You are on page 1of 6

Analytic Solutions for Determining Naturally Fractured Reservoir Properties by Well Testing

A. de SWAAN O.

..

lNSTITfJTO MEXICANO DEL PETR6L50, MEXICO CITY, MEXICO

EXPLOTACION

ABSTRACT
A completely unsteady-state theory is developed that describes the weil pressure response of infinite, naturally fractured reservoirs of the type composed by high -permeability fractures and tight matrix blocks. The theory exclusively involves flow properties and dimensions of the fractures and the matrix blocks; no extra adjusting parameters are needed for predicting the pressure response of a reserrmir Comparison of compu[ed with known P tuperties. pressare-draudown theoretical curves with results of a numerical model applied to idealized reservoirs shows a good fit. It is concluded that an evaluation of the following reservoir properties is possible {mm well-test pressure plots: the fractures kb product and the 01 the matrix porosity by a average product characteristic dimension of the matrix blocks. Thus, the model allows comparison of well-test pressure response with the formation measurements.

the actual shape, dimensions, and fluid flow properties of the reservoir. Furthermore, these two parameters have been found of limited value for determining some reservoir properties. 3 Warren and Rootz give a comprehensive bibliography of the instrumental evaluation of fractured. reservoirs, A promising method to this end, using sonic and resistivity logs, has been presented by Aguilera.4 The need to directly relate the fractured reservoir properties to the well-test plots, just as reservoirs, is obvious. in the homogeneous THEORY The theory, detaiIed in ti]e Appendix, involves the following assumed mechanisms of the fluid flow and their corresponding mathematical expressions. I. At early times of the well test, the flow takes place only in the fractures and is described by the approximate solution of the radial infinite reservoirs as applied to the fracture medium:

qw P
Pf= 4*hfkf

In (

d~f+)

.-.(1)

INTRODUCTION Barenblatt and Zheltov 1 were first to present the formulation of the problem of radial flow of a slightly compressible fluid in a naturally fractured porous medium. They assumed that flow occurs only in the fracture porous medium, in which matrix blocks of contrasting physical properties deliver their fluid That is, the matrix blocks act as a contents, uniformly distributed source in a fracture medium. In their paper, Barenblatt and Zheltov assume this sources strength is proportional to the local pressure loss at the fracture and to an average internal block pressure. An approximate solution to this problem was presented by Warren and Root, 2 resulting in a characterization of the fractured reservoir by two parameters ambiguously related to

~r;

Original manuscript received in Seciety of Petroleum Engineers office Jan, 27, 1975. Paper accepted for publication Dec. 22, 1975. Revised manuscript received Feb. 19, 1976. Paper (SPE 5346) waa first presented at the SPE-AIME 45th Armuai California Regional Meeting, heid in Ventura, April 2-4, 1975, @ Copyright 1976 American Institute of Mining, Metallurgical, and Petroieum Engineers, Inc. I Re femnces given at end of PaPer. This paper wiil be included in the 1976 1ransactiorrs volume. JUNE, 1976
q

for the first part of the This equation accounts fractured-reservoir typical pressure response that, in a semilog plot of pressure vs time, starts as a straight line and then shifts to another parallel straight line (see Figs. 1 and 2). 2. As mentioned above, the matrix blocks act as a uniformly distributed source in the fractured medium. The effect of this source is evident after a certain time lag, since the response of the matrix blocks is slower than the fracture-medium response because of the blocks tight nature. 3. Assuming that the shape of the matrix blocks may be approximated by regular solids, their internal and hence the flow through pressure distribution, their surfaces, is a known function found in the theory of heat fIow in solids (see Ref. 6). These solutions are given for solids with unitary pressure (temperature) loss at their surfaces as a boundary condition. 4. Considering that the pressure in the fractures surrounding the regular matrix blocks is variable the outflow from the blocks is described through a convolution:
117

. . . ..

. ..

. .

-2 qm. (Apf, t) =
. ...*. Amahf . .

t r?Apf f
o 61T . . . . .

TABLE

1-

IDEALIZED case 1 Matrix 0.05 10-5

RESERVOIRS

PROPERTIES11 caee3 Matrix -= 10-3 SS2.3 149,352

(t-r) m . . . . .

dr ~

Fraotures

Fractures 0.45 5.521 3.551 x 105 0.762

.(2)

for the flow caused where qumd is the expression by a unitary pressr,we loss at the surface. 5. This injection term in the continuity equation in the fracture medium,

+ k ; P c

0.45 7.2SS39 1,094 x 0.762

105

1.s6
275.S44

1. 14.7 x 10-5

0.235 14.7 x lo+

rw

11.4s
166.52

11.43
3ss.

=
t3r2

f32Apf

t dApf 1 dApf 2p . ma(t-r)dr J .l qf E3t kf Amahf o dr q

Elw

. . . . . . . . . . . . . . . . . . . . .(3) results in solutions similar modified hydraulic diffusivity to Eq. 1, but constant, with a

1 q,, = 1/(+ Tf
for infinite and slabs

kmahme 1 ),
kf hf ~ln.

(4)
(see Fig. 3),

separated

by fractu~es

where is the average value of the product of a characteristic dimension Xma (which is either the slabs thickness or two-thirds of the blocks equivalent spherical radius) and the corresponding porosity. Thus, the behavior of a fractured reservoir, for 1arge times, is equivalent to that of a homogeneous reservoir with a hydratdic diffusivity constant that depends on the fractures and blocks properties. RESULTS

2 km. rma
~ SP
= 1/(++

1
), .
q

J5)

kf

hf

~m.

for the case of matrix blocks whose pressure behavior may be approximated by that of spherical of blocks of blocks. For the case of a distribution either shape with varying dimensions and porosities, the hydraulic diffusivity constant is

1 v c omp = 1/(+
~f

++
ff

Xma @ma)

, . .(6)

4000 . 3s00 aaoo $70o ~ W !2 i 3600 a600 [ c \ . \ 4 \\A \~ . .A A . .

The compariscm of the results for the infinite-slabs case (Eqs. 1 and 4) with numerical simulations of ideaIi zed reservoirs presented by Kazemi 10 is shown in Figs. 1 and 2. The data of the idealized reservoirs are shown in Table 1. When plotting the theoretical results in Figs. I and 2, the straight line for the early pressure response, Eq. I, was drawn as a full line up to the time when, the large-time solution (that is, Eq. 1 with the modified diffusivity constant), Eq. 4, starts. From the corresponding value of the pressure thereafter, the large-time solution was plotted as a full line. The slope of the lines depends only on the kb product and the production rate. A fractures practical point to use in the plots is the time at which the pressure response starts; this is the

:1 ;-y::,., :Aj
TIME, HOURS

hf

FIG. 1 PRESSURE
woo

DRAWDOWN FOR CASE 1.

4W0 c1 ii a

I
FIG. 3 HORIZONTALLY
SOCIiTY

repetitive element

4S00 4700 b
4000 --4600 THIS WOW ,0.0

.. -.. *.
KAZEIAl NUIAERICAL MOOEL1l

% -.% . %
--b -a %

10.?

to-~ TIME,

10-1

10

low

Houffs

FIG. 2 PRESSURE
11s

DRAWDOWN FOR CASE 3,

FRACTURED

RESERVOIR.

OF PETROLEUM ENGINEERS JOURNAL

. .

_.

inverse of the factor of t in Eq. 1, with the adequate diffusivity constant value. The theory presented here does not account for the transition from only fracture flow to fracture flow plus block contribution. Nevertheless, the values of the properties of interest intervene in the straight-line portions of the well pressure response. COMMENTS AND CONCLUSIONS The well test is a check of or a guide to a From this . point of view, the geological model. usefulness of the equations obtained is obvious. A slide-rule computation of the group of properties involved in the equations and its plotting over the well-pressure graph test the validity of the geological model. Compared with this method, the Warren and Root 2 procedure is less convenient; after obtaining two adjusting parameters without direct physical a third parameter, a shape factor, is meaning, needed to arrive at the probable values of the reseivoir properties. The advantage of the present theory over making a computer run of a numerical simulatorl for every assumed set of values of the reservoir properties is also evident. Whether the slabs or the spheres solution applies in a particular reservoir depends entirely on the geological evidence supporting either model. Notice that the resuIts may be extended to heterogeneous reservoirs with tight zones that are equivalent to matrix blocks ,embedded in an interconnected hightramsrnissibility medium, which is equivalent to a macrofracture grid. It is anticipated that discrete distributions of Xma+ma obtained by we fl-logging and/or core-sample measurements and Eq. 4 may result in computed pressure responses matching well-test curves in reservoirs. Since crossflow is composite fractured assumed, this model should be applied to reservoirs composed of Iense-shaped blocks of varying properties and not to actually layered reservoirs with no other connection between strata but the producing well. The early-time solution, Eq. 1, can be considered only as an asymptotic solution at zero time, since :he matrix blocks near the hole may contribute to the flow from the beginning of tk ~ drawdown. This complement to the theory was formulated becausti approximations for short and medium times of the Eqs. A-6 and A-12, outflow equations, blocks result in transformed equations of the solutions, Eq. A-5, that do not lend themselves to a symmetric Laplace transform. (The Barenblatt and Zheltov 1 formulation, which results in a description of the pressure response in the Warren and Root paper2 at short and medium times, is easily seen to be equivalent with assuming the blocks outflow in the form of a negative time exponential. This is a fortunate choice, being similar in. shape to portions Eq. of the actual regular blocks outflow functions, A-6 or Eq. A-12, and amenable to a symmetric transformation; however, there is no Laplace straightforward way to rei ate the exponential S
JUNE, 1976

decay constant to the actual blocks and fractures flow parameters, which results in the necessity of adjusting constants.) The following conclusions have been derived. 1. A mathematical model of the well pressure response that is caused by slightIy compressible fluid flow in a naturally fractured reservoir composed of high-permeability fractures and tight matrix blocks has been developed. The model S predictions and results of the numericaI simulation of the well pressure response of idealized reservoirs show a good fit. 2. The model character zes the fractured reservoir exclusively by the fluid flow parameters of the reservoir that may be obtained by well-logging and measurements on core samples. laboratory 3. The two straight-line well pressure responses typical of fractured reservoirs may be used co calculate the fractures M product and xma&a, the weighted average of the product of the porosity by a characteristic dimension of the matrix blocks. 4. The approach presented does not lead to an analytical description of the transition between the two straight lines of the fractured-reservoir well pressure plot. NOMENCLATURE
A= c. b= k=

surface thickness,

area, cm

sq cm atm- 1 darcy

compressibility, permeability, pressure, incremental flow rate, atm

p= Ap=pi -p= q=

pressure, cc/see cm

atm

r= radial coordinate, LapI ace transform t=


x.

parameter cm

time, vertical Eulers

sectmds dimension, cm = 0.5772 . . sq cm/sec of order n coordinate, constant

generalized

exp (y) = 1.78. hydraulic theta viscosity, integration porosity


v= SUBSCRIPTS

diffusivity, cp parameter operator

function

gradient

AND SUPERSCRIPTS

~ = average

of the variabIe transform

a F

comp = composite F*= Laplace of function properties properties system


/ = fracture-medium

i = initial
ma = matrix medium r = radial

coordinate

S1 = slab sp = sphere

-..

. ..

. .

u = solution with condition w = well value

unitary-step

boundary

aAp~
2rrrwhfkf dr

qw P
r=rw

t~o

ACKNOWLEDGMENTS The author thanks the directive of the Instituto Mexicano del Petr61eo for permission to publish this paper; Eduardo Piiia and Humberto L. Najurieta for their help in the development of the theory; Joaquin Teyssier and Gaelo de la Fuente for their and Necmettin Mungan for his suggestions; encouragement.
REFERENCES 1. Barenblatt,
Equations

Apf = O,

r+

m.....

(A-2)

G. I. and Zheltov, Yu. P.: Fundamental of Filtration of Homogeneous Liquids in Doklady (1960) Fissured Rocks, Soviet Physics, Vol. 5, 522

An additional assumption is that either the fracture and the matrix compressibilities are equal, or they are different and the contrasting expansions of both media upon decompressing do not influence the fluid flow significantly. Assuming that the matrix blocks are shaped as regular solids so that their internal pressure distribution, Apuma, is a known function of time for the conditions of constant initial pressure and unit-step pressure loss at the surface for t >0, the outflow across its surface may be known:

2. Warren, J. E. and Root, P. J.: The Behavior of Naturally Fractured Reservoirs, Sot. Pet. Eng. ./. Tram. AWE, VOL 228. (Sept. 1963) 245-255 3. Crawford, G. E., Hagc~dorn, A. R., and Pierce, A. E.: tCAnalysis of Pressure Buildup Teats in a Naturally Fractured Reservoir, paper SPE 45!58 presented st the SPEAIME 48th Annusl Fall Meeting, Las Vegas, Nev., Sept, 30- Oct. 3, 1973. #~Analysis of Naturally Fractured R: 4. Aguilera, Reservoirs From Sonic and Resist ivity Logs, J. Pet. Tech (Nov. 1974) 1233-1238. 5. van Everdingen, A. F. and Hurst, W.: {The Application of the Laplsce Transformation to Flow Problems in Reservoirs, Trans., AIME (1949) VOL 186, 305-324. 6. Carslaw, H. S. and Jaeger, J. C.: Conduction of Heat i~ Solids, 2nd ed., Oxford U. Press, New York (1959) 97,233. Functions, McGraw-Hill 7. Higher Transcendental Co., Inc., New York (1953) Vol. II, 354. 8. Tables of Integral Transforms, McGraw-Hill ~ok . . Inc., New York (19S4) Vol. 1, 224; Book ~.,

q uma
. .

==
.

k
P . .

Am~(-VApum, )~,,he block . . . . . . . . surf ace . . . .

(A-3)

For variable pressure, Apf, at the blocks surface, the blocks flow per unit fracture porous volume is
qma

-2 t3Apf (Apf, t) . s dr Urea(t-r) dr , Amahf o

. . . . . . . . . . . . . . . . . . . . , (2) where it has been further assumed that the outflow from every matrix block is evenly and instantly distributed in one-half the fracture volume that envelopes every matrix block, and that for fractures that are thin compared with the matrix blocks may k characteristic dimension; this volume approximated by Ama h /2. Inserting Eq. 2 in Eq. A-1 and applying thp Lap {ace, transform gives

9. Handbook of Mathematical Frnzctiorzs, fourth printing, U, S. Government Printing Office, Washington (1964) 811. 10. Kazemi, H.: l~Presaure Transient Analysis of Naturally Fractured Reservoirs with Uniform Fracture Distributions, t Sot. Pef. En& J- (Dee 1969) 45i46~ Trans., AIME, Vol. 246.

L32 Ap: =
f3r2

Ap;

1
S [+ k A

2
~ q:ma

(s)1

Vf

fmaf

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

(A-4)

APPENDIX Under the conditions of radial, slightly compressible flow plus mattix blocks outflow injection in the fractures, the pressure equation in the fracture medium is
kf

1 ile solution to Eq. A-4 plus the transformed equations corresponding to Eq. A-2 may be found following van Everdingen and Hurst. 5 For large times the solution is given by the symmetric Laplace transformation of -1

Ap; =

2rrkfhf; 2P

w p

In [~

S12 (~

* ma

62 APf

= ~
f

~ dAp~
dt

~~

-qmS(Apf,

t)

,
(A-1)

+ kf Am, hf

(s))121

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

. (A-5)

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

where the last term is the flow provided by the matrix. blocks as the pressure is lowered at their surfaces. The initial and boundary conditions are

Apf=O,

t=o

Since the fracture porous medium has a very large permeability compared with the matrix permeability, it may be assumed that at the beginning of the test, fiow caused by fluid compressibility takes place only from the fractures. This is equivalent to assuming zero matrix block flow and that the pressure response at the well is-the homogeneous reservoir equations for the fracture medium.
SOCIETV OF PETROLEUM ENGINEERS JOURNAL

124)

APf .in

qwP

4~f

,)

.(1)

4rrhfkf

yr~

For very high ratios Eq. 4 simplifies to


kf %jl+k
ma

of fracture-to-matrix

diffusivity,

After some time, the effect of the blocks outflow is felt, and the pressure response at the well changes in a gradual transition that is not described by the model to the transformed solution (Eq. A-5). The solution depends on die shape of the matrix blocks through quma(t), the blocks outflow caused by a unitary pressure loss at the blocks sutface. HORIZONTAL
For an infinite reservoir and blocks in the shape pressure distribution inside to the plane at one-half the 1). The pressure increment

hf h
ma

kf hf ~rll.=pc+
ma

h
ma

. . . . . . . . . . . . . . . . . (A-n) SPHERICAL MATRIX BLOCKS

FRACTURES with horizontal fractures of infinite slabs, the the block is symmetrical slab thickness (see Fig. since time zero is6 -~ma (h+l)z~z
t/h~a

A sensible approximation for cubical or parallelepipeds matrix blocks is to consider their behavior similar to that of spherical blocks of equal volume (see Ref. 6, Page 55). The internal pressure distribution in a sphere at null pressure with the with radius rm initially surface subject to a unit-step pressure increment for t >0 is6
Apm ~ (r, t) =

(-1 ) Apma(z, t}= 1 -~ 2 e ?7 n=O 2n+l (2n+l)nz


q

Cos l-l ma

2rma l+ frr

~
n=l

(-l)n e n . . . .

-~man2n2t/rma

nfrr
sin .

r . . . . . . . . . (;:12)

#.

.(A-6)
. . . . . .

and,

at the slab boundary,

dApm, ()z=,m,,, az

=:
ma

~~o (-1) e-qm(2n+)m2th:

A treatment parallel to that of the slabs case leads to equations similar to Eqs. A-7 and A-8, but with a (34 function and a hyperbolic cotangent (see Refs. 7 and 8) instead of the theta function of order one and the hyperbolic tangent; the arguments depend on rms. After approximating

(2n+l)ir . sin

2
2 = h 01 (1/2, ma eqma*2 (hm62)2).

.(A-7)

the solution An alternate

for spherical development


._~e ma

blocks, Eq. 5, is obtained. is presented: 2m -~


n=l
. . . . . . . .
n2772t/r2 mama,

for which the Laplace

transform

iss

(
h
2=
~1/2

a ),= ;:m
. . . .

r
.

ma . .

a Ap (d ~m)

-1
~@ =

(A-14)

tanh (

and
r=r .
ma

. . . . . . . . . . . . . . . . . .(A-8) Approximating the hyperbolic tangent arguments (s small = t large), value for small

2 i
ma

1
n=l s+ 7,1me7r2n2
~
ma

(A-15)

results

in

2 assume

that

s << ~ ma by

so

Eq.

A-15 may be

approximated Substitution in Eq. A-5 and symmetric transformation gives a solution equivalent to a homogeneous reservoir solution with a modified diffusivity constant:

. . . . . . . . . . . . . . . (A- 16) The infinite series is the with exponent two,g and is Riemann zeta function

JUNE, 1976

x~=~ n=l
(For

. . . . . . . . .
6

rr

(A-17)

the horizontal-fractures case the development rrz 9 Substituting into Eqs. leads to ~ 1 tz=o(2rz+ 1)2= x ) A-3 and A-4 leads to a formula similar to the Eq. 1, with homogeneous reservoir equation, diffusivity 1
1 SP = l/(
Tf

tortuosity factor of the fracture. For the case of infinite slabs presented above, the fractures tortuosity is unity, so effective and intrinsic fracture permeabilities are the same. EXTENSION COMPOSITE TO COMPLEX RESERVOIRS

2
3kfhf

kmarma

1 ~)(5)

and with a simplification


3kf hf

similar

to Eq. A-11: (A-18)

7p

A2pcq5marma

Note that, permeability from the first response, Eq. bility of the permeability

in this case, k., is the effective of the fracture medium as computed straight-line portion of the drawdown 1, and is not the cintrinsic permeafracture medium. The ratio of intrinsic permeability is the to effective

TiIe equations developed above are readily extended to composite reservoirs with blocks of varying properties uniformly distributed in a fracture medium with definite permeability and porosity. Assuming that, in stratified reservoirs, there is crossflow among the strata, calling Xm a characteristic dimension (either hma for slabs or 2rma/3 for spherical blocks) and defining = as a weighted average of the distribution of the product characteristic dimension by porosity, results in an equivalent homogeneous reservoir solution, Eq. 1, with diffusivity:

rl comp

.l/(

1
~f +

/lc

. . . .(6)

Xma$ma)

kf hf

***

SOCIETY OF PETROLEUM ENGINEERS JOURNAL

You might also like