. ●

Sodety ofPetroleumEfl@lom I

SFE 26937Perhxmance Evaluation of Waterflood Project in SouthernWest Virginia$hahab Mohaghegh,● Samuel Ameri,* Kha~hayarAminian,● and Ujjal Chatterjee,West Virginia U.

●SPE Members

COPYWIt19S3, Sod@ of Petroteum Engineere, Inc.

This paper waa prepared for preaerttation a! the 19S3 Eeetern Regional Conference 6 Exhibition he”r’ in Pittsburgh, PA, U. S.A., 2-4 Novamber 1SS3.

This paper wee wlocted for preaentdon by ●n SPE Program Committea fotbwirq review of intormatbn contained in an ebatrect eubmitted by the author(e). Contents 01 the pecw,as preamod, hew not been rdewed by mo socii d PeMeum EIWWW9 m .M a~- to CONOCI~ by t~ ●tiwa). T~ met~ial, w w-nt~, do’= @ n~-ily ref~t

MY POCMIO+Iof the 9CCMY Of Pawdaum EIIIWI- M -, of ITW*. p- rN-WCJ III spE -iww W* a- to @l~t~ rw~ by Edltofiel *mffl*a of the ~evof Petmhwm Err@neem. Permii to copy b metdded roan tired of not more then M we & Iliuetretiom may nd be copied. The ebatrect ehoutd contain wnepicuou$acknowtedgrnentof where end by whm ttro paper b preeanmd. Writs Librarian, SPE, P.O. Sox S3S838, Rkhdrdeon, T,! 7YM3-3KM, U.S.A., Talex 1S3245 SPEUT.

ABSTRACT

Granny Creek oil field is located inSouthern West Virginia in the AppalachianBasin. R has been producing from the Big Injunformation. Primary production was initiated inthe 1920’s and continued until early 1970’s.Around the mid 70’s, new wet:: were drilled anda waterflood project was started which is still inprogress. During the course of the waterfloodproject, some problems were encountered.These problems included, inconsistency insweep efficiency in adjacent patterns, and highinjection pressure.

In this field often two adjacent five spotpatterns with similar injection to por~ volumeratios exhibit totally different oil ai~d waterproduction rates, While water does notbreakthrough in some patterns for months,other patterns experience instantaneous waterbreakthrough. Unusually high injection pressureis obsewed almost throughout the field, and yetsome injection wells exhibit normal injectionpressures,references and Illustrations at Ihe end of paper,

Presented in this paper is a summary ofapproaches, methodologies, results andconclusions that have been reached during theperformance evaluation of this waterfloodproject. With the aid of resei :oir simulationstudies, some major heterogeneities werect~aracterized and modeled for this field.Interpretations of seismic studies were used toconfirm the orientation of such heterogeneitiesin the field.

BACKGROUND

Figure 1 shows the approximate locationof Granny Creek oil field in Southern WestVirginia. This field is producing from Big Injunformation, which is about 1800-2000 feet deep.Big Injun sandstone in this field has a netthickness of about 35 to 45 feet and is cappedby Big Lime, a gas bearing limestone formation.Big Injun sand, a Mississippian age formation,has been divided into three layersl’2, namely A,B, and C. These subdivisions of Big Injuncouesponds to grain-size as well as densityvariations.

429

.

2 PERFORMANCE EVALUATION OF A WATERFLOOD PROJECT IN SOUTHERN WEST VIRGINIA SPE26937

Layer A is the shallowest of thethree and is characterized by coarse grain-

size and low density values, with an averagethickness of 5 to 15 feet. Layer B, the midsection, is also coarse grained which exhibitshigh density anomalies. This layer is furthercharacterized by its small thickness (5-1O feet)and low porosity and permeability values. Themost productive layer is C, which is the deepestof the three, and is the thickest (20-35feet). This section is fine-grain and ischaracterized by low density sandstone. It hasbetter porosity and permeability definition whencompared to the other two. Big Injun has alsobeen divided into other subdivisions on thebasis of Depositional Environment andLithofacies. The typical division of Big Injunsand are presented in Figure 2.

Primary production in Granny Creekstarted in the early 1920’s. It was not until thelate 1970’s that waterflooding was initiated.During the course of the waterflooding operationtwo major phenomena was observed: (a)encountering high injection pressure at theinjection wells and, (b) inconsistent sweepefficiency or relatively low secondary oilrecove~ from some five spot patterns. As aresult, attempts have been made to addressthese problems, and provide possibleexplanations for their c~ccurrence. Granny Creekis being operated by two separate companies,One is developing the notihern part of the fieldwhile the other operates the southern part. Theaforementioned problems (namely, highinjection pressure and low secondaryproduction) has been observed and reported byboth companies,

In this paper, two adjacent five spotpatterns have been chosen for study. Thesepatterns are located in the southern part of theGranny Creek and have comparable pore

I 430

volumes. The behavior of these two patternsare typical of the entire field. These twopatterns involve six injection wells and twoproduction wells. The injection wells in thisstudy are referred to as 1-1, 1-2, .... I-6 and theproduction wells are referred to as P1 and P2,co~esponding to the first and second patterns.Injection wells 1-1,1-2,1-4, and I-5 belong to thefirst pattern while injection wells 1-2,1-3,1-5, andI-6 are part of the second pattern. Figure 3 is aschematic diagram of these two adjacent fivespots,

SOURCES OF DATA

Geophysical well logs are available frommany wells in Granny Creek. More than 20wells were cored during the course of primaryproduction and core analysis data are available(the injection and production wells of the twoadjacent five spots that are being studied hereare not among the cored wells, although four ofthe cored wells are quite close to the wellsunder investigation). Most of the data, such asporosity and formation thickness, have beenderived from well logs while permeability datawas measured from cores. Using conventionalgeostatistical methods, these data wereextrapolated and mapped to the entire field.Fluid properties (both oil and water) and relativepermeability data were obtained from theoperating companies. These companies alsoprovided the injection and production rates aswell as injection pressure data. For most of thesouthern part of the field, !!w detailed primaryproduction data were not available in a formthat could be used for modeling and simulationstudies.

SPE26937 MOHAGiiEGH, S., AMERI, S., AMINIAN, K., ANO CHATTERJEE, U. 3

INJECTION PRESSURE

Figure 4 shows the injection pressuredata in the six injection wells in the twoadjacent five spot patterns that are beingstudied. Itshould benoted thatwhilefive outofthe six injection wells have pressures in thesame order of magnitude, well I-4demonstrates a much lower pressure. Thepressure in I-4 is almost half as much as theother five injection wells. It must be also notedthat given the reservoir pressure, injection rate,formation thickness, well completion information(mostly open hole), formation permeability,viscosity and compressibility of the injectingfluid (water), a theoretical (approximate)injection pressure was calculated. This valuewas found to be less than half of what isobsewed in the field.

During the course of the simulation,although injection and production rates werematched, the injection pressure that wascalculated by the simulator was again muchless than what was repoded from the field.

Since flow of fluids in porous media isgoverned by the permeability of the formations,once rate and pressure gradient were adjusted(matched with the field data), the integrity ofpermeability values became questionable. Thevalues that were used in the simulation studywere those measured from the core. Severalcore samples from the field were obtained andwere independently measured in the labs.These samples were taken during the primatyphase of production and therefore were notaffected by injected water. A detailed samplingof cores and their analysis revealed that thevalues of permeability used in the simulationstudy were quite reasonable,

The next step was the analysis of twoseparate fall-off tests that were performed ontwo wells in the southern pad of the field.Figure 5 shows the location of these two wells.Figures 6 and 7 illustrate the graphicalrepresentation of these two tests. These wellswere stimulated during the injection process.The result of the analysis showed a negativeskin which is explained by the stimulationprocess. The interesting finding from both fall-off tests was that the calculated permeabilitieswere at least one order of magnitude less thanthose measured from the cores. Consequently,two points needs to be addressed. First, thecalculated permeability from these testsrepresent an average values for the radius ofinvestigation of the test which was determinedto be about 300-400 feel. Thus, th ~permeability at the vicinity of the wellbore islower than the calculation revea%. However,the permeability of areas farther away from thewellbore are higher. Second point is that it maybe questionable to compare the calculatedpermeability values for these wells to measuredpermeabilities of cores from other wells. Themagnitude of the permeability in differentlocations in the field reveals that , although thepermeability changes from place to place, itranges between 5 to 14 md.

The permeabilities calculated frwv fall-offtests were about 0.3 to 0.4 md. Since the coreswere taken prior to the water injection, theconclusion is that introduction of water into thesystem is causing the damage or permeabilityreduction. To further test this theory, thepermeabilities of the blocks around the injectionwells were reduced and simulation was earnedout. The results sh~wed that by modeling alocalized low permeability area around theinje~lion wells, the injection pressures observed

I

in the field could be simulated.

431

.

4 PERFORMANCE EVALUATION OF A WATERFLOOD PROJECT IN SOUTHERN WEST VIRGINIA SPE26937

SWEEP EFFICIENCY

A puzzling probiem that appeared duringthe waterflooding process, was theinconsistency of sweep in adjacent patterns.The two adjacent patterns that have been usedin this study are examples that typifies suchinconsistencies in the field. These two patternshave similar injection rate to pore volume ratios.There is no reason to believe that one part ofthe reservoir is depleted to a higher degreeduring the primary production. Figure 8illustrates the reported oil production in thesepatterns vlhile Figure 9 shows the reportedwater production. The inconsistency of the oiland water production in these two patterns isquite notable.

It has been a well known fact that Biglnjun sandstone is a highly heterogeneousformation. Geological studies112have indicatedthat this formation as well as its adjacenthorizons are fractured and/or contain manyfaults. SRismic studies have shown theexistence of such faults and fractures. Anotherpossibility is that, since the injection pressuresare so high, the formation may have beenfractured during the injection process, Mhoughthe operators strongly disagree with this theory.Looking at the injection pressure behavior, onemight conclude that, the pressure data does notshow typical behavior of a damaged/fracturedwell (unless the formation damage took placeearly during the injection process), since theinjection pressure bu!lds up and remains steadyat a high level. This supports the operatorsview regarding the fracture initiation during theinjection process. It could be concluded thatsome fractures might be present in the field,The irnpor%nt task was to see if the existenceof a fracture, whether the fracture(s) have beengenerated during the injection process or arenative to the formation, could explain the

behavior of these five snot patterns.

The results of tracer tests performed byboth operators further proved that somecommunication exists between different wells.In addition the seismic interpretation suggestedthat a fracture could possibly exist in that are.Having this information, an obvious fact thatwas being overlooked became more visible.This was the pressure data of the injection wells(Figure 4), Looking at this graph, it is obviousthat well I-4 has the lowest injection pressure.Therefore, it is most probable that, if a fractureexists between production well P2 and oneinjection well, it must be 1-4. This orientationwas much closer to the interpretation of seismicdata.

The next step was to decide whether thefracture is actually in Big Injun or in anotherformation, which is acting as a thief zone sinceseismic data was unable to detect the exactdepth of the fracture.

Several simulation studies wereconducted and the final conclusion was that theexistence of the fracture in the Big Lime, whichis the adjacent horizon on the immediate aboveof the Big Injun, is the most probable scenario,The results of the simulation study in the fwrnof a history matching is presented in Figures 10and 11 for both five spot patterns. In this studythe Big Injun and the Big Lime formations aremodeled as four separate layers. The top-most layer is the Big Lime while layers 2, 3,and 4 are subdivisions of Big Injun, layersA, B, and C respectively. Both injection andproduction wells are completed as openhales. The fracture runs from injection well I-4to production well P2. The fracture exists onlyin the Big Lime formation.

II

432

SPE26937 MOHAGHE13H, S., AMERI, S., AMINIAN, K., ANO CHA7TERJEE, U 5

In order to simulate the effect of thefractures in the reservoir, the transmissibilitymodi?cation method was used.3 This methodhas been tested and verified for modelingpermeability interfaces in black oil reservoirmodels.

the oil recove~ to decrease considerably. Itwas also concluded that in this field the iso;atedfractures are most probably located in the BigLime, which is above Big Injun sand, acting asa thief zone,

REFERENCES

CONCLUSIONS

A waterflooding project in Granny Creekfield in Southern West Virginia was evaluated.This recovery process has displayed someunusual behavior. These behaviors weremainly high injection pressures and inconsistentoil recove~ from similar five spot patterns. Itwas concluded from engineering calculations,simulation studies, analysis of injection fall-offtest data, and Iaboratoty measurement of thecores that injection of water has adverselyaffected the permeability of the fcrmation. Thismay have caused the hydri%.dicfracturing duringthe injection process. It was also concludedthat the inconsistency of secondary oil recove~is a function of isolated fractures that may bepresent.

These permeability interfaces, whenoriented parallel to the streamlines can cause

a

1. Donaldson, A,, et al.: “Measuring andPredicting Reservoir Heterogeneity in ComplexDeposystems’’,Annual Repod, DOE/BC/14657-7, Aug. 1992.

2. Donaldson, A., et al.: “Measuring andPredicting Resewoir Heterogeneity in ComplexDeposystems’’, Anrwal Report, DOE/BC/14657-7, Aug. 1993.

3. Nlohaghegh, S., Aminian, K., Ameri, S.,and Chatterjee, U.: “ A New Approach forModelling and Simulation of HydrocarbonResemoirs Containing Isolated Channels andBarriers”, paper No. 203-052, The InternationalAssociation of Science and Technology forDevelopment (IASTED), internationalConference on Applied Modelling andSimulation, July 21-23, 1993, Vancouver,Canada,

a,.

1 0! .. %,.. ‘ .,. I

,,,: :T .,,: .,,;,

ZONAL EN VIRONAfENT LJTIIOFX CIE

Figura 1. Granny Creek field in southern West Virginia, Figu~ 2. Geological subdivisionsof Big Injun formation,

I

WIA 1.1 + M .*- 1.3 + I-4 . . .. 1.s — 14

M I-5 1-6800. Y

0“ 500 iooo 1500 2000

Fgure 3. Two five spot waterflood patt..ms being studied,

2500

17me (days)

I

sPQ69372000

!.* -& 1

1-2.-

1-1a I

‘“- II If

/ \\

I

INor!hernPort ‘

\

\

I~,y,,,,,,: 1I

oI

1

1 Figure 5. Location of two wells with injection fall-off test data.

Figure 4. Injection pressure data for the two five spot patterns.

well # 2055

2000

u T

9&e=33 1.6

~~ P.,’

. . ,.~ , ~oo ‘. ’’:’,,

E!E!=i

g

i 100 0 @ia :’:. ~

I .S=.-4,2 .I

. . . . . . . .:. ” I500

10’ 10’ 10’ 10’ 10’ 10’ ‘0’ ‘0’h++

Figure 6. Fall-off test data for well # 2055.

WC1l# 21975

s =-5.5\

\

. 10’ 10’ @ 10’ I@’10’*

Figure 7. Fall-off test data for well # 21975.

434

‘sPE26937

2.27x104[ 1 1 1 1M

-[

0 5- Spol P2

x 5 - spot PI

~ 1,82X1042

co.-

1 1.36x104 ./1eoII

, *OXIO1*

30 517 1004 1490 1977 2464

Time (days)

2.22xlo4”

~~ 1.78x104 -

c~YJ

: 1,33X104 -

<

g

g 8.90X103 -

al:0‘iE 4.45X103 -

‘3

o30 517 1004 1490 1977 2464

Time (days)

Figure 8. Secondary oil production from two adjacent patterns. Figure 9. Seconda~ water productionfrom two adjacent pattern:

2.27x104[ 1 I 1 1‘a

4~ 1.82X1O

~

=

> 1.36x104 -:

:>

~

i!~:

: 4.54X103 -

01 501 1001 1500 2000 2500

Time (days)

Figure10. Simulation match for oil and water production forfive spot pattern #1, including the Lsolatednatural fracture.

2.22X104

tI,78x104

t

~ Simulated water

~2

Fkld Dst- (oil)Pield Dais (water) ,

n t

~ 8.90x10J.=“ t

E

: 4.45X103 k 4!2\\.Gi#i

OL1 475 949 1422 1896 2370

Time (days)

Figure 11, Simulation match for oil and water production forfive spot pattern #2, including the isolated natural fracture,

I 435