Está en la página 1de 61

.

. .

c,
<

~ov~

/ ca>-g~

,.,

NEW PRESSURE TRANSIENT ANALYSXSMETXODS


FOR NATURALLY FRACTURED RESERVOIRS

.:,

....

. .. . . . ...

,.

K.

,.

.:,

..:

SerraL A C. Reynolds

Raghavan

. .

. .. .

.. -

.-..
..
,.
.,...-;-.,.
,.. ..............

.
.
.

NEW PRESSURE TRANSIENT ANALYSIS METHODS


FOR NATURALLY FRACTURED RESERVOIRS

ABSTRACT
The objective of this paper is to grzsent new methods for analyzing
pressure data obtained at a well IocsNed in a naturally fractured reservoir and
produced at a constant rate.

As is typical, a naturally fractured reservoir

indicates a reservoir system in which the conductive properties of the rock are
.
due to the fracture system end the rock mat~ix provides m&t of the storage
CWaCity of- the system.. _~e modei -@_ in this .xudy is Jd_gntical to one pro&sed by deswann+

and also considered by ??ajurie~

&tstea& state flhd titer


. ...

from-~

that is,. we consider

matrix to the. fractumk

~.

~ A new flow regime which Me not been reported. until now is identiML

This flow regime sitouid enable the engher

to determine matrix perm-

eability if the matrix porosity is known or vice versa

It may also be possible

to determine the ratio of the matrix storage capacity to fracture storage


capacity.
The discovery of this flew regim,e also exphins fieid behavior
which has been considered unusuaL

A@icstion

of the theoretical

results to

fieid data is also discussed.


. ..

.-

.
-.

..,, .
,.

:.

&

...

.
.

INTRODUCTION
In this wc?k, we investigate the pressure response in a naumaily
fractwaed reservoir.

Several models of naturaiiy fractured

presented ig the literature.

Warren and Fmtl

assume unsteady state flow.

Sawyer6

have been

and 0deh2 assume gseudo steady

state flow in the matri~, whereas, Kazemi3, deSwn.n4.4,


m.d

reservoh

L~a@k~~s

KUCUk

SSId

The model used in this study is

identical to one proposed by deSwann-0.4 and also considered by Najuriet&5


ti

mOdd assumes that the matrix is divided by $ ~tof

e~q~y

sga~d !iorizont~ fractures and also assumes unsteady state fluid transfer from
the matrix system to the fracture system.
-We consider cgn@jtt rata- production foiiow~ by a buii&
and

present

new-:meth~:
,..

for anaizying both d@iKIowq ,ti

,Smdup ~t~..

Fesuits. are baaed ..on the. analytical. solutien and, thus, ~me
pressible reservoir fluid of constant viscosity.
here can be applied to gas wail tasting.

period

-.

.W

a Q@tiy qm:

- ~~~:

However, the results presented

The results presented here a&sume

infinite reservoir behavior; however, we have completed work on the bounded


reservoir case and these results will be presented in a subsequent paper.
We identify three possibie flow regimes.

The first and third flow

regimes correspond respectively to the classical eariy and late time _behavi_or
reported by Wam%nand Root, deSwarm+. and others.

In this paper we refer

two flow -regimes as flow regime 1 and flow qime 3, reapectiveiy.


.

.
.Dw+~. both_of thQse flow periods, a ~milq..piot
of pw~- (dimen@o~~ -wellto these

bore ....
..presmre drop) Versue ~

($irnensioniess time) exhibi* a_st@@t_rmO:_

s@e 1.191. The classical Warren ani Rat q~


-.
...
.
-----

.
-,__.,__,
,. .. .
..,:
.. . . .
.
. . . .. .
.=- .-, .. :-. ,.. -. . . . .,. :.- +.

p&~

.~ _

on .&

: ___
---- -.,

._- __ -__:_ ~

: : ~~

,..,..=
.

.
.

-.,
...,-

:,:, -

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

-; :.,.: -;: .::;,

:;::::- :.,

..

,
.

existence oi both straight Iinae.

However, for unsteady state fluid transfer

from the matrix to the fracture,

References 3-5 show that the eariy time

straight Iine of Warren and Rootl may occur too eariy to be observable in
practice, and, thus, analysis must be based on the late time straight iine.
The late time straight line yieids directly only an estimate of the fracture
flow capacity.

Najurietas suggests that additional information can be estimated

using a trial and erxmr procedure based on regression anaiys&


we establish the existence

intermediate

of an

In this work,

time flow regime.

We show

this flow regime provides ~ethods for preparing a com@&e analysis of pres-

... .

sure data obtained from naturally

reservoirs.
.

fr9ctur&i

In this work we refer to this intermediate


refjime 2.. Duri~ .f!ow regime 2, a- timil+

noted

prdoualy

on field

flow regime baa- riot

tests,

time flow regime as flow

pltit of pwD vemM tD .exhi~its a

to our knowiedg&,the a&@icai @ty

been previotisly

dfscuss&L

The

iifentiiication

of this

Ot-this-ffow

regime aiso resoives the quandary of Adams -et SL7 who observed two semilog
straight Unes in several sets of field data with the siope of the second straight
line equal to ~onehdi

the siope

of

the first straight line.

tie show that if the

reservoir is known to be naturally fractured, then situations of the type confronted by theee authom can * resolved
.-

without

invoking

a compoeite reser-

voir modeL
In this work we establish the

- the -s@l@g sgwig$-ljne


tions of matrix-fracture

of slope-0.575% ~

results:

timc3@sight-lin*

.(3)if :th-:rn&ti.:por&ity

..

and Iaietirne %emilog straight lims

- ~

---

flow capacity_ ciin be

-. is kn6wn, the matrix permeability Cias-ti eetimatecf&n

;. . ..
---------Imnmsdiite

(1) the intermediate

be apparent fw mga-gombirw

pareinetem: (2)the frsicture

.--calcuiates-.frorn: the-intermediate

following

inteSect:

t~

~.:-..----:-

time=when the ~

or,.iithe :m.=*

~. ~-

-.,-.
-...

4
permeability is known, the matrix porosity can be estimated from this intersection time; (4) the skin factor can be estimated using the intermediate
$ata; end (5) the ratio of matrix storage capacity to fracture
can be estimated using data from the intermediate

storage

time

capacity

end late time straight lines.

These results apply to both drawdown and builcitm analysis and to our knowledge
have not been previously reported in the literature.
In addition, we delineate the time periods when each flow regime
exists

and

indicate the- information that can be obtained from each possible

combination of flow regimes.


in practice.
. ,-

We aiso show how to identify these flow regimes

This is important since for certain combinations of matrix-frac-

ture properties and producing times, a semibg plot of drawdown or buildup


data may exhibit only one straight line, and this Lirie may cmespocd

to either

Slow regime 2 or flow regime 3 or even flow regime 1 in the ease of gas $vell&

.,.

-+

.-,

...

. . .. .

.. .

--

.. . . .

:..

,.

MATHEMATICAL SICDEIi
We consider the flow of a slightly campressibie fluid

of

cmmtant

viscosity in an isotropic, naturaiiy fractured reservoir of uniform thickness.


The top and outer bmndarids are ciosed and the drainage radius is infinite.
TLe wail is produced at a constant rate and then shut in to obtain buiiclug
C!at& Initiaiiy the pressure is uf;iform throughout the reservoir.

We assume

ail production is via the fracture system and that we have onedimensional,
unsteady state flew in the rnatrk

tures
,.

A schematic diagram

of

The matrix structure

the reservoir geornetq

consists of Fectan-

@ Show in Fig. 19

We-consider an &initesimaiiy thin skin region and nagjaot w-e

deSwann-0.4 and also considered by Najurieta5.


Sn Appendix A; we present

the

relevant initial bomdary value p

biem, the analytical solution of this probiesn in Lapiace space, and intermediate
and long time approximations to the analytical soiutiom
approximation indicates the possible .qistence

The intermediate

of a new and

useful

flow

time

regim~

We show that this flow regime does indeed exist by inverting the anaiyt@M
soiution using the Stehf&t aigorithm8.
,_,

. ..

-.. _

in this study am obtahed *.

We emnhasize that the results presented

the antiiyfieei s@utfon .-,=


.
.
mdiateS KldhgWnta
aQpradmMms
. .---.
,-

.
-.

fron..the

intee.

_- ._. __ __ _______

.
-;D=OtiS-,:- ---:
--T- - - ~~~~ =
,.
..

.
,.. . .We-&&~j&
:~~~%-~
&*--G*~&&&&
~~
:; ~
..
.
...
. . .:
---
.< -.. ..,.,,,- .,. ....... ...,- ..... .,..-. . ... .,.-,.- ------ 4.-..~,-, -~--,..-, ,--. s

6
time respectively by

(1)

and
2,637 X 10-4kft
.

D =
9

(2)

qcfur:

Note that these definitions are besgd on fracwre properties snd that oil field
units sre use&

(If Sl.
metric
units ere used, the co!wta~~ in Eq. l_anc_2
..
=6
should be replaced by I/7.2z x 10 and 3.5 x 10+, respectively.)
We.cfefiie t~e. &meneiotieas matrix st&ativity;
sioahei framm
. . .. .

tifi~<~l,+~y,~
tiansfis
.

- ,

w, and _the ~me~


-

~;
=

.:

.
. ..0.

. (3)

- -

I#h

A?

mm
= 12 ()
~4,

(%

(4)

h2In

All symbols used in Eqs 14 we defined in the nomenoiatum


\
..

.-

To avoid oonfusion. we compare our syetem paramet-,


a, and A,
.. .
to the_.@as@c_@
u ~, ~_definitiq. d -WmeXI m. ROOtl. .For the deSwann@
m@ei which we consider, u and x. are given, reepeetively,

by.
(See
,.

Refs
L 3
T ,.
...

__
- .
.

,.

.-, . . .

... ....
... . . :----- -----

,.
.-
---- .,--- .,.

._
,.
=. .,.
.
.... .... :..-..?:,.-,...++. . ..... ...:..... ...

.
.
,. ....
. ,+ ..... .1.
.

... ...: ,
: -..,, .Z. ...::- .-. .

7
and 9).

andkfie the bulk fmmire

Hefqanomsreb%tlk

(5)

(6)

permeability.

For our modeL we have

?+f

Qf-=~

.
...

in=MI.-.
--

where h is the formation thickness.

1-

-.

..-

T-

{7)

Frem E-

(8)

**.*9

(9)

3, 5, ? and 8 we see that

(u

U.=*w

Moreover, if hm g h,. then. ~

4: 6-.and,9 tiPiy that

(lo)

~.

-..

Equations 10 and 11 previde the relationship between our parameters and those
of Warren and Root 1.

We preier our parametem u and Abeermse they involve

intrinsic instead of bulk properties and S&O bemuse of their physical hterpr~
tation.

Note M is the ratio of matrix storftge capacity to fracture storage

capacity and x invoives the ratio of the matrix flow capacity to the fracture
flow capacity.

(Note that the preceding discussion aiso implies that the Warren

and Root 1 soiutions can be rephrased in terms of intrinsic properties.)


-Raghavan and Ohaerig state that the range for u is 103 ~ u ~ 20-1
and the range for A iS 109 : A : 10-3.
to 10 ~ u ~ 103 and 104 ~ A s 10-5.
.,

These ranges correspond apgroximateiy


h this work, we expand these ranges

slightly end eomsider 10 ~ s ~ U3: and 104 g A ~ 10-3. Note that this fnaSSS
+1
that the riqe for i J is 104- : IS&: 10

BACKGROUND
. .
As indic.atkd kti

the *&d
. .
down prkssure responses are given, respecti&ly, by

... L
. &.4:a&i5,

&e-time

dra~:
. . . ..

..
\ ..

PWD = 1.151 [kg tD + 0.3511


.

(12)

and

s..

Throughout this work we refer to ES


1 and flow re@ne3.
may _

at

This terrninol~

,iiiteim*ate
.,

S& :KbedcgroUid:foF
-.
-.
..
. . ..
.. .,. - ..
,...--.+..:. .
-,.? .
+.-

* (13)

12 and 13, respectively, as flow regime


implies that another

tifuf

fftiw. regime

times

We -discuss this flowre@ne @er we pr~


._
-.
thi ZWwariiiys& -tGdiru~-.
tfotehiiii-:~. Mrni&g+iot -of--

.. .

.
,-,
,,. >,
. . .. .
...-.
.
...
... . ..- :.. ..-:.... ....-.--.-.!-. .,,.,, -..,.. . .. ...-.

-.
....,, . .,,:..;..

3
.

PWD WX$US
tc wiU exhibit a straight line of siege L151 during both

flow

1 and 3.

~=

Equations X2 and 13 may b~ mvritten,

Pi - ?wf =m

f%

logt+kq(
{

respectively,

.# 3.23

YVCFW

as

+ 0.869 ms ,

***e*

(14)

.
.

..

o.a69 rns ,

-where

(1s)

.-

m= 162:S q Bu
A

When either Eq. 14 or 15 is vaiid, a sernilog plot of pwf vemm t will exhibit
a straight line of slope m. The fracture flow capacity can then be estimated
.-

by,:

*W***

..

,.

.If the-matrix @@M@-k


- known, tM sti factoF can th&nbe estimatti
.

(17)
-.

from

---

~
1

..

assuming

(18)

The appro~riate formuia is given by

+ 3.23] .
m

The skin factor cannot


porosity-thickness

.-

be

be

(19)

$mcmhur~

estimated from the early time data unless the

product for the fracture or the fracture diffusivity is known.

tU discussed in References 1-6, the dimensionies

matrix storativity,

u, _can be ccrnputed fmm the verticai separation, Dp, between the eeriy and

9te

time straight Iines (Eqs. 14 and 15). The appropriate formula is given by

..

I
a-f: =l@Dp/m=i

--

;.

._.-

.
.

*#-**,

.~a)
I

The results of References 3-6 and the results presented here indicate that the early time straight line, Eq. 12, will not always
on fieid data

be apparent

In this case we can estimate the fracture flow capacity and

skin factor only by using Eqs. 16 and 19. .As already mentioned, we esmbi.ish
the existence

of an

intermediate time flow regime from which additional infop

mation can be obtaineti

We present methods of analysis assumt~ the eariy

time flow regime, E% 12, is not apparent.

However, the, reader should bimr

in mind that if the early time flow regime is apparent, then d can be com--

@ted .froin ~
ar@yeie

20. The

techniques

that

abo~-~ori
we present below.

pt%videithe

baekgrouix-for

M&-new-- --

DRAWDOWNSOLUTIONS
The analytioai solution is eompktely

determined by the Yaks

and J (see Appendix A). In Appendix A, we derive the intermediate

of ~t

time

approximation given by

P,M~ -= 0.57S6 [l% tD + 00452 - l% (Au/3)] +

(21)

. . .

This intermediate time flow regime mews between flow regime 1, (E-

22

and 14), and flOW regime 3, (Eqs. 13 and 15), and is, thus, referred to thrcugh.

:,

out this work as flow regime L

Note that during flaw regime 2, a semilog

-plot of pwD verzns: tn exhibits a straight. line of siope O.SWM(1.1 S1/.2). ,


Using the definitions of tD, e and A(~
~q, -2$.-

2-4), we may rewrite

Pi- pwf =. m*. [log


. . t + iOg
-

49+2..:..
km4metiur:

-,
:
- 3.?29] +- 1.737
-~~~
* *.

*.*

.,

-:.:

M*S,

(22)

where
m* a 8129 q u .
f%
(Note that m* = m/2.)

.*

*.**

(23)

Equation 22 implies that during flow regime 2, a serni-

flow oepadty oan k caio!!iated fhm this siopwusing

the foiIow@g f~u@

---

--

K estimat~

of the matrix ?ermeabil.ity and porosity can be obtained, the skin

factor s can be obtained by e.xtrapoiating the straight line of slope m* to t =


1 hr. The appropriate formti

is gfven by

Figures 2* represent semibg plots of pwD versus tn for various


VdUe!SOf A!and @. ~ each figure (Figs, 2-5), the skin factor is zero, Q is
fixed

and

A v*=

folIowing observations are aOparenL

The

First, the early time straight line, E* 12, h apparent oniy for u
g 102. The duration of the eariy time straight line (flow regime 1) increases
.
as Asnd u decrease. For the case d = 102, our results indicate that the
eariy .tirne stnight -line of siope 1.151 exists oniy for A~ 104.

However,

. .. . .

..

~ 10-.

- For the results shown, the longest eariy time straight line occurs

when 1A= 10 and A = 10-9.

In this case, the eariy time straight line ends

at approximately tn = $ x 105. TM red time mrrespondin@ to


depends on the properties of the fracture system.
value of tD may correspond to a minute

or

tD

3 x 105

For the liquid case, this

Iess (see

Refs. 3-3).

However,

data for a-gas carbonate reservoir gmtsidered by Adams -.


et aL7 seem to cieariy
indicate the presence of

both

w*m:~,mt
-y

the eariy and intermediate

time straight line

ush&_@
d~~, (m~2ti

SJl Of the pFeSSlk- C@tamay M

3),...

on the late time &@@t ~e~ ~

.h .Q@i dese,- &-.nQte&by Od@2, the -pr&k~


. ..,-- _____ . =.-,..--.,
,=.

reapnsw o!- a &&@,ly&@t@.


..

... . .
.:. .-.
. . ......-u.,..-.: .,.-. . . . . . . .. .. .... ,. .:.. . . ..., ....
-.-
-. ..., .. ... . ,. ..:.-...... .. ... .. . . .

i3; --

~~-I-..

.. _ ..J
. ,.-. .-,7,... .-,

reservoir will have ail the characteristics

response in a singie

of the pressure

layer homogeneous system.


that flow regime 2, (Eq. 21) is

The most important observation

not apparent for U = 10 but ia apparent for ail values of A if Q ~ 102. In


Figs. 3-3, flow regime 2 occurs between the points labeled X.

By making

expaiii.ded piots such as shown in Fig. 6 snd .%mputfng siopes, we me able to


We find that flow

correlate the beginning and anding times for flow regime 2.


regime 2 be@m and ends, respectively,
. .

. . ...*

(26)

*9***

(27)

and
. :.

0.5 (AI
tDe=.
2

The approximation given. by Eq. 27 applies for all vaiues of x and u. The
approximation given by Eq 26 applies only for AJ ~ 102.
Eq. 26 overestimates

the time when flow regime 2 begins

For Id

>10-2

For exampie, for

~d = 10-L flow regime be@?~ at tD = 2 x 103. Note that E- 26 and 27


n
indicate that if w ~ 104 we will obtain at Ieast one log cycie of data corres
pending to flow regime 2.
.

,.,

Moreover, as u inoreases, the Iength of this inter-

mediate time straight line will increas& Figure 6 shows an expanded semilog
.
piot for the case W =102 and A = 10+. For thhoase--~Db = 5 x 10~ and
lDe = S x 107 and theee times are
do qbaqve 8 W**.*-.

,.

!m~ ~Db
De
.,-.
.
.
.
.
.
.
.
.
.. .-.
.:... 1..-...:.:.-.,,,.?..C:..- ..:-.
.._

indicated

by X. on Fig. &

~.ygi.lw Q@@!!..q~..of *.-

..

. .

o~?s?

Note that we
@egw.wn

the ... . . .. .::..:


-..

-..

..

14
We can also correlate the time when the Iate time straight line,
Eq 13, begh

This time is

given

by

(28)

9.****

Equations 27 and 28 indicate that the transition from flow regime 2 to flow
regime 3 w

approximately one log cycle, (aim see Figs. 3-3).

Several other points regarding the above analysis deserve emphas~

First, if flow regimee 2 and 3 are evident, then the. dou&Ung


of the +opq on
.
. .
,
semilog coordinate is usually interpreted aa indicative of a seaiing fauk
Second, if slow regimes 1 and 2 are endent,

then one is inciined to attribute

the reduction in siope to the existence of more permeabie awe

seriously considered, changes in s@e appear to be


of weils

fundamental

or an increase

to the behavior

in fractured reservoirs and changes in slopes can be consider&i to be

a reflection of the flow regimes that govern the well response. In

this

context,

it shouid be noted that the discussion given here can be used for other idealizations given by deSwann (spherical elements) and Kucuk and Sawyer.
Physical Interpretation
As is welMnown (see Refs. 1-5), flow regime 1 ocmrs when
pressure response is dominated by the fracture system and the duration
@flow

..

period ~.

.as the dimemionleaa fraoture &anaf* .Coeffieknt,

A,and the -dimensionless matrix storativity, M, deereaeia


regime
the peasure
..
-----

of

reegmse

behaves

ss ~~..=
O.:(,-. ,

During this flow

F:.=.. !&..._

-.

, ~ . ..

..-

The

remarks

p-

indfoate -that

the

time reqqired to reach the

semnd semiIog straight line (Eq. 21) increases as AU deereases and this is
indicated by E+

2S.

During flow regime 2, the pressure response is irulluenced

by both the matrix and the fracture systems but the matrix boundaries~
(centerline of the matrix element) do not intluenc= the pressure response.

As

indicated by E% 21, pwo is determined uniqueiy by the vaiue of 1 d during


ffOW

regime L

More generaiIy we find that PWD iS ~que~y

determined by

xJ for aiI thee

prior to the end of flow regime 2. This Iaat statement k


,.
:.qisc illustrated by ttw resul~ shown in Fig.
tX .vW. @ A.a@
0- 7.. I/nF~ 7U
fhd

but

the

individual vaiuea

of

l and d vary.

Note that in thie

figure, the pressure response prior to the end af flow regime 2 does
not d-

real time

on the individual W3hleSof

~ehours

~t end Q .

Flow

regime 2 ends at IM

where

(29)

Similariy, the time when flow regime 3 begine, iDg cmresponds to a real time

+mcmm:
= 1S88
i%
.,.
.!m

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

,z

_<_

..,.

_i.

,_ -:.-.:

bduwior

. .. .. . .. . . . .
.- .-, ..

and,

thus,h&n6s

.4-.

.,. .,:

we bv8--kkhw

..

.,.

. . . . ..

. .. . . .

ps6~0-~e~~-

...

.>.

(30)

.,-

~a~~n

...=
flow

in

-..
.

16
the matrix systefn.a
Identiffcatiori of FIow Re@mea
In the event that two or more sernilog straight lines are apparent,
the corresponding flow regimes can be identified by the siope ratios.

However,

if U is small and a is Iarge (see Fig. ~)


. oniy the late time semilog straight

line may be apparent.

On the other hand,

Fig. 5) only the intermediate

if

U is large

and A is small (see

time semilog straight line may be apparent unless

the test is run for an exceeding

long time.

In cases where the pressure data

exhibits only one semilog straight line, misidentifying the correct sernilog
straight line will yield sri katimate -of fracttn% flow ca@ity
error by a factor
~
.

of

2, (see E-

whicfi ~

~ in

17 and 24). Our results indicate that for

~ 10-1, the pressure data on a dimensionless plot always approach the

intermediate

time st@ght I@e

line from abovregimes,

f~rn

Th@ o@svatj~

However--we

beiow

ti

the

Iate t$me .sexnihg straight

may be useful in identifying the flow

emphasize

-this observation- can fl&tbelfsedto

that

identify flow regimes if kwg 10-1.

But, if the Rag&wan and Ohaerig ranges

for A and U hoid, then we always have i~ ~ 10-1.


be used

27 can ah

identification

for

fractures properties are avaihbie.

Note that Eqs. 2S and

purposes if gross estimates of matrix-

In fact, Eqs. 2S-28 may be used to design

a well tAnaiysis Using Flow Re@mes 2 and 3


In thie section we present procedures for analyzing data when flow
..

*m&

2-~

flow c~ci~
:

,,.

.-t.

----

3 &: -L

~t the f&et&g
.
eesl be estimitted from either the interrywiiate or late time
---

-Un--(see

-.

We-~ve s~~

-------

____

-E&17&-24)e- ----

.. .

-.,

~~~~y

-- ------

Can.
be korpora
the.,.
frtuliewom
o$: *.
;.. ..,:. ted
... ...with@
..- ,.
..,,..
.. ...-, .<...-,
. ,.-

~ ----

-~.~~e.rn.@*--..-..
. . . . . . .. .

--

,.

- ;:.,;-.., .

. .

17
13 ind 21, we find that the intermediate

m+--

and late

time straight lines intersect at


3 *I

.*

This vafue of t:

(31)

corresponds to a real time t* (hrs) given by


532.3 @mcmh:
t* =

(32)

*.*..

Thue, from the intersection time t* (on a pwf versus t semileg piot), we can
estimate the matrix diffusivity by assuming hm
@ @own, we ean eetimste the maeetimate

U the matrix porosity

permeability from Eq, 32.

Afrom E* 4 ueing the &ate5

We can then

of km and k~f and @@~

hm i

hw

.
We can aiso estimate

S USi!lgE% 19.

~apolating
,,/

k in the following manner.

We first compute

E4 15 to t = 1 hr. and rearranging, we o&tain

Pi - Plm, ~= m

i-

r
-~
- I% (1 + a) -3.23
+ 0.869 rns ,
)
$fcfilrw

--+
9

(33)

where the su&cript L refem$ to the late time sesnilog straight Jhte. E.xtrspoia. ..
=s
w.= @ t = 1.JW**.~~mlmw
th re=wng.equatiall
by 2 anc. ream@l&
$Vee
,--..- -. . .. . .-----..
,,
.
. . . ..
,., . .. . . . . .. . ,:,.
.-. .
.,.:.,- :---- -----


.,
,7,

~~

\
..

...

?-(~i::. F;M ;; = m {-l; #+*


~~ (~hi)_o. 2.83;..}
--. :. _-,
.
.- ?=.
--4fopFw
-.
-.
.
.
::
:
,:
-,,:
:
...-.
:..
.
. .. .. ,.
>., ----- . . ..
. . . . ...
:.
7- .,
+ ~m(Oa~69$)
-o-.
; :--0: ~3~ ---; ,_:

We know that if the intermediate

time straight line exists, then 1 + u t = U1.

Subtracting Eq 34 from E+ 33 yiehis an equation which can be soived for A,


.
The remit is

x
i = 10

.*..,

(35)

ee e..

(36)

where
- ~lhr, 29i
Xab 2 lhr ,int ~

We can obtain an estimate

,.

of $~C~f or squivaientiy u by estimating

whenthe flow regime 2 _

The appropriate equation is

he ime b
obtained fram Eq. 26 and is given by

4@f

+ 08$9s + O5?8

= 7.2S2 X 10
m

Note that once the fracture poroeitycompreesibility-thicjmess

(37)

pl%duot

is eeti-

mated, then E+ 3 can be used to obtain w.


If all observed data correspond to either flow regime 2 or flow
regime 3, we can oniy obtain en upper bound for lb.
fieid an wper bound for $~~f

and 4.

This means Eq. 3? will

3 will then yieid a Iower bound for J.

The preeedhg observation qn be demonstrated if we oonsidar the relevant


solutions

If flow regime 2 exists, then our previous resuits indicate that

-1 + u .* d, and hence, E% 13 may be WIitten as

PW* = Ll$l

[log(tD/u) + 0.3511 + s

*****

(38)

-.

. ...

19
Equation 21 can be written az

Pw~ = 0.5736 Ilog(tD/w )+ 0-452-

Equations 38 and 39 indicate that

if

log (A/3)] + s . . . . . (39)

the solution is graphed as pwD versus

tD/ wI then the soiution depends uniquely on A during flow regimes 2 and 3.
Since

t~a:=.,

2.$37 X 10+ k~ft


.$mcmhm+

we see that the solution is independent of the porosity-thickness

*.***.*. (40 .-

product of

Equ@ions 38 and 39 kuggeet,@t we can abo adimate A and


olkin

a Iower bound for u;- U&

sernihg type curves prwided that we have

data from both flow regimes 2 and 3.


versus

tD/d

FiguEw8 praaents a @@

of PWD

for various values of X. The dashed lima oorreapond to vaiuea

of u t, and tlteae dashed iinez intersect the X curves at the points where flow
regime 2 begins for the cwresponding values

of

A!

and

w.

It ia important to

note that Fig. $ only presents a graph of the two semilog straight lines (Es
38 and 39) and that Fig. 8 assumes that the skin factor is zera
cenbeused

aaasemiiog
The-first

...-.

typa curve in two ways,

way-assumea that- we-haveait

which ia sufficiently a~te

estimate of-matrix po~ity-

so that we are ebie to obtain aoc~te

:-rnaise-.a _swn*
@x -Of.& .(pl- o +w~. v_*_!

., ..+ ...-.....4 --T soaiei +3f-FigO-8.I We

Figure 8

m. .tra@ng:pag4r

then simply ini@h - da-t~

eatimatee.

~-_&_-:.,

-;.:.:.-:

k eppmgriute -A--- .:. ~~~~


.-

curve in Fig. 8 and read off the lower bound for u 1. (If we have data prior
to the start

of the

straight line of slope 0.37S6 we can,

sernilog

obtain a direct estimate

of Q from this procedure.)

of course,

If we remmd a pressure

match point Qp ~ and p~D ), we can dso check the accuracy of our -~imates
.
M
of the matrix-fracture properties and the skin factor since if all our estimates
are acc~urate, we should have

Pw~

f% %4
M s=~

****.

(41)

Note the addition of the skin factor to the Ieft-hand side of Eq. 41 is neces,. ..- ,.
. ,.. . . . . . .
,,
since the PWD soiutione of Fig. 8 assume that the skin factor is zero.

..-.

The type cxrve cf Fig. 8 can also be used ?0 estimate i ywk!ed


we have an accurate estimate of the skin factor.
same~veraue$pM&aerihd

Inthiscaee

we make the

ah0*e9 wetiK&Ap*dr

pute the ~rres@nding pwDvaiue from the foibwing equition


.
Wb #
pwD=i~ti

. . . (42)

This fixes the verticai scaie so that we can perform the type curve match

by siiding the tracing paper (Ap versus t data) oniy in the horizontal direction.
This mstching prwedure yields an edrnate

of Aas well as a lower bound

(or estimate) for Q. Using tMs estimate of A, the matrix permeability can
-

be estimated using E~ 4.
If we ti

of the accaiFaq

of tie sidn faaw

-ate,

we

-.

can still use Fig. 8 as a type curve provid~


.

.:. .

Iidius&acepi#
._,

.. .

.. .

.weiI&&diua:rD

:3hu:-effe&ive

.._ ._.

.,___ .

~.-d~

I& -. .. ..-. ~.....

...=.... . . .
__= ~_=_
.-.,=-: .::::=

:r*-=-.rwe:. :.

we use tM effeotfve wellbore

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


,..,, ,: . .. . ..

..

:.
.S.

: ~~~

-(43)

_.

c *.w.e ...

:__=

_:__

..

..

..
,.

-.

. .

.---.

-,

.:

.:.,:.

-.

.,-

:;,

. .

-,.

. .

..

1...-..).

,.

-:...

. .. . .

..

, ..-.....:

.,

-..

. . .

.. .

. . ..

...

21
If we re@aee rw by r; in the definitions

of tD and l, then the s term may


. .

~d~etdfmmE~.12,~3~d~~.

Using the effective

weUbore radius concept, the semilog type curve

matching procedure would proceed as foilows: (1) PIot @ versus t data on


tracing paper using the scales of Fig. 9.

(2) Choose a Ap value and compute

the corresponding pwD using the estimate of k~~


scale.

This fixes the venicai

(3) Perform the type curve match sliding Ap versus t data oniy in ihe

horizontal directiom

Record the value of A for the curve matched and denote


..

:-

thk value of .4 by AM. Note that here


9

12kmhm
M=

TT

r&
%}

(4) Reeord the time nwtah pshts

..

vehes, tM and (tn/a)w


b

(?

*.*.

From

(44)

the

time match p6int we have

2.637

+mcmhm(r#2 =

104 k~f

lJ

****.

(45)

***9*

(46)

(6) Estimate the matrix diffusivity by

k-

(t*/@)..

46 kobt&d

- ~ti~

matrix ~fusivity
.. .

-- -tie
..

., . . . . . ..

by -divi~

obtained

from E% 46 *&i

:f~m _E&3~. ::~ -~;we:~ve:

.__.-*

. . .-.,....-.-.;..

Eq M -by -E*4s0 lhe-~imate

mat~~-

Of --:

agree with the eitimate


&e:-d&&g~~yO:
.

obtb

~~te th~-. ...

:::::

_*_:ov&ss::g_
Ews__.*md=j.*_:~_*8gg)=f~
:S::: FMO_2_:@=:::,..:.: =:-.
..
.. ....,. .__ .. . . . ..
. _.. . .
.. . . . . ---. . . -.
. . ..
,
..+.- .....
+.. ..+,., .,. ,- .+ ..:,..,..-,
..,-. --,+7 .:-...+. -.,.., ..;... ..,.... . . .= ---- .-.,.,.--+

22
procedure yields estimates of AMend M,we must still use E% 15
to ~timate

the. skip
factor.
,.

or

Eq. 25

Analysis Based on Flow Re@mes 1 and 2


Here we consider the case where the drawdown data exhibits both
flow regimes 1 end 2. The fracture flow capacity can, of course, be computed
from either the early or intermediate

time semilog straight line (see Eqs. 17

and 23, respectively).


~

equating

Eqs. 12 and 21, we find that the early end intermediate

time straight lines intersect at t~l


. ..

.
;1

Ue f %

by

giVen

3
eY~f~

(47)

corresponds to a time t; in hours given by

If estimates of the matrix permeability and porosity are avaihbie;

(48)

for e.xampie

from care data, then we can estimate the effective porosity~empressibilitythickness product for the fracture system from Eq. 48.
The major obstacie to a successful analysis is the computation of
the skin factor.
time

straight

line

The equation for computing the skin factor from the early
data

is

given

by
.

23
Thus, to compute the skin factor using early time data reqdres an eEtim8te Qi
fracture diffusivity or of ~fh~f (k$tf can be computed via Eq. 1?).

On the

other hand, the equation for csmputing the skin factor frem the intermediate
time semilog Iine requires an estimate of the matrix properties (see Eq. 25).
We noted previously that if the intermediate

time straight line

exists, then for tD ~ ~De, pwD is uniquely determined by the values of Ad


and the skin factor s; that is, the soiution does not depend on the individual
Vt3hMS Of A

and d,

(See

Fig. 7)0

This obaervtst$onSu#geats that we em estimate Au aa WeU ae other


paramete=

using

type

curve

such as the one shown in Fig. 9.

points X denote the times at which the early and intermediate


lines. int~

Et Fig. 9, the
time straight

Note no data beyond flow regime 2 is shown in Fig. 9.

The

m-mat-~we~
belawean beusedwf~mfner
,modifications even if the .
early time semilog
straight U_nedoee not exi%
.
In the event that both the early snd intermediate
lines exist, we suggest the following procedure:

(1) Mimate

time straight
the fracture

flow capacity from either straight Iin& (2) Based on the available data,
estimate the sidn factor from the most appropriate formuia (either Eq. 49 or
Eq. 2S). (3) Plot Ap ve~
type curve (Fig. 9 ).

t data on tracing paoer using the scales of the

(4) Choose a value of AP and compute the cumswidirig

PWD f~m EqO42 @ng the estimates of k~f and s.

(Note Fig. 9 assumes

. . .

24
match points using the equation
.

,.

..

where the subscript M denotes mat ch point values.

(St))

*.*.

(Note $ ~cfhf cQuM dso be

computed directly from Eq. 47 using the veiue of Au obtained in the matching
procedure.)

(7) &timate

the matrix permeability-pormity

product from Eq. 48.

(Note that if an estimate of $m is available, we can estimate km from Eq. 48


and then estimate

U and A from Eqs, 3 and 4, respectively.)

The aeimra+ of the above pjjk


the estimate of the skin factor s.

&&s

on the a~a~y

Since the potentially unknown terms

invoived in the estjma.tion of s occur in a logarithmic


WF COmpU*tiOn of s w~ ~ sufffdtiy
the. follo*.

ch!eWng ~

of

terin,

it is Iikeiy that

If it ~ .~t, ~hm.Q~ ~f

a~t~

--

wUl indicate the inaccuracy in A

~ ~ 49 k US8dto computes, th~ the ~~:~f

~fc~f- &~.in

@ 49 shouid h in reasonable agreement with the vaiues obtained from


E* 50.

If these vaiuee ~o not agree, we should

try

a new vaiue of +~~p

recompute s from E4 42 and repeat the type curve matching procedure.


Thus, we see that our procedure can be viewed ss en iterative

one which

should converge to the correct solution.


If we have used an estimate of kmt ~ to compute s via E% 25,
then the estimate

of

+#t

curve matching pSoCM~


.

.
-.

. . .. :.:,

If

-.

obtained from Eqsa48 and 50 should agree.

_:- +@@!

_*: .*$.:to:.E_

rnay*

~ ft~ti~*_:eff&i+_

-Y.-.

~~~
:&g*:

--

_.:---_-_: :_:

*S.*--rn8t@_*_ &_
-.-_&thii&i.
r. ~ . -<.-:2.:-:.
-- ~
.=_w_-=:.
.
. ... .. .._
=
..
.
. . . ?..
., .=.. . . . ... . -. ..
,...
..4 :,... . .. . . .......... .... . -. .,,. ,. .:.:..
.-..---.:. .... -L..~-...- .- .,;:.. . . . . ... . ....... ...--.*.. ,,-,. :. ....

,=. .

:$@+@t- *m.*!+*

W*.

A@@ t~

....+-

.-. .

. ..

,.

29
wouidbe re@ecedby$/n

Since we would have to compute r& using

Eq 50.

E* 43 in order to oomplete the type curve procedure, the effective

weiibore

radius concept offers no advantage in this case.


If the eariy time semilog straight line does not exist, the preceding
type curve matching procedure can still be used.

In this case, we would have

to obtain the fracture ffow capacity and skin factor estimates

Moreover, we would have to estimate the

mediate time sexnilog straight Iim

matrix permeebiiity=pewoeity product (step ?)


;

= 1.58
b

usfng the inte~

e
10= (I$f~+2U
km$ ~Cm

using
the following

equation

(s1) - --

51 is derived from E% 2S.

Note that Eq

BUILDUP BEHAVIOR
We define the dimensioniees shut-in pressure, ~~ by

P~~ = .

kfhf (Pi - ?WJ


141.24SU

.*

*...

(52)

when pw @ the shut-in presmme h psL The baeie buildup equation is weii
known (see Refs. 11 and 12) and is given by

..
.

PWD
[(t+-M)n-] - Pwn(Atnl

whei? mpreeenmme.. *1*


-t~w

sohtth

k uiiiqueiy- deWFrnined

-,

-N?!q---m?:eww!x:x: .W!!!!!!!!i!-.

by=the flQw:regimes

. .
that are repwsentad

_:

in...:. . ;-:

2s
advantageous to know the flOWre@me that OC=UFS
at the end of the production
period, and this could be determined if drawdown data is available.
in

O&
analysis

that t~

PWD[6tDl refkts

We assume

b@3LIpbehaviorrefieets
only ttio ~OW N&=,

that is.

oniy flec~regimes 1 and 2, or flow regimes 2 ar~d 3. In the

unlikely event that PWD[AtD] reflects all three flow regimes, we can, of cwrse,
cgmbine the anaiysis to enhance the reliability of our computations.
The meet commonly used method of anaiyzing buildup data is due to
Horner Is.

Here we consider the buildup behavior at a well for the various pos-

sibilities that can occur based on our drawdown solutions


.----

. .

Flow Regime 3 Exists at Shut-in


First we consid~ the csse where flow regime 3 occxm prior to shutin and pwD [Atnl rei3ects either flow regime 2 or 3.
Q

In this case we can use

21 to obtain pwD ((At)D) for small vaiues of atn and thus the buildup

QDs

=1.1s1 {
Iog--(t
[ A%]+ ~c125 +

1)2

& E3(1A~ut)gI
} -

..

****

(54)

For @@r VahIeS of AtDs PWO h tD] can be obtained from E4 13 and then we
obtain

Ds

Equation.34

indicates

t+At
= 1.151 Iog (~)

that a semibg plot. of .Pm V-

(t + At)/.~

.w.&

(53)

. .

..

l.

27
If the producing time is

large

so

that t +

At

= t, we can replace

E% 54 by

Ds

t A;t ) +
0.5756 log (
{

I09J tD

0.125 +

~0~ [-j}

(56)

Figure 10 presents a dimensionless Horner plot for the case A = 104 and J =
.
-103 for various vaiues of producing time: tu
The time {Dg where the late
time straight line begins on the drawdown solution is indicated on Fig. 10.

.. .

The semilog iine ot siope 0.5756 occurs between the points marked ~ for each
tn cumm
,..

Note the vaiidity of @ 56 is indkated

semilog straight line ef..~.

0.5753.

by the existence of the

A&o mte that .Fig. 10 indioatea that we . .-

..

need approximately one Iog cycie of flow regime 3 data during the drawdown
period in order to observe flow regime 3 (line of siope 1.151) on the Horner
plot.

Other computation not shown indicate that this latter observation is

valid.
&uming

that Iog (t + &) = log t we find that the two semilog

straight lines, Eqs. 55 and 56, intersect at a reai time ~t* in houm given by
.-

332.3 + ~Cmyh:

At*

-.,

...

,-...

siope m* corresponding to ~
of slope m ~
drawdown CUL
. .

(57)

..-

1bus,if a Ho~er piot of pm vemis

. .

.* .***

k
-m

(t +

&)/k exhibits the stfaight Une of

36 and aiso exhibits the sernilog straight line

to lZq.5& we -cam c?m~ete


Fht,

of ~either Mvsigiit ~;

,.

the analyiis aS -in the

we compute the fracture flow ~a@y


%60ndly, -we-~matii

from .the slope


:
&. rndrix ~p&Fmeability &m
-- ~ 7.

28
Eq 57 provided an estimate of the matrix porosity is available.

We

then

estimate the skin factor from one of the following equations:


(k+@2
(

- Pwf
1.152 %hr
~
m
{

s =

+ 3.729 ,

- Iog[

-Iog[

(58)

km4mc~r~

J+ 3.23\.

(59)

It is important to realize that in Eq. 58 m* is given by Eq 23 and is the siope


of the intermediate

time straight line on the Horner plot, and plw ia the

extrapolation of this straight line to At = 1 hr.


59 correspond

to

Similariy the parameters

in

the Iate time straight I.ine on the Homer pioL

Our computations indicate that the line of slope 0.37S6 begins on


. ..

the Homer plot at

tDb

=5X102
l@

.*

.***

(60)

,./

Note that the formula for A~Db is the same as the formula for A~Db, Eq. 26.
This result is not surprising since we should expect to see flow regime 2 on
the Horner piot as soon as pwD Ut~] represents flow regime 2. Thus if Lb
-p

is the red

time corresponding :0 ~{~b, we can still estimate the fracture


n

porosity-compressibility-thickness

product using E% 37 provided we replace


.-

=*

4&

Our analogy with drawdown case is now compie~,


qe the prooedw
.--

@sssgd

for
.-.

&a@oyfp

_3y@a
.- .-

Howevdr, we should note that M we. yish to. w


,.

to

.-w
.....

b
.

that is, we can


@ill

@ta

.... ...

the type curve. matching

29
we shouid plot pm - pti~
ma@mum value of At used) satisfies
ments are discussed by Raghavan 14

versus At and require that 4tm=

&m=

<

t/10.

The basis for

thee

(the
requir~

Atmu

C<t, we can of course use a WIIer-Dyes-Hutchinson 15

(MDH) type plot, that is, we can plot pw~ versus At snd again use the same
to analyze the buildup

procedures

data.

We now consider the case where

flow

regime

but pwD [AtD] reflects only flow regunes 1 and 2.


flow regime 1, we have

Ds
,..

Orquivaiemiy

at shut-in

3 mists

When pwD [AtD] refk!ts

. 1*151 [lq (+)

- log (1 + $))1,

(61)

..

Pi - Pws =m[log(-)-kg(l+

d)]

. . . . ..

(62)

Equation 62 suggests that if pi is known we can cempute Q by extra@atin~


this semilog straight line to (t + At)/& = 1.

Letting Pws,e denote the- corres-.


pending e.xtrapolat@ pressure, ~e equation for u is given by

The fractti

flow ciipacity CG be comput&3 from the siope m.

for the skin factor using the eeriy time data, E+ 62 is

s =1.isl[
..

. .

. . . .

.. . . . ..
.

,-.

..

Pl~ - P~4
m .
-.
. ..

.~+(+):
.,.,..- +$ep.ew

@?8Xi

(63)

The equation

by

~g~] ,.,

(,~j

-..-*. ~, ..*:,,:., .-...-

30
The skin factor can of course be estimated from the intermediate

time data

(slope m? using E* 38 but this requires that estimates of the matrix parameters be availabie.

Again, we can use the drawdown type curve (Fig. 9) to

complete the anaiysis provided &mu


If an accurate estimate

c t/10.
initiai pressure is available, we can use the

of

following procedure which is based on having available an estimate of matrix


porosity.

(1) Compute J from E~ 63. (2) Estimate +~~,

using

estimate *

(65)

**V*.

(4) Cempute

(5) Compute k from the obvious rearrangement

of E4 60.

(6) Using I@ 4, estimate kmhm.


I%epswedhg
3 exists

at shut-k

pmwdureisbesedon

the knowmdge that flow reghe

If the drawdown data necessary to make this determination

is not available, we propose two ways to determine if pwD (tD) represents flow
regime 3. The fim

way

to compute PWD (tD) and tn based on the results of

is

analysis and see if this point corresponds to flow regime 3 on the appropriate
The second way is to compute the skin factor

A-d curve, (see Figs. 2-5).

using Eq. 58 and cheek to see if the two estimates of s agree.


Flow Regime 2 Exists at Shut-in
In this section we assume that flow regime 2 exists at shut-in and
\

..

flow regime 1, the dimensionless shut+

.. . -.

.-

.-

.-

.. .

. . .. . .

. .=.-.

.,.

.-

.. .

. ..

31
~e

isgiven

by
~(t

pD~ = 1e151{ iOg[ ~]

For

At)
- 0e125 - 1/2

10g (~)}

c c

(66)

Atn <<tD, the foilowing equation is approximately satisfied

Using Q. 67 in E% 66 and rearranging

log~+
Ds = 1.151 { -

logy-(t

indieatae that for smail wiuae ef ~


of dope L151 on a dimensions

~*125

+ At)

- 1/2

log

+37
+

(68)

we shouid see a sefni.log

Horner pioL

When PWD &tD] represents flow regime 2, the dimensionless shut-in


pressure is given by

Ds

= 0S756 iog ~)

8****.

(69)

In this case a dimensionless Horner piot shouid exhibit a semilog straight iine
of Siope 0.37s6.

Figure 11 dispiays a dimensionless Horner piot for three vaiues of


j?OdUCi~

time

for ~

Aarid ~ flow-regime

&ee # = 104 sridw = 203: For these V8iWS of --2 exists

for

tn in the range

$ x 108 ~ t~ : s x 1011,

32

and the dimensionless Horner plot exhibits a straight line of siope approximately equal
to 1.151 for these same vaiues of AL Also note that the siope of Q.5756.is not well
defined for the case tn = 109. This indicates that we need. at f~ast me log cycie of
drawdown data during flow re@me 2 in orde~ to see ffow regime 2 on the Horner plot.
Also note that as expeoted, the semihg straight line of siope
D = S x 108, (see Eq. 26).
The above results indicate that on a Horner piot
(t + At)/At) we ti

begins

0.S756

of (pW

at

versus

see two Semilog straight lines provided the drawdown data

reflects at Ieest one Iog cyeie of flow regime 2. The first (snail vaiues of
At) straight Iine will have siope m and the second will have siope m*.
fracture tlow capacity can be caictdated from either straight lin~

The

Extrapolat-

ing the straight line of siope m to plw we can estimate the skin factor by

3.23] .

. . .. .. . . (70)

Alternatively we can estimate the skin factor using Eq. 58.


The analogy with the &awdown case ia now complet~. However,
to use the drawdown type curves in the anaiysis procedure, we must again
plot pm - pwf~ Veins At and require that

At ~=

10t.

Buildup Examole
Here we consider the analysis of buildup data for a naturaily fra--

tured limestone

reservoir. The data used is given in NajurietaS.

Sit&S ..??ajurietafs.dam-eofivertedtfioil fia,ldunf~


iQt4@reta-tion ye ~ee@

iS @eeonabie -.

-We em-

on .OW.theiketicd

Tabie 1
tit

*-

re@ts but is

not n~y
the @y...-.reea4@3ie
However,
. . ..ht~tg~o~
..- .-.
. . . ...
. - wehave ilot~
---- . . .. .----several sets. of data h fractured. reeermimthat agree with the interpre$atJon

-.

33
given her%

These inciude oil, gas and geethermai


Figure 12

SilOWS

a Horner

piot

of

we&

the buildup data of TaMe 1.

For

a matrix permeability on the order of 0.1 md, we expect the fracture permeability to be so large that oniy flow regime 2 and flow regime 3 will be apparent
on the Horner pioh
of two.

Thus, we searched for two straight Mae with a slope ratio

The two Iines we choose are shown in Fig. 12.

Note that there is a

transitional period of approximately one log oyoie between flow regimes


3 as our

theory predicts,

..

and

We now proceed with the com~utatiom


.

Using Eq. 24 we have

k~f = 81.29 (2S92.2 STB/D) (2.3 RJ3/STB)(1.0 C@ ;


17.5 Psiilog cycie

that

is kpf

= 27694.8 md-fL

The iktesrndte

&d Itketime straight &as i&&eot

Thus usingthe obvious reammgement

km =

that k
,-

..

S32.3

(0.21)

of &

(8.1633

at it* = 2.83 ho~

S7, we find

x 104

gsi-l) (1.0 CQ)(17.062 ft2);

km = 0.094 md.

Computing the skin faotor from Eq. W, we find

-97094.0 md+t
(0.22) .(8.16S x 10* psi-~) (la~ cp) (0.37S2
.. . .;
.L ... .

-Iog[
,.,

1
.ftz).(17.06 .ft).

34
that i% S = 2.624.
(Computing the factor using Eq. 58

@feS

S =

2.628).

Using Eq. 4, we have

12 (0.094 md) (17.06 ft)


A?=

(0.3752 ft2)
(17.062

ft2)

that *,
From Fig. 12, we estimate that flow regime 2 begins at (t + At)/At
= 1.0S x 103; that is, Atb = 0.082 houm
the porosity-compressibility-thickness
..

Thus using Eq. 37, we can estimate

product for the fracture system.


.

We find

. .

%cA,=?.262x1O

+ /(27694.6 m&ft)(O.2l)(8.l655tiO-@-i)(l7.O6ft)(3.36tiO-7)(O.O82hr)

.-

Locp)

that

(0.3754

ft~)

is

$fw?

= 2.a9 x 10-7 psi-~. ft.

Finally from E% 3, we find

, - (0.21) (8.$153 x 104 !?si-i) (17.06 ft)


6J ( 2.89 x 10-7 psi-~ - ft )

thit k; d =
.. 106.

.-

The results obtained by N@rietaS

&Jl x -107,
km4

-.
.,. ,
... . .

..

. ..

&12@
rnd,

are kpf = 30308.4 md-ft, s = 3.61, Y


2242;

~i2S@ f@rt@

ViihiC

of-ti, -o&

--

.
*
--- ..
;

. .

3s

results are in good agreement

with those obtained by t?ajuriet~

The discrepancy

in the U? values could be e.xpi.ained if we knew that the first two data points on
Fig. 12 are influenced by wellbore storage and skin ss Najurieta seems to suggest.
However, the standard lo@g
bore storage

effect&

piot of AP = p,~s - pwf,~ versus At shows no weU-

Moreover, to

obtain

2242, we would have to have

flow regime 2 begin at Atb <1.0 seoonds, and for the values of k and w
obtained by Najurieta, there wouid be no early time straight line of slope m,
(see F@ 4).
.
-.

CONCLUSIONS
In this work we have established that the pressure response in a
naturdiy

fractured reservoir may follow three distinct flow regimes.

Although

we have considered only the modei shown in Fig. 1, the flow regimes discussed
here are evident if the matrix is assumed to emaist of sphericai4 or cyiindricai8
block

For these clbses, the existence of the three flow regimes can be

identified directly from the soiutions presented in Ref. 9.


methods to identify these flow regimes on fieid data
phy&icd interpretation

We have presented

We have given a

of each flow r&fme and have detieated

periods ch.rringwhich 9ac5 flow rqlme

exkts.

the

time

Under the assumption that an

estimate of matrix permeability is available the foiiowing conclusions are


warrante&

(1) Using data from flow regime 2 and flow regime 3, the fra-

ture flow capac~tyy matrix ~eability,


coefficient

can

be

eetimated

skin Pactor Ad @M@O~eSS f~~t~

A Iower bound for the dimensioniees matrix

etorativity can also be obtaimi


(2) Using data from fXowreffimes 1 and 2, aU matrix and fracture

properties

can be estimated by using a combination of semilog and type curve matching


proced~

..-

NOMENCLATURE

A=
B=
c
h=

ira.insge area, m2 (ft2)


formation volume factor, RB/STB, (res m3hock-tank
= compressibility, PSI.-1 (Wa-l)
reservoir thickness, feet (m)

f
hm

fracture height, feet (m)


=
= matrix height, feet (m)

= Ipermeability, md (umz)

KO(~) = Modified Bessels Function

of

second kind, zero order

K1(X) = Modified Bees&s Function of second kind, first


.,

m=
m=

m3)

OrdtW

sernilog straight line slope (E% 16)* peiflog cyeie (kPa/log c@e)
,.
= semikg straight line siope (E4 23), @/log cycie (kPa/log oyeie)
ressure, psi (kPa)
1?
flowixg wellbore pressure, PA (kPa)
wellbore pressure at shut-in, psi (kPa)
shut-in weilbore pressure, psi (kPa)
dimensionless pressure
dimensionless wellbore preawre
production rate, STWD (stock-tank m3/hr)
radfai distance, ft (m); raw
dimen@Onlue__r@M-..

E@g:.ft:(.rn)
.

..

INWM+Oiriii-dhs, ft (m)
w@l

~~te

--

3a
AA6

bte$t
t*,ti

*Aa

beginning or ending time of a semilog straight

intersection

time, hrs

dimendoniees time based on fracture

properties

dimensionless time

DA

dimensionless time based on drainage area

DbtDetD

line, hrs

dimensionless beginning or ending time of a semilog


straight ILle
dimensionless intersection time

= Lapiace transform variable

= vertical coordinate, ft (m)

= dimensionless vertical distance

Supem?ripts and subscripts


f

fracture

initiai

-m

nlwrf%

match point

Lapiace Transform

Greek
viscosity, cp (PM)
Euk.rs oonstant, 0.57722
P@rQ$W.fl%ction
defined by Eq. 6
dimensionkss fracture transfer coefficient
-1
~@.

E?&.s

..

..

-..

-.

-.

dimensionless matrix storativity


.

.-

.-

...

.
.-

..

..

... -.,. . . .. ....-

.
. .

.-l
. ... . . . . ... -,,
.

. .. . . . .. ..,.

..

,.

. .

REFERENCES
1.

Werren, 3. IL and Root, P. J.: %e Behavior of Naturally Fractured


Reservoirs, SPEJ (September, 1963), 245-255.

2.

Odeh, & ~:
SPEJ (hfWCh,

3.

Kazemi, H.: Pressure Transient Analysis of NaturaLly Fractured Reservoirs,m


m,
MME (1969), 256, 4S1461.

4.

deSwaru@., .Q &@tical Soiutions for Determini~ Naturally Fractured


Reservoir Properties by Well Testing,m
0SPEJ (June, 1976), 117-122.
~@fet& H. ~: *A Thiitiry fti- I%siure l?ransieftt Analysis: in Naturally
Fraotured Reservoira,ff J. Peh Tech= (JdY, 1980), 1241-12S0.

$.

6.

Wnste@y-State Behavior of Naturally Fractured Reservoirs,n


196S, 60-64.

Kucuk, F. and Sawyer, W. K.: Transient Flow in Naturally Fractured


Reservoirs and Its Application to Devonian Gas shake, Riper SPE 9397
presented at the SPE Wth Annual Fail Technicai Conference and Exhibition, Dallas, Texas, September 21-24, 1980.
.

7*

Adam& JL R-, Ramey, H. J., Jr., and Burgess, IL J.: Gas Well Test@g
in. a Fractured Carbonate Reservoir,q ~. &L Tec~ (October, 1968), 1187-

1194; Trane. AIME, 234e


8.

Stehfast, H.: Numerical Inversion of Laolace Transforms, WCommunications


of the ACM (January, 1970), 13, No. 1, i7+9=

9.

Raghavan, & and Ohaeri, C.; 7hsteady Flow to a Well Produced at a


Constant Pressura in a Fractured Reservoir, Paper SPE 9902 presented
at Wt Annual SPE California Regional kleeting~ Bakemfield, California,
March 2S-27, 1981.

10.

Brom F. and LMiUer,W. C.: .4 Simple Method for Comcti.ng Sgot Pressure Readings,q & ~.
Tech. (August, 1961), 803-80S; Trans., .&ME, 222.

u.

Matthews, C. ~ and RueselL D. G.s Pressure Buildup and Flow Tests in


Wells,w Monograph Series, Society of Petroleum Engineers
of AIIW
Dallas

.-

(1967), L
.

i2.

.-

1%

Horner-.D. R.: -_.!!Pmssure2hiiid4P h.-@iS,w:

:~w~

c-;

-------

~--H~e(3g51)~*

q-,5a3+230

!!@!@ ?e~~ie~
.=

. ..... .

.. . _ ,........

14.

15.

%ffect of Producing Time on Type Curve &alysis, ~.


Tech. (June, 1980), 1053-1064.

Raghavan, IL:

&

LMiUer,C. C., Dyee, & B., end Hutchinson, C. A,, Jr.: The Estimation
of Permee&ility and Reservoir Pressure from Bottom-Hoie Preseure Buildup Characteristics,n Trans., AIME (1950), 189, 91-104.

.-.

..

.. .

.... . .

. .. ..

APPENDIX A
ANALYTICAL SOLUTION AND APPROXIMATIONS
By symmetry, we can formulate the relevant initial boundary vaiue
~ problem on one of the repetitive

elements shown in Fig, L

We assume aiI

fluid production is horn the fracture system and let q denote the rate at which
fluid is produced from one of the horizontal fractures which comprise our
mhtrix-fracture
the fract~e

modeL

We Iet pDf denote the dimensionless pressure &cp in

and PDm denote the dimensionless pressqre drop

We assume flow in the matrix is one-dimensionw


.snd that we have radiai flow. in the fracture.

in

the fracture.

that is, in the z+lirection,

Thus, the matrix-fracture

model

we consider is identicai to one that was originally proposed by deSwann-Q.4


. .. .

We let z denote the vertical coordinate and let z = O. denote the top of. the . .
repetitive

eiement

of

Fig. L

We define the dimensionless distance, ~s

D = z/ti:/2)

bY

(A-1)

The initial boundary value probiem which represents flow in the matrix can
then

be

written

..>

..,. . . . .

-k-

as

.2

----:--

a pDm

~.
a%..

.-,

3 Ut
=,.~
.

?..

. ..**m,*.**

(A-2)

. ...-

pD~l=o,
tD=Q

(A-3)

-:

= pDf

Dm I

. . . . . .

(A-4)

(A-3)

%=1
Note that Eq. A-5 simply requires that the matrix and fracture pressure are
9
equai at the matrix fracture interface.
We let FDf and ~Dm Cfe310te, Respectively, the La@ace transform of pDf and pDm with respect to tD, and we let
u denote the L@ace transform veriable.

Using La@ace transforms, we find

that the analytical solution of the matrix pmbiem, Eqs. A-2-.4-5 is given by

PDm = pDf
. . . . . ..

..

. . ..

Cosh(m

(A-6)

problem describing flow in the fracture is

The initial boundary

given by

*Df
-L (rD~)aro
D

@Dm
=-a~Df
1
D
% ZD=I Ot

(A-7)

(A-8)

D=fl
. .

..

-..

(A-9!

..

...-.

.-.

,-

...

. . .. . .
... .

..

.. .

. ..

43

hpnf=o

The soiution of the fracture

(A-1o)

initial boundary value problem (E% A-7-A-IO) is

given in L@ace space by

(A-H-}

**9**

&

where

f(u) = 1 +

Q=n
..

t-

..

(In deriving Eq A-U, we have used Eq. A+

pressures.)

(A-12)

The dimensionless weilbore pressure drop in L8piace space can be

obtained by evaluating Eq. A-H

at rn = 1, and

The solutions presented

..

the Stehfest algorithm.7


F= smell veluea af u,

. . . ..... ... . . .-

..

. . .. .

.,, ....

. . .:-, .~- ~1

(-.+~)

(A-13)

by inverting Eq. A-13 using

in this study were

KJmm

@ven by

..

to relate the matrix and fracture

--

-I

In
+jq (

=~
/rti,

.-$-:

.,,

..

(+14)

(A-15)

-.

:.:_:[

I
.!

and

f(u)= l+w

(A-1$)

(Xote the approximation of Eq. A-16 uses the fact tanh(x) u x fcw small
vaiues of x.)
Using the approximatio~

given by Eqs. A-14-A-16 in Eq. A-13 gives

e2yu(l +
[1
4

i5w&+rl

The long time approximate solution

w1)

obtained

b.,

(A-1?)

inverting Eq. A-1? and is

given by

PWD = + [h

tn

0.80907 -In (1 + w)]

Equation

13 can be o&tained from ~

(A-18)

...

. .

A-M by adding the skin factor s

converting from natural Logarithms to common logarithms (log bese 10).


Since tanh(x) rapidly approaches the value of one 8s x increases, we
can ako approximate f(u) by

f(u) =&

(A-19)

Using Eqs. A-14, A-IS and A-19 in Eq, A-13 @S


.,,

8****

(A-20)

...-.

.. ,. . ...- :: -.-.

. ....

. . . . . .-..,.:

.. . .

-.

,..

.. . . . .

. ...

45
Inverting Eq. A-20, we

obtain

the foilowing intermediate

time approximation:

(A-21)

By adding the skin factor s to Eq. A-21 and converting to common logarithms,
we obtain Eq. 21.

,,.

.-

,.

..

. .

:.

,_..

,_

. . ..
.

ACKNOWLEDGEMENTS
This work waa partiaUy supported by funds provided by the Department of Energy, Morgantown Energy Technology Center (METC), Centract No.
DE AC21-80MC14645.

..

,:

,.-,

,.->

..

-,.

----

..,

hf

FRACTURE
MATIWC
RE$SiTllVE
ELEMENT
}

h*

.,

.
.

,,. . . . ...

. -.
.

.,. . . ..--:-. F&:I-

..

..&&&;~~+m--Res&=voiFivitl----
.
. ...

:-2--

..

. ..
.,

I
8

.-

s&o$E=
MM

z%
-

~~

m3

-6
= 10
-9
= 10

2
DIMENS1ONLESS TIME, to

Duration of

,.
-.

-.
.

. ... ..

..

. .

12

,:.

..

Fg.

.
..-

. ...

. ..

:-<.

. .. ,.

3- .ifi~t Of .1 m.W .Eciiie.mc*


ad Qurition
d
how ia@Iiaer.
al?802
.

.-.

. .......

::.

,.

..-

. .

. .._

-..

~.

.,.

V-

..,-

.
.

--

.,

SLCWE=L151~

&-

0 9
a

Id

5
m
U3

IA

a
&

QJJ8 &

$80

. .. .

..

.-

H&
.

4- Effeot of Aon ~

_~q*

DUr@on of .-

Flow Re@aes% *..= 111~


.
..
. .. ::. .. . ......
.

,,.

.
,
..

....

w
e

It

u)
*

-3

IQ

..

-w%

-.

. .. . ..

. .
..

..

...

.
>

..

?a
220

a-a

..

-43us

-p

---...

.. .
.. .

.. .

.. . . ..

3 -3

am

4 -4

od

-..

..

-.

-.

. . .

.. .
. .

. ...

-..

h
\

-3

L
.

f-

\\.

\\\\\

y\

-3

-4

\\=\

\=v

. .

..

..
.

. .. . .

..-. . . .. .
..

..

. ..

.:
..

..

.
,-

-..

. . .

.. .

. .

1#

..

\-

- ....

.. .

. -

I!i

-a

.
.
.

!3
x
U)

.
.

SLOPE=1.151
0

HORNERTIME RATIO,(t +At)/At

,.
.!

1,

{
,

. ..

..

0
.-

It

/-~~~
,, .. . .. . ... .

-% -3

,. .,.

. . . . . --

..

. ..

..

.,
. .,.

.,

ii

1
e4

N
a

.
.
.

. ..

TABLE 1

NAJURIETA BUILDUP EXAMPLE, NATURALLY FRACTURED RESERVOIR

&

Reservoir, Weil and Fluid Property Data

o.37s

Weilbore Radiue, rw, feet


Approximate Matrix Permeability,

km, md

Approximate Matrix Porosity, ?m

0.21

ThiCkTlt25S,
hm, feet

17.08

~8tFiX

~awty,s,~.
Initial

Praesure,

Compressibilities,

,. ,_
pi, psi
cm = cf, psi-l

2.3

Formation Voiume Factor, Bo, RB/STB

0.1

,.

10
6789.S3
8.1655 x 104

&

Buildup Data
Production Rate, q, STB/D

2592.2

Producing Time, t, hours

8611.1

Pressure at Shut-in, pwf~, psi

At, hours

6352.27

t+At
T

-3
2.7a x III

3.10 x 106

6616,73

1.87 x 10-3

5.17 x 105

663S.69

3.33 x 10-2

2.58 x Io;

6643.62

6-67 X 10-1

1.29 x 105

6650.09

1.33 x 10-1

6.46 x 10!

6653.61

2*S7 x 10-1
5.33 x 10-i

3.23 X 104

6660.96

1.62 x 104

6666.11

1.067

8.074 x 103

6669.34

2.133

4.038 X 103

6678.01

4.267

2.019 X 103

6685.06

8.533

1.010 x 103

6696.83

17.067

S.056 X 102

6704.46

34.133

2.533 X 102

6711.52

También podría gustarte