 p 309ndash335

Vasseur G and L Demongodin 1995 Convective heattransfer in sedimentary basins Basin Research v 7p 67ndash79 doi101111j1365-21171995tb00096x

Wang G Q J F Qi and Y F Yue 2003 Formation andevolution of the Cenozoic tectonics within and surround-ing the Qikou sag (in Chinese) Chinese Journal of Geol-ogy v 38 p 230ndash240

Wang G Z S Q Yuan and L Xiao 2006 Source condi-tions in Chengbei fault-step area and analysis on hydro-carbon migration (in Chinese) Journal of Oil and GasTechnology v 28 p 200ndash203

Weber K J GMandl F Pilaar F Lehner and R G Precious1978 The role of faults in hydrocarbon migration andtrapping in Nigerian growth fault structures 10th An-nual Offshore Technology Conference Proceeding v 4p 2643ndash2653

Xie X S T Li W L Dong and Z L Hu 2001 Evidencefor hot fluid flow along faults near diapiric structure of theYinggehai Basin South China Sea Marine and PetroleumGeology v 8 p 715ndash728 doi101016S0264-8172(01)00024-1

Xu JW 1980 Lateral shifting of the Tanlu fault and its geo-logical significance Proceedings of the Symposium ofInternational Exchanges for Earth Sciences p 129ndash142

Xu Z H X Wu S F Qi and D N Zhang 1996 Charac-teristics of contemporary tectonic stress in Beidagang oilfield and its vicinity (in Chinese) Petroleum Explorationand Development v 23 p 78ndash81

Yang Y and T Xu 2004 Hydrocarbon habitat of the off-shore Bohai Basin ChinaMarine and PetroleumGeologyv 21 p 691ndash708 doi101016jmarpetgeo200403008

Yielding G B Freeman and D T Needham 1997 Quantita-tive fault seal prediction AAPG Bulletin v 81 p 897ndash917

Yielding G J A Oslashverland and G Byberg 1999 Charac-terization of fault zones for reservoir modeling An exam-ple from the Gullfaks field northern North Sea AAPGBulletin v 83 p 925ndash951

Yu C H J Q Su S Q Yuan D J Sheng R M Duan andL K Zhang 2006 Analysis of main control factors onoil and gas accumulation in Chengbei fault-step areaDagang oil field (in Chinese) Sedimentary Geologyand Tethyan Geology v 26 p 88ndash92

Yuan S Q X L Ding J Q Su and S F Zhang 2004 Theresearch on forming conditions of reservoir in Chengbeistep-fault zone Dagang oil field (in Chinese) Journal ofJianghan Petroleum Institute v 26 p 8ndash9

Zhang L K X R Luo Q J Liao J Q Yuan D J ShengD Q Xiao Z M Wang and C H Yu 2007 Quanti-tative evaluation of fault sealing property with fault con-nectivity probabilistic method (in Chinese) Oil and GasGeology v 28 p 181ndash191

Zhou X G B S Sun C X Tan H B Sun R Z Zhengand C X Ma 2000 State of current geo-stress and ef-fect of fault sealing (in Chinese) Petroleum Explorationand Development v 27 p 127ndash131

Zhang et al 1399