8/9/2019 Spe 07921

1/10

SPE79

SPE

of~

3@_@mac AlME

GASOCCURRENCENTHEDEVONI ANHALE

by E. C. Smth, S. P. Cremean, and G. Kozai r,

Col umbi a Gas SystemServi ce Corp.

Copyri ht 1979, American Institute of Mining, Metallurgical and Petroleum Engirv?ers, Inc.

f

This paper was presented at the 1979 SPE Symposium on Low

rmeabi it y Gas Reservoirs, B y 20-22, 1979. Denvar, Colorado.

The material le subject to correction by the author.

Permission to copy is restricted

an abstract of not more than 300 words,

Write; 6200 N. Centra~ Expy., Oallas, Texaa 75206.

\

ABSTRACT what f racti oncan be producedeconomcal l yw thin

a reasonabl et i me fr ame. Si nce the shal e has a

Esti matesof recoverabl ereserves i n the

very l owpermeabi l i ty, economc producti onrequi re

eastern Devoni anshal es range f roma fewt ri l l i on

a fr acturenetwork to provi de suf fi ci ent fl ow

cubi c f eet to several hundred tri l l i oncubi c feet.

except whece the secti cnsontai ns permeabl e si l t-

The most pessimsti cesti mates assume that al l

stones or sandstones. Gas producti on rates w 11

recoverabl egas occurs w thi n natural f racture

depend upon the s ze, spaci ngand geometry of thes

porosi ty. More opti msti c esti mates assume a

f ractures.

substanti al gas contr i buti onf romthe shal ematr i x

close to fractures. Thi s paper presents the avai l -

The extent of these control s i s the subj ect of

abl e evi dence for both vi ews i ncl udi ngl ong- term

a l i vel y debate whi ch focuses on the dynamcs of

shal e producti oncharacteri sti cs, l aboratory

shal e gas producti on. Two basi c school s of though

measurements on shal e cores, and numeri cal model i ng exiW

a) the bal k of the recoverabl egas deri ves

studi es.”The wei ght of al l avai l abl eevi dence

f romthe matr i x, but f ractures are necessary to

favors the posi t i onthat the mat ri x provides a providerapid f l owto the wel1, and b) the bulk of

maj or contri buti onof the recoverabl egas fromthe the recoverabl egas i s contai nedw thi n natural

Devoni an shal es, w th a smal l er amount deri ved f rom

fr actures i n the shal e secti onand the contri buti o

the f racture voi d vol ume. fromthe matr ix over the l i fe of the wel l i s uni m

portant.

I NTRODUCTI ON

Thi s debate on f racture vs. matr i x producti on

For several years, the Col umbi aGas System

began at l east 45years ago. The proceedi ngs@ a

Servi ce Corporati onhas shown an i ncreasi ngi nterest

1935 Devoni anshal e symposi ummenti onedboth con-

i n promoti ngDevoni anshal e researchand devel opment

‘ 2‘ g) Each concept cl ai ms the support of pr

i n the Appal achi anBasi n. A systemati cexpl oitati o

n

cepts.

programwoul d hi nge upon the dri l l i ngofa l arge

ducti on data and hi story matchi f i gof producti on

number of test wel l s i n key parts of the Basi n.

decl i ne curves. Both si des can account for the

Thi s, i n turn, requi res pri or demonstrati onof the

total gas producti on (usi ngappropri ate assumpti on

basi n’ s potenti al gas suppl y. Our researchcan-

and i nput data).

Both have the support ofnumeri -

ti nues to support estimates of 200 to 900 tri l l i on

cal model s.

cubi c f eet of potenti al gas suppl i es based on pro-

ducti on records, core gas content measurements,

Col umbi a’ s i nterest i n these producti onmodel

wel l l ogs and wel l cutt i ngs descri pti ons.

stems f romthei r i nf l uenceon reserve esti mati ons.

We must resol vewhether matri x gas can be recovere

before usi ng the matri x gas content as a basi s f or

suppl y proj ecti ons.

Thi s paper revi ews the assump

Several reports publ i shedby the Of f i ce of

ti ons, i nput data, and cal cul ati onsused i n each

Technol ogyAssessment, Federal Power Cormni ssi on,

approach.

Al thoughwe may not def i ni tel y prove

Federal Energy Regul atory Cormni ssi on,TRWand Lew n

whi ch i s correct unti l after a w despreaddri l l i ng

and Associ ates on the Appalachi anDevoni an shal e

program we can compare the l i kel i hoodof each con

provi de other resource esti mates usi ng di f f erent

cept. Compari ngthe two arguments may hel p us

assumpti ons.

One very comprehensi vereport by the

narrowthe range of val ues for the assumed i nput

Nati onal Petr ol eumCounci l shoul d he i ssued i n the

data and gai n some i nsi ght i nto f racture si ze and

near future.

spaci ng. These factors shoul d prove useful i n

- -

/ “

determni ngrecovery factors, wel l spaci ng and

Al thoughwe knowthat the shal e matr i x contai ns

f racture desi gns.

a l arge vol ume of mobi l e gas, the questi on remai ns

99==

8/9/2019 Spe 07921

2/10

?“

HI STORICAL I NFORMATIONON SHALE GAS PRODUCTION

Devoni anshal e expl oi tat i onhas endured si nce

1820 when the fi rst gas wel l was dri l l ed i n

Fredoni a, NewYork. By 1890, numerous very shal l ow

wel l s were dri l l ed f romDunki rk, NewYork to

‘ Sandusky, Ohi o.

Duri ng thi s century, mo j tshal e

product i oncame f romeastern Kentucky, western

West Vi rgi ni aand southernOhi o. To date, devel op-

ment has concentr atedon onl y 4, 000 of the over

110, 000square ml es of the Appal achi anBasi n under

l ai n by t l e shal e.

I n . he earl i er years, very shal l owwel l s w th

pressurss barel y above atmospheri c provi dedan

easy . S{J Urc( ?f smal l vol umes of gas for l ocal use.

From1927- 1962, th[ economc cl i mate encouraged

sha?e gas expl orati onas can be seen f romthe

number of compl etedwel l s by Col umbi a (See

Fi gure 1) .

However, the i ndustry typi cal l y

expl oredf or other targets, and i n some cases sti -

mul ated the shal e i f the other targets fai l ed to

produce. We knowof l arge shal e gas shows whi ch

i nterrupteddri l l i ng, but whi ch were not expl oi ted.

Product i onrecords(l G) showthat 40%of the

Shal e wel l s had no measurabl enatural showand

onl y 5%had shows of a commerci al si ze (Fi gure 2).

Most gas shows occur w thi n the bl ack, organi c-

ri ch sectionwhi ch contai ns an avera e of si x ti ms

7

?s much matri x gas as the remai ni ng gray) shal es 6

Af ter tr eatment w th gel l ed ni tr ogl yceri ne, 89%of

the wel l s have producedgas commerci al l y. An addi -

ti onc~

fractionof the fi rywel l s Woi:ld

have pro-

duced gas but for mechani cal probl ems. Past wel l

spaci ngpracti ces ref l ect no syst( :mat i cmethod, but

showthe eff ects of 1) competi ti vedri l l i ngw th a

hi gh wel l concentr ati onnear l ease boundari es,

2) l ess dri l l i ngnear dry hol es, 3) avai l abi l i tyof

proper l and ti tl es, 4) topographyand cul t

features, and 5) company spaci ngpol i ci es.

yf ’

Consequentl y, on? must questi on the accuracy of

drai nage vol umes, recovery factors or other

cal cul at i onswhi ch assume that wel l spaci ng equal s

the drai nage radi us.

Duri ng the past si x years, Col umbi a and

several tens of research, servi ceand support i ng

compani es have studi ed the characteri sti csof the

Devoni anshal e, recentl y i n cooperati onw th the

U. S. Department of Energy. Several of these i nsti -

tuti ons have provi deddata whi chw l l hel p us com

pare the val i di ty of the two gas prodl : i . i onon-

cepts.

DI SCUSSION

Matr i x Gas Model :

Schett l er( l1) deri ved a numeri cs

model fr omFi ck’ s fi rst and second l aws of

di f fusi on(5) whi ch generateda successful hi story

mat: h of actual shal e gas producti onf or hi gh,

medi um and l owvol ume wel l s.( Fi gures3 and 4) .

The model consi ders two separate cases:

1) wherei n constri cti onsi n. thef racturenet-

work (i . e. , mneral deposi ts or parti al cl osure)

l i mt the f l owrate of gas di f fusi ngfromthe

matr i x i nto w del y- spacedfr actures. The corres-

pendi ngequati on accounts for al l but the hi ghest

product i onrates:

‘=%ib)w~-it)

 

2) wherein f ractures are so cl osel y- spacedth

the drai nage zones of adj acent f racturesef fecti v

overl ap or i nterf ere. Thi s rel ati ondescri bes th

producti onof very hi gh vol ume wel l s:

q=-+-b

(

Kuuskraaet al

‘ 8) claimed that di ff usi onf ro

the shal e proceeds too slowy to account for the

observedproduct i on.

They cl ai med that Schett l er

sampl e preparati onmethods ( i . e. , crushi ngand

dryi ng) coul d have al tered hi s measured sorpti on/

desorpti onrates yi el di ng an erroneousl y hi gh di f

si on constant . The constant depends upon the rad

or thi cknessof sorbi ng grai ns or sl abs. They

rcesonedthat mcro- cracksand l amnae se arati on

 

assumedto exi st i n the preparedsampl es coul d

i ncreaset he val ue of the constant.

However, Schett l er’ s earl i er studi es

(11) sho

that the parti cl e size has a smal l eff ect upon the

di f fusi onconstant.

I n fact. Schett l er’ smodel

al ready consi deredthi s eff ect. Col umbi a’ s of f ga

rates measuredon unprepared, whol e core sampl es

l end further support to the d“ff usi on rel ease

rates, or showthe constant to be conservati ve.

Kuuskraaet al (8)

voi ced a second obj ecti ont

Schett l er’ smodel .

They contend that a l og- l og

pl ot of fr acti onal gas rel easeversus ti me shoul d

have a~sl opeof O. 5 for a di ff usi on- control l ed

process unti l hal f of the gas has been produced.

They rej ect di ff usi onas a control l i ngprocess

based on observed i ni ti al sl opes of 0. 8 to l . Owhi

decl i ne to 0. 4 to 0. 6 over 20 years.

However, thi s assumedproducti onsl ope ofO. 5

corresponds to d+f fus?onf romthe surf aceof a sem

i nfi ni te sl ab rel easinggac i n responseto a sudde

pressure drop to a constant pressure (3 and 7). I

contrast, Schett l er’ s f l owconstri cti ons i mpl y tha

the pressure i n the f racture syctemdoes not

suddenl y drop to a constant pressure. Carsl awand

J aeger( 3)

provi de equat i ons whi ch descri bedi f fusi

f l owunder more real i sti c pressure- f l owcondi ti on

i . e. , for gas rel ease rel ated to a l i near pressur

drop versus ti me w th a siml ar producti onsl ope o

1.5, or for a constant f l ux rate at a f racture

surf ~- ”ew th a sl ope of 1. 0(7~.

El ki ns(7) esti matesan expectedproducti on

slope of sl i ghtl . v ess than 1. 0, provi dedei ther

di f fusi onor- Dar~y f l owaccounts f or matr i x gas

rel ease i nto the fr acture system He reasonedt ha

the gas fl owrate i n the fracture systemcontrol s

both the pressure drop i n the reservoi r and the dr

between the f racture surf aces and the wel l head. T

pressure drops i n turn, determne the producti on

rate. Matri x gas rel ease i ncreases as the pressur

i n the fracture systemdrops, but the fr acture fl o

capaci ty decreases. Thi s anal ysi s made use of the

fol l ow ngthree rel ati onshi ps:

.——

8/9/2019 Spe 07921

3/10

For a stepw se l i near pressure decrease i n the

f racturesystem

Pt= Pi -mot- m( t- tl )- . . . mntn)tn)

(3)

..

For esti mati ngstepw segas.i nf l ux to the f rac-

ture systemwhi cn decreases l i nearl yw th ti me:

%

= rmot1”5+rm1(t- t11”5+. . . rmntn)1”51”5 (4)

For the average gas producti onduri ng each

step - based on the back pressure gas fl ow

equati on:

.

‘t- -l

. K~: - p:+: t- l -p: ) (t(t - , ) ) (5,

By consi deri ngthe f ractureporosi ty smal l com

pared to the gas storage capaci ty of the matri x,

he di dn’ t need to I nvol vepressure transi ents i n

the fractures. El ki ns i ni ti al l yassumedr= 1. 0,

P. =

50Gpsi , and P = 490 psi . The above sol u-

t] on hol ds unti l th~producti onf romadj acent

fracturesbegi ns to i nterfere- i .e. , unti l a

l arge f racti onof the gas content has drai ned f rom

the matr i x.

Schettl er has shown that the rate of gas di f -

fusi on f romthe Devoni anshal e matr i x can sati s-

factori l yexpl ai n the observedproducti on rates.

The obj ecti ons to the expected sl ope of the

producti vi ty- ti medecl i ne and the di f fusi on

coef f i ci ent do not appear warr anted.

FracturePorosi ty Model : Accordi ng to the f rac-

ture porosi t. ymodel , vi rtual l y . I l lproduci bl e

gas resi des w thi n the natural l y- occurri ngf rac-

ture system

Rel ease f romthe shal e matr i x pro-

ceeds too sl ow y to contri bute si gni f i cantl y.

The support for thi s model rests on a hi story

match of producti ondata w th cal cul atedDai - ; y

f l owprod~~~i on i n a uni formy f ractured

reservoi r~8].

A f racture porosi ty and perme-

abi l i ty so deri ved can account for the produced

gas vol umes.

However, a revi ewof the i nput data,

assumpti ons and data handl i ngprovi des further

i nsi ght i nto the probl em The matched producti on

data consi sts of averages for several groups:

hi gh, medi umand l owproducti onwel l s pl us

averages over separate geographi cal areas. Af ter

smoothi ngthe averageddata w th Marquardt’ s

Al gori thmto f i t the curve defi ned by

f ( t ) = ( l -e-Bt ) (6)

Kuuskraaet al (8) separatedthe data f romthe

resul ti ngcurves i nto annual producti onaft er

5, 10, 20 and 30 years. Fi gure 5 shows the

resul ti nghi story match for cumul ati veprod ct i on

after 5, 10, 20 and 30years based on the fj l -

l ow ngassumpti ons:

Wel l spaci ng= 150 acres

Gas gravi ty= 0. 6

Net formati onthi ckness

= 580 feet

Reservoi r temperature= 560 R

I ni ti al reservoi r pressure= 500 psi a

Fl ow ngbott omhol e pressure = 100 psi

The resul ti ngporosi ty and permeabi l i ty’ val ues

range f rom0. 31 to 0, 8%and f r~m0. 0175 to 0. 027

red. , respecti vel y.

Cert ai naspects of the hi storymat~h meri t

furt her att enti on.

Fi gure 5 shot ‘ the hi story

match for several cases.

Fi rst, note that onl y f

data poi nts def i ne each producti oncurve. Each

poi nt represents an average over a ti me i ~terval

for a number of wel l s.

Aft er fi tti ng the ori gi nal producti ondata t

an equati onw th Marquardt’ s Al gori thm the shape

of the resul ti ngcurve di ff ers fr omthe ori gi nal

data’ s trend.

Fi gure 6 compares the shape ofactu

producti ondecl i ne data (averagedf or groups of

shal e wel l s) before and af ter thi s smoothi ng

process. Consi deri ngthe di sti nct di f f erences i n

the curve shapes, val i d hi story matches shoul d

probabl y use unsoothed data and comparemore tha

four poi nts al ong a curve.

Ei ki ns(7) and others

have questi onedthe use of thi s smoothi ngfuncti o

As shown by Fi gure’ 6, equati on 6 i mposes a consta

percentagedecl i ne Pate on the data. Thi s yi el ds

l i near producti onrate vs. ti me on sem- l ogpaper

A val i d smoothi ngprocedure woul d use an al gori th

whi ch accords w th the observed producti onbehavi

Proponents of the f racture producti onmodel

cl ai mthat the f racture network can contain the

vol ume of shal e gas produced duri ng a wel l ’ s f i rs

30 years(8). Thi s assumes that al 1 of the gas

occurs w thi n a 500 foot net’ pay thi ckness, over

150 acre area w th a fracture porosi ty of 0. 3 to

0. 8% Actual l y, certai n areas woul d requi re i n

excess of 4%f racture porosi ty. For i nstance, on

part of Fl oyd Co:my, Kentucky dri l l ed on l ess th

a 40 acre spaci ngaveraged bett er than ’ 100MMcf p

wel 1.

Usi ng the recovery factorof 65%used by

Kuuskraa et al ‘ 8) and the fol l ow ngequati on:

“= AhTs@Pf

z‘ fps

(7

def i nes a requi redporosi ty of about 4% (Rememb

that the many vari abl eswhi ch have determnedthe

wel l spacingmake i t ri sky to estimate the true

shal e vol ume d~’ afnedby a gi ven wel l ). I n ei ther

case, these requi redf racture porosi ti es seemmuc

hi gher than those i ndi catedby shal e core sampl es

Core anal yses and descri pti ons ofover 2500

feet of Devoni anshal e f romproduci ngareas of th

Appal achi anBasi n (Li ncol nCounty, West Vi rgi ni a

and Mart i n County, Kentucky) suggest porosi ti es o

0. 0003%to a maxi mumof 0. 01% (Thi s i s based on

observati onsof l ess than one natural f racture pe

feet and fr acturew dths f rcm0. 001 to 0. 03 cmas

measured by Terra Tek of Sal t Lake Ci ty‘ 6)) . Mor

over, the ef fecti ve f racture porosi tymay not rea

these l evel s as a resul t of observedcarbonate

fracture fi l l l ngs.

Col umbi a has reported ef fecti ve permeabi l

i ti es and fr ac l engths cal c:l l atedromtransi ent(

reservoi r tests i n two West Vi rgi ni ashal e wel l s

The test i nterpretat i onscome fromseveral source

usi ng a vari ety of assumed f l owperi ods (i deal

l i near, radi al and spheri cal fl ow). Because the

testi ng took pl ace duri ng the post- f rac cl ean-up

8/9/2019 Spe 07921

4/10

14

GASOCCURRENCEI NTHE DEVONI ANSHALE

SPE 792

------- . ...-. .-— -.. . ..—

process, we used the resul ts to compare the ef fects

of our f racture treatments, but di d not- use the

 

resul ts quanti tati vel y.

Kuuskraaetal (8) men-

ti oned that our report edpermeabi l i ti esl end

support to those generatedby thei r hi storymatch.

t

El ki ns(7) questi onedt he basi s of thi s support

We found that our proppant vol umes and cal cul ated

fr acture l engths requi re fr acture w dths of 0. 7 to

11. 5 i nches.

I n contrast, Hal l i burton’ s rock

mechani c$speci al i sts suggest a maxi mumw dth of

0. 5 i nches and a probi ~bl eaverage w dth cl oser to

0. 1 i nch.

Due to the tests’ short product~on(71

peri ods comparedt o the bui l d- up ti me, El kI ns

rej ects the use of type curve matchi ng for ti eter-

mning permeabi l i ti es, f rac+ti re engths or for

demonstrati ngradi al versus l i near f l ow Because

of the test peri od l engths, resul ti ngl og- l og pl ots

of bui l d- uppressure vs. ti me may di spl ay a fal se

‘ l i neari ty(Fi gure 7) .

El ki ns(7) noted a l i near rel ati onshi pbetween

P2 and ( - - @ for extended ti mes i n al l of

our tests. Taki ng thi s as an i ndi cati onof l i near

f l owf romthe matr i x i nto the i nducedf ractures

(O.1“ f racturew dth w th 25%porosi ty), he cal cu-

l atedan zff ecti ve permeabi l : ”: yange f rom0. 8 to

3. 2 mcrodarci es. Thi s cal cul ati onempl oyed the

foll ow ngequati on for l i near fl owof constant

compressi bi l i tyfl ui ds f roma sem- i nf i ni tesl ab.

The matr i x porosi ty (0.7% was determnedf roma

projectedul ti mate recovery of 575 MMcf per wel l ,

a 500- f oot target secti on, a 150- acrewel l spaci ng

and a 400 psi average pressure drop.

I n vi ewof the data smoothi ngand curve f i t-

ti ng procedures, the present f ractureporosi ty

hi storymatch does not’ appear val i d. Moreover,

thi s model requi resporosi ti es and permeabi l i ti es

at l east one order of magni tude hi gher than those

i ndi catedby core anal yses and themost careful

i nterpretati onof our reservoi r tests. Conse-

quentl y, thi s model does not adequatel y support

producti onf romf racture porosi ty al one.

A Compari sonof the Two Model s: As a further test

of each model ’ s practi cal i ty,we can revi ewthe

f racture vol ume and spaci ng requi rements: the per-

meabi l i ty and porosi ty requi red for the f racture

porosi tymodel ; and the expected gas f l ux f romthe

shal e matri x. AL thi s poi nt, fi el d data, core

descri pt i ons, anal yses and materi al bal ance curves

provi de useful i nsi ghts i nto the probl em

8oth producti onmodel s agree that the f racture

spaci ngcontrol s or l i mts shal e gas recovery -

’ 12) f, ~nd that a singl e fracture

ates. Schettl er

of i ndefi ni telateral extent w th a w dth of 0. 01

cmwhi ch i ntersects40 meters of wel l bore coul d

prov?dea f l owof 50 Mcfd aft er one year. Some

readars may obj ect to the use of f rac;~~esof

i ndef i ni teextent. However, Kuuskraa~u] cal cul ated

that the di f fusi onmodel requi res a f racture

spaci ng (2L) of 100 cm They determnedthi s

‘ Paci y

usi ng a pl ot ofnnrmal i zed ti me

(kt/ L versus the recovery fr acti on(Fi gure8).

Thei r esti mate of the recovery ef f i ci ency(O.43)

deri ves f romthe 30- year producti on”di vi dedby the

matr i x gas content ofa vol ume of shal e 500 feet

thi ck over a 150- acre drai nagearea. Thei r matr i x

gas content ofO. 21 cubi c feet of gas per cubi c

foot of shal e refl ects Col umbi a’ s report edaverage

for the enti re Devoni anshal e secti on i n an area

whi ch appears to be hal f depl eted. I n ar?ycase,

we found that the shal e w thi n the 500 foot stimu-

l atedzone nowcontai ns about three ti mes the

averagecontent.

Thi s woul d suggest a f racture

f requencyof l ess tharlone f racture per200 cm

El ki ns’S(7) esti mate of the spaci ng requi red

for producti onf romthe matr i x deri ves f roma

Car; l awand J aeger’ s(3) equati on:

n.m

~p.mt+ M&Li+,.m@~d..

n=o

KW3

(9

@) - e

-K(2n+l )2r2t/ 4L2

. @s ~ . X

( 2n+l) 3 : ,

Thi s re?~ti onshi phol dsf or l i near fl ui d fl owfr om

sl ab i nto a f racture where pressure decreases

l i r, earl yw th ti me.

These resul ts suggest a f rac-

t~re spaci ngof 1000 feet for a 0. 57 mcrodarcy

permeabi l i ty (wherem=

0. 0488psi / day for1944-

1960, and m= 0. 0422 psi / day for 1960- 1975) .

El ki ns pl ott ed the cumul ati veproducti onvs. P/ z f

38 Col umbi aGas shal e wel l s i n Li ncol n County, Wes

Vi rgi ni a.

He noted l i near pressure drops f rom504

to219 psi duri ng 1944-1960and f rom219 to 183

duri ng1960-1975-

the two sl opes ref l ecti ng

di f f erent hi stori cal devel opment tr ends.

Compared to the matri x model , the f racture

poroAi tymodel demands a greater f racture f requenc

Usi ngMuskat’ s equati ons and the porosi ty and per-

meabi l i ty def i ~~$ by the f racture vol ume model ,

Kuuskraaet al ~u) found a requi redspaci ngof

5 X 10-3 cmw th fracturew dths of2. 5 X 10-5 cm

, 011W3 w

k

‘ Fand T=+

(10,11

Such a cl ose spaci ng contrasts s~f~pl yw th t

actual f requencyof natural f ractures

~o~observed

cores f romthe produci ngarea (up to one f racture

per10 feet).

I f the spaci ngwere suff i ci ent to

perm t the f racture porosi tymodel , the shal e core

shoul d have much hi gher penneabi l i t i esbecause the

must each contai nnumerous f ractures (or converse

theef f ect~ve permeabi l i tymust be much too hi gh).

Fracturespaci ngs whi ch do al l owf or si gni f i cant

f racture porosi ty producti onhave a much greater

potenti al for matr i x producti ondue to the enormou

exposed surfacearea.

Accordi ngto the f ractureporos~ty model , f l o

f romthe shal e matr i x amounts to vi rt ual l ynothing

over a 30year wel l l i fe. I n order to quant i f yth

matr i x and f racture contr i buti onsover any ti me

i nterval , we woul d need to knewthe actual f ractur

spaci ngs and the pressure prof i l e i n the f racture

network. However, we can f i nd the ti me requi red

to refi l l a fracturevol ume per uni t area of

f racture for any gi ven pressure drop AP as fol l ow

8/9/2019 Spe 07921

5/10

For f l owby di f f usi onf romboth f racture surf aces:

I

‘=(*JandG=( fwi~i)

12$13’

ti me for the present reservoj r:

t=* , (15)

 

(The l ast termin equati on 13 deri ves f romequati on

7). Al ong 1 cm of fracture, the maxi mumw dth

shoul dbe . 03 cmaccordfnta o Terra Tek’ s core

anal yses(6). Based on vai ues of d = 2 X 10-7 for

Li ncol n County, West Vf rgi ni a core

‘ 12), fl owfrom

the matrf x coul d reff l l the fracture fn 0. 65 hours

foran. y pressure drop no matt er what f ractf onof

the f racturegas f s produced.

Thus far, we have onl y consi dereddi f f usi onas

the process of matr i x gas rel ease. Hovever, f or

l fnear Darcy f l owf nto a one square- foot f racture,

the ti me requi redto recharge the f racture f s:

F

=~96X103GTz ~

~ kc+

A(Pf -Pt )

(14)

(derf vedby i ntegrati ngequati on8 w th respect to

ti me).

Then, f or pressure drops i n the fr actures

rangfng f rom10 to 400 psf, the recharge tfme

ranges f rom6. 8 X 10-4

to 1. 9 x 10-3 hours (roughl y

2-7 seconds).

Actual l y, the pressure doesn’ t drop i nstantan-

eousl y throughout the f racture system However, we

may consfder the matrf x to respond i nstantaneousl y

to the actual pressure drop anywhere fn the system

Therefore, whether by Darcy f l owor di f f usi ve f l ux,

the matr i x f l owcapaci ty as a whol e exceeds the

fracturef1owcapacfty as a whol e. Fromthe ta

$

n the matr x gas content and fracturew dth( , we

al so knowthat the matri x wf thfn one foot of a

fracture contai ns at l east 40 ti mes that fn the

f racture vol ume.

Whi l e the above recharge tf mes are cal cul ated

esti mates rather thanobserved rates, Col umbi a has

measuredof fgassi ngrates for thfr ty core sampl es.

These rates are sl ower than those expected by

Li near Darcy f l owbut faster than those cal cul ated

by the di ff usfvefl ux (f or gas rel easeto atmos-

pheri c pressure).

Our average of f gas rate was

0. 0013 cfgas per square footof surf acearea per

day at 70°F.

However, the fni tf al rates exceeded

ten tfmes thfs val ue. Al so prel i mnarydata at

el evatedt emperaturessuggest that the rati 2may

fncreasemore than an order ofmagnf tude at forma-

ti on temperaturesover 90°F.

Another l fneof evidencewhi ch suggests (but

does not prove) a dual porosit y i s the slope of

n%t eri albal ancecurves (P/ z vs. G). Ffgure 9

shows the sl ope change duri ng 28 years of produc-

ti on, characteri sti cof Devoni an shal ewel l s. The

cl assi cal expl anati onof thfs observedbehavi or

woul d rel ate the earl y decl f ne as domnatedby

f racture vol ume depl etfonf ol l owedby the more

gradual matrfx depl etf on.

‘ 8) have poi ntedout that the

uuskraaet al

48 hour rock pressuresused to generate these

curves may not be the true vol umetr i c average

reservofr pressure.

They refer to the foll owfng

equati on for dete?mnfngthe requf redstabi l i zati on

Usfng the fol l owfng fnput data, he f fnds a requf red

tfme of 3. 4 years:

A= 150 acres, c= 5X 10-3psf-

u = 0. 012 centi poi se, $ = 04, 005,k = 0. 025ml l f-

darcy.

I fwe bel feve thfs equatf onappl f es and the

assumed data f s corr ect, we woul d have to concedeo

thf s poi nt.

However, a revfewof wel l pressure records

bri ngs to l fght an i ntr ’ stfng contradi ct i on. I n a

careful l y studfedarea ncl udfngLfncol nCounty,

West Vf rgfni a, wel l s compl etedpri mari l y between

1947 and 1956, El ki ns(7) found that fni tf al 48 hour

bui l d up pressures had depl etedval ues approachi ng

the annual 48 hour pressures of ol der produci ng

wel l s (Fi gure 10).

The ol dest wel l s had fni tf al

reservof r pressures i n excess of 525 psi. Pressure

fn the f fve wel l s compl eted i n 1956 ranged f rom350

to 390 i ncreasi ngtoward the undevel opedarea.

Three newtest wel l s compl etedf rom1976-1978

pressuredup to about 250 psi. Al l thi s strongl y

suggests an i nterconnectedfracture network and

that areas wel l above 150 acres can be af f ected

wf thl n 30 years.

El kfns cal cul atedan ef fectfve

permeabi l i tyof O. 004md f romthe data for the 1956

wel l s whtch f l owed 30 Mcfd for three weeks before

thei r fi rst48 hour bufl d up (usi nga fl ow ngBHP o

about 100 psi a at an average formatf onpressure of

400 psfa, and assumnga homogeneous formati onwfth

0. 7%gas ff l l ed porosi ty, an eff ecti vewel l radfus

of four feet and 500 feet of net pay).

I fwe consi deredan unsteady state radfal f l ow

(as fn the f racture porosfty model ), the drawdown

hal f way between arrewwel l and anefghbori ng ol d

wel l produced for sevenyea~s shoul d be onl y 10 psf

I n contrast, the i ni ti al 48 hour pressureof the

three newwel l s (up to 4000 feet f romthe ol dwel l s~

were wf thfn 5 to 18 psf of the annual measurements

fn the nearby ol der wel l s. I t appears that the

pressure tr rthe f racture systemcan equal f zeover

more than a thousand feet w thfn days.

Thi s agai n suggests an extensi vef racture

systemwfth a very l owf l owcapacfty drafni nga

l arge vol ume ofmatr i xwf th a greater potenti al for

gas rel ease.

I fmatrf x gas rel easeaccounts for most shal e

gas producti on, the requf redf racture concentr ati on

i s four orders of magni tude l ess than that neces-

sary for fl owsol el y fr omfracture porosfty.

Actual f racture spaci ngs observed fn shal e cores

compare w th those requi redf or gas producti onf rom

the matrfx. I n addi tfon, the shape of the G vs.   i

pl ots, and the siml ar bufl d- uppressures fnol d

and recentl y compl etedwel l s suggest a dual porosi t

and an i nterconnectedf racture systemdrai nfngthe

shal ematr fx over a l arge area. Perhaps the most

convi nci ngevidence formatr f x gas producti onfs th

rapi d rate at whi chmatrf x of fgassfngcan ref fl l a

f racturevol ume af ter the removal of any port f onof

the gas. Thi s sfmpl e feature fndfcates that matri x

gas rel easeassumes a domnant rol e earl y i n the

l ff e of a Devoni anshal e.

8/9/2019 Spe 07921

6/10

CONCLUS1ONS

A revi ewof the proposedmodel s whi ch purport

to@xpl afn Devoni anshal e gas producti onreveal ed

that’the matri x gas rel easerate i nto the exi sti ng

f racturesystemexceeds the bul k f l owcapaci ty of

the fr actures.,,The matri x w l l begi n to contri bute

to a wel l ’ s producti onas soon as gas begi ns to

fl owi n the fractures. Becauseof the shal e’ s gas

content rel ease rate (by ei ther di f f usi onor Darcy

f l ow), the matr i x must be consi deredthe maj or

source of producedgas.

Consequentl y, the di str i -

buti onof matri x gas concentrati onforms a val i d

basi s for estimati ng the potenti al gas suppl y i n

the Appal achi anBasi n.

However, unti l we know

the actual f racture spaci ngs, di mensi ons, f l ow

capaci ti es and pressure prof i l es for a gi ven wel l ,

we can not quanti fy the matr i x and f racture poro-

si ty contr i buti onsaccuratel y.

I fwe consi der that al l recoverabl egas

i ni ti al l y resi des w thi n a fracturenetwork, the

requi redf racture spaci ngand permeabi l i tybecome

suspi ci ousl yhi gh.

Core anal yses and descri pti ons

of 2500 feet of core-attest to the error of thi s

fr actureporosi ty model . I n contrast, the para-

meter val ues requi redby matr i x gas rel ease fal l

w thi n the range actual l y observed. I n addi ti on,

the cal cul atedrel ease rates compare w th the

natural of fgassi ngrates measuredon shal e cores.

Fi nal l y, the f racture porosi ty hi storymatch

w th shal e gas producti oncontains several faul ts

; ~~; ; t; aveapparentl ygi ven ri se to the msl eadi ng

.

So far, the hi storymatch based on di f fu-

si on appears val i d and coul d provi de addi ti onal i n-

formati onon f racture geometr y, pressure profi l es,

drai nagevol umes, recovery factors and on the gen-

eral nature of shal e gas recovery.

Two other ~pproaches coul d shed more l i ght on

the factors control l i nggas producti on: 1) Per-

f ormng l ong- termtests i n cl osel y-spacedshal e

wel l s to defi ne f l owconmni cati on i n the fr acture

system and 2) Mni ng a vol ume of shal e to study

the natural f racture f requencyand other rel evant

characteri sti cs.

TheU. S. Department of Energy i s

currentl y consi deri ngboth i deas. Of f gas anal yses

on cores f romthe depl etedarea i n Fl oyd County

w th a 37-acrewel l spaci ngmght al so be reveal i ng

The present study shows that any numberof

numeri cal model s can expl ai n shal e gas producti on

usi ng conveni ent data and assumpti ons.

I n wei ghi ng

the cases for matr i x and f racture producti on, thi s

revi ewdepended heavi l y upon core data, because

these provi demore cl ear- cut and reproduci bl e

evi dence.

Resourceestimati oni s a tri cky task at best,

and especi al l y so for the Devoni anshal e. Conse-

quentl y, we woul d warn the reader not to accept

any esti mate of Devoni anshal e’ s gas resources

w thout a thorough understandi ngof thei r basi s.

(For thi s purpose, al 1 of the 1i sted references

w l l be hel pful ). We bel i eve that the forthcomng

report by the Nati onal Petr ol eumCounci l shoul d

better def i ne the vol umeof the shal e’ s possi bl e

gas.suppl y.

.-.

NOMENCLATURE

a = a constant -

7164mcf / day%

b= a constant- 40. 5mcf/ dy

c = compressi bi l i ty- psia-

f

d= speci fi c degassi bi l i ty- i n cm3/ cm2/ torr/ s

f(t) = a funct i onof t i me

9’

constri cti onfaotor - ft

h = net pay thi ckness

k = permeabi l i ty- ml l i darci es

m=

pressure decl i ne sl ope - psi a/ day

q = producti onrate - mcf/ day

r

= a constant that i ncl udes eff ects of f ractur

area, permeabi l i ty, and di f fusi vi ty - mcf/ p

day%

s = fracturespaci ng - cm

t

= ti me

- days

w=

fracturew dth - cm

x = di stance i n x di rect ion- f t

z = gas devi ati on factor - di mensi onl ess

A= area - f t2

B = a constant

C = gas concentrati on- cf gas cf shal e

D = di ff usi oncoeff i ci ent -

i

t /day

G= cumul at i vegas producti on i n mcf

K= a constant= 1000 k/ (w$c)

L = fracturespaci ng+ 2 - feet

P = pressure -

psi ~

T = temperature- R

v= vol ume - Cf ,

Greek

o

= porosit y -

di mensi onl ess

u = vi scosi t y - Cp

Subscri pts

f = formati on

i

= i ni t ial

n = nth term

s

= standard

t= value at t ime t

w= wel l bore

x=

open f racture

ACKNOWLEDGEMENT .

The authors grateful l yacknow edget he ef f

of Li ncol n F. El ki ns of Sohi o, Paul D. Schet. l e

J r. of J uni ata Col l ege and Todd M Doscher of t

Doschers Group i n researchi ngthe mechani cs of

gas producti on. Most of the mathemati cal appro

revi ewed i n thi s paper come di rectl y f romrepor

by and correspondencesbetweenthese i ndi vi dua

REFERENCES

1. Brown, Porter J . : “Energy fromShal e - - ALi

UsedNati onal Resource”, I n: Symposi umon

Natural Gas f romUnconventi onal Sources,

Nati onal Academy of Sci ence, Secti on I I ,

Chapter 6, pp. 86-99, (1976).

2. Browni ng, I l eyB. :

“Rel ati onof Structure

Shal e Gas Accumul ati on”, I n: Devoni anShal e

A Symposiumby the Appal achi anGeol ogi cal

Soci ety, Charl eston, Vol . I , pp. 16-20, (19

8/9/2019 Spe 07921

7/10

3. Carslaw H. S. and J aeger, U, C. : “Conducti onof

9. Laffer ty, R.C. :

“Occurrenceof Gas i n the Devo

Heat I n Sol I ds”, 2nd Ed. , 51~ p. , Oxford Press,

ni an Shal e”, I n: Devoni anShal es - A Symposi um

(1959) .

by the Appal achi anGeol ogi cal Soci ety, Char-

l eston, Vol . I , pp. 14-15, (1935).

4. Chase, Robert-

P. C. (Professor, Petr ol eum

Engi neeri ng, Mari ett a Col l ege, Ohi o) .

10. Ray, Edward O. :

“Devoni anShal e Devel opment i n

Eastern Kentucky”, I n: Symposi umon Natural

5. Crank, J . :

“The Mathemati csof Di f fusi on”,

Gas f romUnconventi onal Sources, Nati onal

Oxford Press, London, (1954) .

Acade~ of Sci ences, Secti on I I , Chapter 7,

pp. 100-112, (1976).

6. :=; ~.P. , McKet ta, S. F. , Owens, G. L. and

.

“Massi veHydraul i c Fracturi ng

11. Schettl er, Paul D. J r : “Studyof Hyi ro~arbon

Exper; me~t~”of the DevonianShal e”, Col umbia

Shal e I nteracti on”, I n: Report ORO- 5- 197- l

Gas SystemServi ce Corporati on, Col umbus,

thru 11 for U. S. Department of Energ , J uni ata

Vol s. I and I I , (1979) .

Y

ol l ege, Hunti ngdon, pp. 1-13, (1976 .

7. El ki ns, Li ncol n:

P. C. and conmmni cati onson

12. Schettl er, Paul D. J r. : “ATwo Step Model for

f racturedreservoi rs to Todd Doscher of Lew n

Gas Producti on f romLowPermeabi l i tyShales”,

and Associ ates and Don Ward of U. S. Department

I n: Report ORO-5197-11, Appendi x C, pp. 1- 18,

of Energy, Washi ngton.

(1978) .

8. Kuuskraa, Vel l o A. , Brashear, J . P. . Doscher,

13. Smth, E.C. :

“A Practi cal Approach to Eval ua-

ToddM , and El ki ns, Ll oyd E. : “Enhanced

ti ng Shal e HydrocarbonPotenti al ”, I n:

Recovery of Unconventi onal Gas Sources”,

Proceedi ngs Second Eastern Gas Shales Symposi um

Secti on on Devoni anShal e Gas of the Appal achi a

MorgantownEnergy Technol ogy Center, U. S.

Basi n, Vol . I I I , pp. 1-79, (1978).

Department of Energy, Morgantown, Vol . I I ,

pp. 73-87, (1978).

8/9/2019 Spe 07921

8/10

r COMMERCIAL OPEN FLOW

I

I

i

Ftg. 1- Columbta’s Devonian shale gas development act iv l t y.

240

t

200

- t

I

4

:-

 

e 12

16 2C

YEAR6

Ffg . 3- History match of shale gas p roduction wi th gas recovery

calcu lated by dif fusion for low and intermediate production).

After Schet tler .’11 )

Fig. 2- Frequency of natural open flows in Davonain s;ale w

wells.

320 -

260 “

40

t

Fig. 4- History fiatch of shala gas production ISith gae

recovery calcu lated by diffus ion for h igh product ion) .

After Schat tler .’11 )

8/9/2019 Spe 07921

9/10

400. -

/

HIGHCASE

o

~ Soo. -

MEOIUMC&

a

o

II

-/“”-

LOWCA%E

200

3

8

 

g

ij ,00

u

1~

ACTUAL PROOIJCTIOHTA

0 Awwiq ~ 4’

D Amllwlg datwwltol

fmwwa Prtaolt.

‘-r-

io &

TIME ( TEARS)

/

6

ACTUAL DATA

SMOOTUSDDATA

5

\

a H16H2622

4 HWI CA6E

4

\

: il X&d&E

m Kuw CASE

\

  LWC5S4 300

\

4

00

 

I

0

60

40

30

20

=..,,.,

,0

9i14

6Ulbli? 14161’Stiti24 262jkwAJlS

Fig. 6- Production decl tne of Devonian zhale WS11s.After Elk fns 7)

Ftg. 5- Hi etoryatch of simulation and ffeld data

for thres t y pf cal Devonian nel 1s. Af far Kuuakraa

et a l, 8)

-100,000

0

OwlatiohromLinwily

0

:B

0

K

-10,000

PR@UCTION PERI~ =200 HOURS

[

I

,10 ,100 1,000

Fig. 7- Theoretical prassure but ld up for I f near flow. After Elkfna{7)

8/9/2019 Spe 07921

10/10

[{ /

,4

g

 

.3

 .e

.1

0

~Y,,,,,

 1 2 .3 .4 .5 .6 .7 .8

.’s

......................................

—

I

~a *

I

1

w

_..Awmbllec - –

‘“” - - ““ “-

-.

 

akmah

u ~-

Se.s emmr~-wt

RDmd

Fig. 10-1956 shut-in preesura distribution in

Lincoln/f ayne County ehale nel 1s.

After Elkins. 7)