The Behavior of Naturally Fractured Reservoirs

J. E. WARREN
P. J. ROOT
MEMBERS A/ME
I GULF RESEARCH & DEVELOPMENT CO.
PITTSBURGH, PA.

ABSTRACT eters were sufficient to describe the single-phase

flow properties of a producing formation, i.e., the
An idealized model has been developed for the
absolute permeability and the effective porosity. It
purpose of studying the characteristic behavior of a
later became evident that the concept of directional
permeable medium which contains regions which
permeability was of more than academic interest;
contribute significantly to the pore volume of the
consequently, the degree of permeability anisotropy
system but contribute negligibly to the flow capacity;
and the orientation of the principal axes of permea-
e.g., a naturally fractured or vugular reservoir. Un-
bility were accepted as basic parameters governing
steady-state flow in this model reservoir has been reservoir performance. 1,2 More recently, 3-6 it was
investigated analytically. The pressure build-up
recognized that at least one additional parameter
performance has been examined in some detail; and,
was required to depict the behavior of a porous system
a technique for analyzing the build-up data to evaluate
containing regions which contributed significant! y to
the desired parameters has been suggested. The use
the pore volume but contributed negligi?ly to the
of this approach in the interpretation of field data
flow capacity. Microscopically, these reg10ns coul~
has been discussed.
be "dead-end" or "storage" pores or, macroscopi-
As a result of this study, the following general
cally, they could be discre~e vol~es of lo~­
conclusions can be drawn:
permeability matrix rock combmed With natura~ ~Is­
1. Two parameters are sufficient to characterize sures in a reservoir. It is obvious that some provision
the deviation of the behavior of a medium with "double
for the inclusion of all the indicated parameters,
porosity"from that of a homogeneously porous medium..
as well as their spatial variations, must be made if
2. These parameters can be evaluated by the
a truly useful, conceptual model of a reservoir is to
proper analysis of pressure build-up data obtained
be developed.
from adequately designed tests. A dichotomy of the internal voids of reservoir rocks
3. Since the build-up curve associated with this has been suggested. 7,8 These two classes of porosity
type of porous system is similar to that obtained
can be described as follows:
from a stratified reservoir, an unambiguous interpre-
a. Primary porosity is intergranular and controlled
tation is not possible without additional information.
by deposition and lithification. It is hig~ly inter-
4. Differencing methods which utilize pressure
connected and usually can be correlated with perme-
data from the final stages of a build-up test should
ability since it is largely dependent on the geometry,
be used with extreme caution.
size distribution and spatial distribution of the
INTRODUCTION grains. The void systems of sands, sandstones and
oolitic limestones are typical of this type.
In order to plan a sound exploitation program or. a b. Secondary porosity is foramenular and is con-
successful secondary-recovery project, sufficient trolled by fracturing, jointing and/or ~~lution _in
reliable information concerning the nature of the circulating water although it may be modified by m-
reservoir-fluid system must be available. Since it filling as a result of precipitation. It is not hig~ly
is evident that an adequate description of the reser- interconnected and usually cannot be correlated with
voir rock is necessary if this condition is to be ful- permeability. Solution channels ?r ~ugul~ voids
filled, the present investigation was undertaken for developed during weathering or burial 10 sedimentary
the purpose of improving the fluid-flow characteriza- basins are indigenous to carbonate rocks such as
tion, based on normally available data, of a particular limestones or dolomites. Joints or fissures which
porous medium. occur in massive, extensive formations composed of
shale, siltstone, schist, limestone. or dolomitC: are
DISCUSSION OF THE PROBLEM
generally vertical, and they are ascr~bed to tens10nal
For many years it was widely assumed that, for failure during mechanical deformation (the permea-
the purpose of making engineering studies, two param- bility associated with this type of void system is
often anisotropic). Shrinkage cracks are the result
Original manuscript received in Society of Petroleum Engineers
office Aug. 17, 19 62. Revised manuscript received March 21, 1963.
Paper presented at the Fall Meeting of the Society of Petroleum
Engineers in Los Angeles on Oct. 7-10, 1962. SPE-426-PA 1 References given at end of paper.

SEPTEMBER, 1963 245

of a chemical process (dolomitization) and do not formation may be obtained from spinner surveys,
appear to have any preferred orientation. observed loss of circulation and down-hole photo-
In the most general case, both classes of porosity graphs. It is clear that data obtained from these
are present and the internal-void volume of the rock sources, which reflect conditions in the immediate
is intermediate in nature, i.e., an independent system vicinity of the well, give only a positive indication
of secondary porosity is superimposed on the primary of the presence of secondary porosity and a descrip-
or intergranular system. The obvious idealization of tion of the primary-porosity material. Some additional
an intermediate porous medium is a complex of dis- information can be acquired from non-routine labora-
crete volumetric elements with primary porosity which tory tests on cores 3-5, 12, 13 if the scale of variation
are anisotropic ally coupled by secondary voids as in the secondary porosity is sufficiently small.
shown in Fig. 1. This heterogeneous, double-porosity Multi-well interference tests 14 or tracer tests can
model will be investigated in detail since a significant be utilized to establish the components of the aniso-
number of petroleum reservoirs can be classified as tropic permeability and the orientation of the principal
intermediate. Although most intermediate-porosity axes. The analysis of pressure build-up or fall-off
rocks are limestones or dolomites, other sedimentary data 15,16 permits the evaluation of the apparent
rocks such as cherty shale or siltstone exhibit this permeability, (the geometric mean of the directional
characteristic. Since the double-porosity model yields permeability components) the completion damage
both of the single-porosity systems as limiting cases, and, possibly, the primary and secondary pore
an internal check on the results is provided. * volume. 17 - 19 This well-test information indicates
Sources of useful information for the characteri- the gross properties of the intermediate formation on
zation of a reservoir with intermediate porosity are an interwell scale.
limited and the information itself tends to be more To develop a plausible model for an intermediate
qualitative than quantitative. The analysis of core reservoir, it is essential that all of the available
data 9-11 is complicated by the presence of extraneous measurements and observations, cited in the preceding
fractures induced during coring and retrieval, poor paragraphs, be utilized; furthermore, the model must
core recovery from intensely fractured zones, dis- be consistent with the physical inferences obtained
placement of fracture surfaces and the size of the from the performance of actual reservoirs of this
sample itself. In spite of these difficulties, acceptable particular type. Applicable studies involving pro-
values for the porosity and permeability of the inter- ducing reservoirs located in many geographic areas
granular material can be determined from plug-core are to be found in the literature. Reflecting these
data, and an estimate of the total porosity can be performance studies are a number of mechanistic
obtained from whole cores. The vertical extent, width, models which have been proposed to describe the
spacing, dip, relative strike, degree of infilling and response of an intermediate reservoir to various
the type of fracture can sometimes be qualitatively natural and/or artificial drives. 20-29 Since our ob-
ascertained from a visual inspection of whole cores. jective is to suggest a model which will simulate
The volume and nature of the solution voids can the behavior of a formation with intermediate porosity
be estimated from the examination of thin sections during single-phase flow, the more esoteric features
and/or polished surfaces. Wh~le it is usually not of the proposed models can be discarded; the result
possible to correlate well logs on a well-to-well is a physical idealization which incorporates a
basis, an indication of the presence and vertical common set of characteristics which seem to be
extent of fractured zones in an individual well may significant.
be obtained from sonic logs; similar qualitative in- The model to be employed in this investigation
is based on the following general assumptions:
a. The material containing the primary porosity
.Analogous physical problems arise from the consideration of is homogeneous and isotropic, and is contained
heat or mass transfer in a heterogeneous medium or electrical
power transmission. Mathematically analogous problems occur within a systematic array of identical, rectangular
when two flow processes are coupled by linear functions in- parallelopipeds.
volving both dependent variables; e.g., chemical reaction, ad-
sorption or phase equilibrium. b. All of the secondary porosity is contained
within an orthogonal system of continuous, uniform
fractures which are oriented so that each fracture is
parallel to one of the principal axes of permeability;
the fractures normal to each of the principal axes
are uniformly spaced and are of constant width; a
different fracture spacing or a different width may
exist along each of the axes to simulate the proper
degree of anisotropy.
c. The complex of primary and secondary porosities
is homogeneous albeit anisotropic; flow can occur
between the primary and secondary porosities, but
VUGS MATRIX MATRIX FRACTURES
flow through the primary-porosity elements can not
ACTUAL RESERVOIR MODEL RESERVOIR occur.
Additional assumptions of more particular nature
FIG. 1 - IDEALIZATION OF THE HETEROGENEOUS
will be made at appropriate points in the mathematical
POROUS MEDIUM.

246 SOCIETY OF PETROLEUM ENGINEERS JOURNAL

treatment. regions, PI and P2, (shown in Fig. 2) it is obvious
While the assumed model certainly implies hetero- that the volume considered must be small in compar-
geneity on a macroscopi·c scale, it may be considered ison with the volume of the reservoir and must be
to be homogeneous if the dimensions of the homoge- large in comparison with the size of the matrix
neous blocks are small in comparison with the dimen- elements; i.e., the following conditions must be
sions of the reservoir, see the discussion by Warren, satisfied:
et al. 30 The same type of argument is often used to
justify the measurement of porosity or permeability
on a small core. The integration of the solution and
JV
9, (V)dV= fJ r (3)

fissure porosities is obviously necessary if the model

is to be practically useful; furthermore, it is reason-
able on the basis of similarity of flow performance.
Iv 9 2 (V) dV= fZJ2 (4)

A certain amount of freedom has been allowed in the If the connate-water saturation in the secondary
description of the fracture system to permit the use porosity is negligible, the average measured inter-
of all of the qualitative information that might be granular porosity ¢m is simply related to ¢1 and ¢Z.
available.
Because of the discrete nature of the primary- (5)
porosity elements and evidence which indicates that For a uniformly thick reservoir that is horizontal,
normal and log-normal distributions of porosity and homogeneous and anisotropic, the single-phase flow
permeability are typical,31 the arithmetic-mean of a slightly compressible liquid is partially described
porosity and the geometric-mean permeability should by applying Green's theorem to the volume V to obtain
be used to obtain a "most probable" model. 32 The the applicable form of the continuity equation.
components of directional permeability and their
orientation should be determined by means of inter-
ference tests. While this completes the description
k2x iJ2P2 +.!:ll..
of the reservoir via current techniques, there are two JL iJx 2 JL
parameters which are not yet determinate - the
iJP2 (6)
secondary porosity and a shape factor which describes
~2C2 iJt
the communication between the primary and secondary
regions. An attempt will be made to develop methods where the x-axis and the y-axis coincide with the
of determining approximate values for these parameters principal axes of permeability. In this expression
from the known information and pressure data from C 1 and C 2 are total compressibilities; however, if it
well tests. is assumed 27 that the external forces are constant,
that there is no interaction between the two regions
THEORY (¢ 1 is independent of P2 and ¢ 2 is independent of
PI) and that the variation of ¢2 with respect to P2
Since it has been assumed that the reservoir can
IS negligible, the following approximations obtain:
be treated as though it were homogeneous, let us
define two pressures at each point (x,y,z,t) in the
following manner: (7)

PI (x,y,z,t) = Iv p(x,y,z,t) 9, (V)dV/

For many of the pertinent cases, the second term in
(8)

Iv 9, (V)dV . . . . 0)*
Eq. 7 will be at least as large as the compressibility
of the flowing liquid.
In addition to Eq. 6, continuity on a local basis is

P2 (x,y,z,t )= Iv P (x,y, r,f) 9 (V) dV/

2
necessary. If it is assumed that a quasi-steady state
exists in the primary-porosity elements at all times,
the following equation must be satisfied in the volume
Iv 92 (V)dV. . . .. (2) V surrounding each point in the reservoir:

where 9 (V)={I, in prilnary porosity

, O,outside of primary porosity

g. tv)
2
=fl, in secondary porosity
l.O,outside of secondary porosity
rul[[iJ
\ '\
MATRIX FRACTURE

If useful solutions are to be obtained utilizing the

"smoothed" pressures in the primary and secondary

*The subscripts I and 2 always refer to the primary and

secondary pore volumes respectively; for definitions of other FIG. 2 - SCHEMA TIC REPRESENTATION OF THE
symbols, see NomenclatlD"e. PRESSURE DISTRIBUTION IN THE MODEL.

SEPTEMBER, 1963 247

 (9)
It is obvious that the well, which becomes an ellipse
The parameter a has the dimensions of reciprocal after transformation, causes the pressure distribution
area; it is a shape factor which reflects the geometry to lack axial symmetry. To avoid the necessity of
of the matrix elements and it controls the flow between including K as a parameter in the solution, it is as-
the two porous regions. * The assumption of a quasi- sumed that the well-bore pressure obtained by solving
steady state will introduce some error into the solu- (10-12) differs by a constant amount from the well-
tions when times are small; but, since the results bore pressure obtained by solving the symmetrical
are ultimately to be used in conjunction with well- problem (K = 1). After a very short time, the difference
test data, the approximation should be adequate. in these pressures due to the distortion of the pres-
Eqs. 6 and 9 can be cast in dimensionless form sure distribution in the region near the well will be
by introducing appropriate transformations. For an constant; 2 therefore, since it has already been as-
infinite reservoir with a uniform initial pressure sumed that errors which are restricted to small times
which is to be produced at a constant rate, the trans- are permissible, the following expression completes
formed equations, initial condition and boundary the statement of the problem:
conditions can be written as follows:

where s* = In ( K+I )
2#
• . • • • • . • • • • (10) t/J2* (l,T) = solution of symmetrical problem, de-
scribed by (10-12) with K = 1, evaluated at g = L
Using the Laplace transformation and solving the
resulting equation subject to the boundary conditions,
the desired solution can be written as follows:

If'I = 1f'2 = 0; for T =0, a II e and (}

*
1f'2 0, T) =
L-I[ Ko(vsl(s» ~ ] (14)
s .jsl (S) Kj (.j sf (s) )
"'2= OJ for T > 0, e CO, all (}

I
217"
{7
0
alf'2
- - d(}=-I·
an '
T > 0, R =
where I (s) =
w(J-w}S +x
(I-w) S+X
The asymptotic solution which is valid for T > 100**
is obtained by making the usual substitutions, Ko(v)

= ~Kcos 2 (}
= (-y + In v - In 2) and Kl (v) '" l/v, and inverting .
+ -k- sin 2 (} . . . (12)

"'2* 0, T) ~ 1/2 {InT+.80908+E;[-~t"/W(I-w}]

where - E; [-~t"/(I-w)]} ..... (15)
CO
exp (-u) du
e= + Ky2 where - E.
I
(-V)=! u
r w2 II

The solution for the case of a finite reservoir can

(} = tan-I (~y) be obtained in a similar manner if the condition at
infinity in (12) is replaced by

T = k 2 1<¢ICI +¢2C2)JL';

*2 = v'Ik2 x k2y I K=.jA2x/k2:Y c)1f'? _

ae -0; for T > 0, e = H, all (} . . (16)
2
ak,rw If R 2 »K, Eq. 13 is valid for this case; and,

*2 **This condition is sufficient for all values of'\ and W;

however, if,\« 1, the condition is T> lOOW; or, if (Ll«
1,
*For a more detailed discussion, see Appendix A. T> (100'11. - 1)/A.

248 SOCIETY OF PETROLEUM ENGINEERS JOURNAL

 12,-------------------,

"
10

• . • (17)

DISCUSSION OF RESULTS 0.
~ 7
The sol utions obtained emphasize the need for o
~ ,1.- 5)(IO-e
two parameters with which to describe a reservoir of ::>
::l... 6
the intermediate type; this agrees with the conclusions 0:
0.
of Stewart, et aI,6 which were based on experimental III

work. The next step is to examine the effects of the .......

III

z
parameters on reservoir behavior; then an attempt to o
0;
evaluate them must be made. ~ 4
2
If the derived results are to be accepted, it is C
necessary that the limiting behavior described by INFINITE RESERVOIR
W • 0 FOR ALL CASES
Eqs. 15 and 17 be physically reasonable. As <u->1.0,
both equations approach the asymptotic solution
obtained by Van Everdingen and Hurst 33 for the same
boundary conditions. * This behavior is proper since
the primary porosity, or its effective compressibility,
must vanish if cu is to approach unity; then, the
reservoir contains only the homogeneous secondary ~ d d d d d d d d d ~
DIMENSIONLESS TIME. l'
porosity. Similarly, the equations indicate homoge-
neous behavior as A .... 00. This is also correct since FIG. 3 - SOLUTIONS OBTAINED BY APPROXIMATE
there is no impedance to interporosity flow when A INVERSION.
approaches infinity, i.e., either kl or a must become
infinite. Furthermore, the necessary condition that
the infinite reservoir responds in a homogeneous
manner for very large values of time is properly
described.
Some results for the special case in which cu = 0
(negligible storage capacity in the secondary porosity) 8

in an infinite reservoir are shown in Fig. 3; this

particular form of the model has been discussed III
quite extensively. 15,27It is apparent that the dimen- ~ 7
sionless pressure drop tf!2*(1,T) increases discontin- "!

uously when production begins, and it asymptotically

...+e INFINITE RESERVOIR
W' 0 FOR ALL CASES

approaches the homogeneous solution whenAr becomes ~ 6

I
large. The pressure discontinuity is the result of a ;:-
lack of fluid capacitance in the secondary porosity;
the magnitude of the discontinuity is ~(-y - InA) for
the cases of interest. Fig. 4 indicates the deviation ...o
from the asymptote for each case. The intercept at ...z
...
IX

T = 1 is equal to the initial pressure discontinuity, ::: 4

and all of the linear segments have the same slope,

c
III
III
i.e., -1.15/ cycle. The apparent linearity of the curves ~
persists until AT > 0.05. ~ 3
0;
z
In Fig. 5, behavior patterns that are associated ...::E
with finite values of cu are shown; once again the o
reservoir is infinite. The most notable feature is the
second linear segment which is parallel to the asymp-
tote but vertically displaced by an amount that is
equal to In(l/~). The transitional curve which con-
nects the two linear portions represents the inter-
action between A"'and cu. Deviations from the asymp-
~OO 10'

*For 1I» I, -E; (-v) -=exp (lI)/lI; and. for 1I« I, --£j (-v) DIMENSIONLESS TIME. l'

~ - <y + Inll) where Y = 0.5772••• FIG. 4 - DEVIATION FROM THE ASYMPTOTE.

SEPTEMBER, 1963 249

tote are presented in Fig. 6. The initial constant these particular results may be rather artificial, they
level is equal to the displacement of the second do suggest the possibility of misinterpretation; e.g.,
parallel segment from the homogeneous asymptote. the latter part of the transitional curve between the
The linear portion of the difference curve has a slope initial linear portion and the asymptote can be ana-
of -1.15/cycle and an intercept of ~[-y-In'\+ln(1-<u)] 12,------------------,
at T = 1. Although the linear part of the deviation
curve will not always be clearly defined, a line of the II

proper slope passing through the point of inflection

will generally yield an. intercept that will permit ,\ to 10
be approximated quite adequately.
No results are presented for finite reservoirs since 9
the behavior described by Eq. 17 is so familiar. The
form of the equation is that which results in the .~ 8
classic time-lag curve 34 encountered commonly in
the quasi-steady state. The asymptote for the dimen-
.
o
0:
o
sionless draw-down tj;* (I,T) is a linear function of T '"
0:

with a slope of WR 2_1) and an intercept approximate! y ""'

"''"
equal to [lnR - % + 2(1 - (i))2/,\R2]. The logarithm ..
0: 6

of the deviation from the asymptote is also a linear "'"'

function of r; the slope is [-,\/2.3cu( 1 - cu)] cycles '"z
..J 5
o
and the intercept at T = 0 is log [2(1 - cu)2/'\(R2 -1)]. iii
z
In principle, it is possible to evaluate R, ,\ and cu '"
~
i5
from the asymptote and the difference curve.
For practical use, equations, expressed in field
units, for the draw-down and build-up histories and INFINITE RESERVOIR

for the deviation of the actual build-up curve from its --W·[Link]
------- W·[Link]
asymptote are listed in Appendix B. Theoretical ---W·O.I
build-up curves for various combinations of ,\ and cu
are shown in Figs. 7 and 8 to illustrate the type of
behavior that can be expected in an infinite reservoir.
~ d d d d d d d d d ~
The assumed parameters and operating conditions DIMENSIONLESS TIME. f
used to generate the data are the following:
It is obvious that the behavior when there is only FIG. 5 - SOLUTIONS OBTAINED BY APPROXIMATE
INVERSION.
primary porosity cu = 0 is so discontinuous that the
build-up occurs almost instantaneously when ,\ is 3.51=:::::======--------1
INFINITE RESERVOIR
very small. The other cases for the smaller value of ---W·[Link]
'\, 5 x 1O-9,give results that cause the reservoir to -------- W· 0.0 I
---W·O.I
appear to have a closed boundary; furthermore, if 3.0
only the early portion of the curve were recorded,
the value of P, determined by extrapolation to
([Link] s + I3.t) = 1, would be erroneously high by the
..
CD
o
o
II!
quantity m log (1/ cu) and the value of the skin resist- +
~ 2.5
ance Sd would be hi~h by 1.15 log (1/ cu). * Although
~
*The apparent value of the skin resistance in an anisotropic
reservoir is actually equal to Sd-S*; therefore, the correction
should be made if well completions are to be evaluated on this
basis. If a negative skin resistance is measured in a well that
has not been stimulated, it may be used to obtain a lower limit
on the anisotropy parameter K.
'"oz
'"0:
TABLE 1 ......
W

i5
"'
If)

k2 40 md '"
.J
Z
o
h 20 ft iii
~ 1.0
Te 00 ~
i5
Tw 0.316 It
¢l C 1+¢2 C 2 2.64 x 10- 6 psi -1
0.5
11 2 cp
fJ 1.23
Sd = S*
oL-~_~~~-L~~~~~-L-~-~
P 4000 psia d d d d d d d d d d ~
q 115 STB/D DIMENSIONLESS TIME. T

ts 21 days FIG. 6 - DEVIATION FROM THE ASYMPTOTE.

250 SOCIETY OF PETROLEUM ENGINEERS JOURNAL

lyzed by Pollard's method 17 or its extension 19 just or misuse. Until the effect of the closed boundary is
as ifthe reservoir actually had a finite drainage radius. felt, the reservoir will act as though it were infinite,
F or A = 5 x 10 -6, the curves, with the exception so the previously discussed methods can be applied.
of ILl = 0, exhibit a double-slope type of build-up Beyond that time, the difficulties inherent in the
performance; apparently, this behavior is typical of procurement of sufficient, accurate data and the
fractured reservoirs. 16 ,17,35 Unfortunately, it is fundamental uncertainties in the available methods
characteristic of stratified reservoirs as well, so the of analysis force us to question the value of such
evaluation of such data may be uncertain. If, however, tests. The time required to achieve quasi-steady
a plot of the deviation from the asymptotic solution state conditions in a heterogeneous reservoir is one
behaves in the prescribed manner, it is safe to con- or two orders of magnitude greater than it is in a
clude that the reservoir can be considered to be homogeneous reservoir. 35 The accurate determination
intermediate in nature for predictive purposes. It may of slowly varying pressures is quite difficult unless
be postulated that evidence of this pressure lag a pressure-sensing device that is operating in the
indicates the presence of macroscopic heterogeneity correct range is being used (it might be necessary
or that A is actually a direct measure of the scale of to .chang: pressure bombs several times to accomplish
heterogeneity; e.g., for A » 1, the reservoir behaves :hlS). Fmally, the analysis of these small changes
homogeneousl y. m pressure may become largely a matter of personal
When both of the parallel segments are discernible, judgment; and, if higher order pressure differences
the build-up analysis presents no problem. The later are involved, the results may be more artistic than
portion is interpreted in exactly the same manner as s~ientific. To emphasize these points, the dimen-
it would be if the reservoir were homogeneous. In slOnless pressure history of a single, undamaged
addition to the possible errors which can result from well producing at a constant rate from an infinite
using the wrong portion of the build-up curve, em- homogeneous reservoir 33 was used together with the
ploying an incorrect method of analysis or assuming reservoir data in Table 1, q = 200 STB/D
the wrong type of spatial permeability distribution, and ts = 10 days, to obtain the synthetic build-up
there is the possibility that a significant portion of curve shown in Fig. 9. The data are plotted in the
the curve may be masked by afterflow effects. There ~rescribed mann:r 17,19 and the graphical analysis
is also a finite probability that the use of the "log IS shown. The mterpretation of the parameters so

(P - Pw) vs time" method 16,17, 19 could result in a determined leads to the following results:
complete description of a reservoir solely on the
pore volume of the coarse voids 1,100,000 cu ft
basis of after production. This difficulty can be cir-
cumvented by utilizing. a recently developed tool
which permits the well to be closed in at the
formation. 36
No results are presented to demonstrate the ex-
pected behavior of a finite reservoir. This is done
deliberately to avoid any chance of misunderstanding

woO
4000

3950 I

.
.~
'":>
0:
In

.'"
In
3900 0:

~
IU
II:
=>
rn 3850
..'"
0:
o
~
..J
rn
IU
II: '"
~
0-
IU
II: 3800
0
CD
I
J
J 3700
IU INFINITE RESERVOIR
~
3750
O INFINITE RESERVOIR
.\.- 5X10- FOR ALL CASES
'\-5Xlo-
e FOR ALL CASES
'l-1I5 STB/D
3700 \'"5 STB/D
t. - 21 DAYS 3650
t•• 21 DAYS

3650

10"'"

FIG. 7 - THEORETICAL BUILD-UP CURVES. FIG. 8 - THEORETICAL BUILD-UP CURVES.

SEPTEMBER, 1963 251

 pore volume of the fine voids 5,670,000 cu ft present in the system.
radius of the reservoir 904 ft 2. These parameters can be evaluated by the
proper analysis of pressure build-up data obtained
radius of the damaged region 130 ft
from adequately designed tests.
It is obvious that this approach yields erroneous 3. Since the build-up curve associated with this
results. The fault cannot be assigned entirely to the type of porous medium is similar to that obtained
model upon which the technique is based since the from a stratified reservoir, an unambigous interpre-
model is physically reasonable during quasi-steady tation is generally not possible without additional
state flow. Rather, the difficulty lies in the nature of information.
the data used. It is probable that any decaying func- 4. Methods which utilize differences obtained
tion can be approximated by a series of exponential from pressure data recorded during the final stages
terms over an interval of gradual change; thus, am- of a build-up test should be used with extreme caution.
biguity results.
NOMENCLATURE
As a consequence of the examination of the approx-
imate mathematical treatment of the idealized physical C total compressibility, LT 2;M
model, it has been suggested that the two additional K Yk2xlk2' degree of anisotropy, dimension-
parameters which are significant in the behavior of less Y
an intermediate reservoir can be evaluated from pres- P = initial pressure, MILT2
sure draw-down or build-up data. The data should
be obtained before the effect of the reservoir boundary R = TelTw' radius of closed boundary, dimension-
less
is felt at the well. Extreme caution should be used
in the analysis of data secured during the later stages S* In [(k+l) /2 v'KI, effective skin resistance
=
of build-up in a finite reservoir. due to anisotropy, dimensionless
S d = skin resistance due to completion, dimen-
SUMMARY AND CONCLUSIONS sionless
An investigation has been conducted for the pur- S we = connate water saturation, dimensionless
pose of improving the description of a formation Co = compressibility of flowing liquid, L T 21M
which contains both primary (intergranular) porosity C P = effective pore compressibility, L T2/M
and secondary (fissure and/or vugular) porosity. C w = compressibility of connate water, L T2;M
Based on the assumption that the primary-porosity
h = thickness of formation, L
region contributes significantly to the pore volume
but contributes negligibly to the flow capacity, an k = absolute permeability, L2
idealized model has been developed to study the k 2 = yk 2 k2 ' effective permeability of anisotrop-
characteristic behavior of such a reservoir. Unsteady- ic m'edi'Um, L 2
state flow in this model reservoir has been described £ = characteristic dimension of heterogeneous
mathematically, and asymptotic solutions have been region, L
obtained. The pressure build-up performance has been
p = pressure, MILT2
examined in some detail; and, a technique for analyzing
the build-up data to evaluate the desired parameters pw = well-bore pressure, MILT 2
has been suggested. The use of this approach in the Apw = deviation of actual pressure from the value
interpretation of field data has been discussed. on its asymptote, MIL T2
As a result of this study the following general q = production rate, L 3 IT
conclusions can be drawn: s = Laplace operator, l/T
1. Two parameters are sufficient to characterize t = time, T
the deviation of the behavior of a medium with' 'double
ts = time of shut-in,T
porosity" from that of a homogenous, porous medium.
One of the parameters, cu, is a measure of the fluid At = t - ts' time elapsed since shut-in, T
capacitance of the secondary porosity and the other, x. y, z rectangular coordinates; the axes coincide
..\., is related to the scale of heterogeneity that is with the principal axes of permeability, L
a geometric parameter for heterogeneous
region, l/L2
INFINITE RESERVOIR
1-00 OR W-I
",-200STBID
f3 = formation volume factor, dimensionless
t.-IO DAYS y 0.5772..., Euler-Mascheroni constant, di-
mensionless
..\. = aklTw21li2. the parameter governing inter-
porosity flow, dimensionless
11 = viscosity, MILT

.a
SHUT-IN TIME. HOURS
100 110 120 130 140 ISO
t: =
s
jf x2
-- + --
Ky2
radial coordinate, dim en-
KT2 T 2 '
FIG. '9 - WORK CURVE FOR BUILD-UP ANALYSIS w w
(REFERENCES 19, 21). sionless

252 SOCIETY OF PE'V-ROLEUM ENGINEERS JOURNAL

 Dallas, Texas (Oct. 8-11, 1961).
T = k2t/(CPtCl + ¢2C2)prw2, time, dimension-
20. Pirson, S. J.: Bull., AAPG (1953) Vol. 37, 232.
less
21. Baker, W. J.: Pmc., Fourth World Pet. Congress,
Ts = time of shut-in, dimensionless Sect. II (1955) 379.
~T = T-T s' time elapsed since shut-in, dimension- 22. Birks, J.: Proc., Fourth World Pet. Congress, Sect. II
less (1955) 425.
23. Jones-Parra, J. and Reytor, R. S.: Trans., AIME
¢ porosity, dimensionless
(1959) Vol. 216, 395.
0/ = 2TTk2h (P-p)/ QI1f3, pressure decline, dimen- 24. Freedman, H. A. and Natanson, S. G.: Pmc., Fifth
sionless World Pet. Congress, Sect. II (1959) 297.
0/ * = pressure decline in secondary porosity at 25. Aronofsky, J. S. and Natanson, S. G.: Trans., AIME
T uJ d"ImenSlOn1ess
2 (1958) Vol. 213, 17.
26. Maksimovich, G. K.: Geologiya Nefti i Gaza (Petro-
0/ w 0/2 *-S*, pressure decline at well, dimen- leum Geology) (1958) 258.
sionless 27. Barenblatt, G. I., Zheltov, I. P. and Kochina, I. N.:
(i) ¢2 C2 /(¢1 C1 +¢2 C 2)' parameter relating fluid PMM (Soviet Applied Mathematics and Mechanics)
capacitance of the secondary porosity to that (1960) Vol. 24, 852.
of the combined system, dimensionless 28. Mattax, C. C. and Kyte, J. R.: Soc. Pet. Eng. Jour.
(June, 1962) Vol. 2, No.2, 177.
SUBSCRIPTS 29. Tarr, C. M. and Heuer, G. J.: "Factors Influencing
1 primary porosity the Optimum Time to Start Water Injection", presented
at the Rocky Mountain Regional Meeting of the Society
2 secondary porosity of Petroleum Engineers, Billings, Mont. (May 24-25,
m matrix (refers to properties measured on small 1962).
core samples) 30. Warren, J. E., Skiba, F. F. and Price, H. S.: Jotn'.
Pet. Tech. (Aug., 1961) Vol. XIII, No.8, 739.
X,y,Z vector component
31. Bulnes, A. C.: Trans., AIME (1946) Vol. 165, 223.
32. Warren, J. E. and Price, H. S.: Trans., AIME (1961)
REFERENCES
Vol. 222, 11.
1. Barfield, E. C., Jordan, J. K. and Moore, W. D.: 33. Van Everdingen, A. F. and Hurst, W.: Trans., AIME
JOUf'. Pet. Tech. (April, 1959) Vol. XI, No.4, 15. (1949) Vol. 186, 305.
2. Mortada, M. and Nabor, G. W.: Trans., AIME(1961) 34. Barrer, R. M.: Diffusion In and Through Solids, Uni-
Vol. 222, 11. versity Press, Cambridge (1951).
3. Klute, C. H.: Jotn'. Polymer Science (1959) Vol. 41, 35. Lefkovits, H. C., Hazebroek, P., Allen, E. E. and
307. Matthews, C. S.: Trans., AIME (1961) Vol. 222, 11.
4. Fatt, I.: Trans., AIME (1959) Vol. 216, 449. 36. Pitzer, S. C., Rice, J. D. and Thomas, C. E.: Trans ..
5. Goodknight, R. C., Klikoff, W. A. and Fatt, I.: JOUf'. AIME (1959) Vol. 216, 416.
Physical Chern. (1960) Vol. 64, 1162.
6. Stewart, C. R., Lubinski, A. and Blenkarn, K. A.: APPENDIX A
Trans., AIME (1961) Vol. 222, 383.
7. Bulnes, A. C. and Fitting, R. V.: Trans., AIME (1945) ADDITIONAL THEORETICAL CONSIDERATIONS
Vol. 160, 179.
8. Imbt, W. C. and Ellison, S. P.: API Drill. and Prod.
Prior to the derivation of Eqs. 6 and 9, it was
Prac. (1946) 364. assumed that the secondary porosity was contained
9. Atkinson, B. and Johnston, D.: Trans., AIME (1949) within an array of homogeneous, geometrically iden-
Vol. 179, 128. tical elements. Since it is ultimately necessary to
10. Kelton, F. C.: Trans., AIME (1950) Vol. 189, 225. introduce average values in order to obtain useful
11. "Geological Report, Steelman Main Midale Beds Unit", results, this simplification is justified. To show that
submitted to Oil and Gas Conservation Board of there is no loss of generality, an alternate approach
Saskatchewan, October, 1960.
will be considered.
12. Stewart, C. R., Craig, F. F. and Morse, R. A.: Trans., Let us define PI (x,y,z,t) within a representative
AIME (1953) Vol. 198, 93.
volume in the following manner:
13. Stewart, C. R., Hunt, E. B., Schneider, F. N., Geffen,.
T. M. and Berry, V. J.: Trans., AIME (1954) Vol. 201,
294.
14. Elkins, L. F. and Skov, A. M.: Trans" AIME (1960)
Vol. 219, 301.
.0, (X,); Z', t) = t-iy
<X>

, 0
(..t) 0,' (1 )p, Lli t ) d.t
. . • • • • • • • • • • • • (1-1)
15. Dyes, A. B. and Johnston, O. C.: Trans., AIME (1953)
Vol. 198, 135.
16. Matthews, C. S.: JOUf'. Pet. Tech. (Sept., 1961) Vol.
XllJ, No.9, 862.
17. Pollard, P.: Trans., AIME (1959) Vol. 216,38.
where ¢, = I
o
<X>
I' (.l) ¢, I (l) d.t

18. Samara, H.: "Estimation of Reserves from Pressure

Changes in Fractured Reservoirs", presented at
Second Arab Petroleum Congress, Beirut, Lebanon
9.1,' (l) = ¢m(.l) [1-5 we LL)] m
(Oct., 1960).
19. Pirson, R. S. and Pirson, S. J.: "An Extension of the
Pollard Analysis Method of Well Pressure Build-Up
and Drawdown Tests", presented at the 36th Annual
Fall Meeting of Society of Petroleum Engineers,
PI (.tj t) =vhr 1. 1.

P (..£ iV, t ) A (v) d v

253
SEPTEMBER, 1963
 vel) d..t = bulk volume of matrix material three dimensions of the identical matrix blocks and
the width of the fracture (or one dimension and the
with characteristic dimension
three fracture widths) can be calculated if k2 ,k2 '
from .J, t,O L + d..l per unit bulk and k2 and ¢2 are known. Since this infor~ati6n
CO will no't generally be available, it is necessary to
vOlume'l
, vCI)dL= I-¢2 resort to approximations. The simplest approach is
o to assume uniformly spaced fractures and to allow
1. variations in the fracture width to satisfy the con-

Vel) = J o
A (v )dv = charcteristlc volume
ditions of anisotropy. Then,

a = 4n (n+2) I../. 2 . . . . . . (1-8)*

associated with the dimension..t. where n = number of normal sets of fractures = 1,2,3.
Let us define t in terms of measurable parameters.
The interporosity flow per unit volume, q, is given
by 1/2 2 )1/2
L = z[n(n + 2)] kl rw .. , (1-9)
q
_
-JL1JI
I
[l~UI"" U It 2lll aW (h-P" The following min-max approximations obtain:
( >..k Z

. . • . . . . . . • . . . . . (1-2) 2 )1/2
where P2 (x,y,z,t) is assumed to be only a function
of time over the region of interest. Let
j.= 4.78
(
~;: j n = 1,2,3. (1-10)

Cl() where the maximum error is 38 per cent.

a~=
1 v(.I)¢II(l)k l (..t)at.l)(p2-PI) d..e__
-=0______________________________
(k 2 )1/2
= 4.30 \ ~:w2 n = I, 2 . (1-11)**
(P2-PI)d~
.J, j
[flJ V (},) f)11 eL)
o where the maximum error is = 24 per cent.
• • • . • • • • • • • • • • • • . • • • (1- 3)
Substituting (1-3) in (1-2), ..l = 6.55 (klrW 2 )1/2; n=2,3. . (1-12)
>"k 2

q= ;~, ~21.~.ll ,,: {:J 1Il"(IIPodl] (.l)<U

where the maximum error is = 16 per cent.
If the blocks actually have the dimensions x = a,
y = band z = c, the equivalent value of t can be
estimated from the surface-volume ratio; i.e.,
. • . (1-4) 30bc
Utilizing previously defined quantities, .l = . n =3· . . (1-13)
( ob + bc + co) ,
ak l
II = p:- (P2- P I)' . • • (1-5) 20b
n= 2 . . . . . . (1-14)
..t= (o+b)
Similarly,
(l)
.t=o; n=l . . . . . . . . (1-15)
q= 0 C
1 1 0:,1 =jV(I)¢II( .J)C1(./) , dL· (1-6)
APPENDIX B
o
BEHAVIOR EQUATIONS
In the equations that follow, these definitions
where obtain:***

Therefore, quite generally,

*Indicative of the insensitivity of a to the exact geometry is
the fact that the results for an n-dimensional sphere agree with
t
(1- 8) if is replaced 2r.
**This is the most probable case since it is unlikely that
It i.s apparent that the characteristic dimension horizontal fractures contribute significantly to the flow capacity.
t must be examined more closely. In principle, the * **Units are psis, STB/D, res. bbl, md, ft, p s1- l and hours.

254 SOCIETY OF PETROLEUM "ENGINEERS JOURNAl.

 6Pw (T's +tU")~.435m{E; [- Xt.T'/w (I-w)] - E;
[-XLh/O-W)]} ;

all values of wand all values of ~T' (II-3)

S *-J. (K+I)
- n \ 2ft bopw(T's+~d~-mlog w; w¢Oand small
values of boT' . . . . (II-4)

Sd = skin resistance due to completion (constant portion of curve)

6Pw( t) =difference between the asymptote ~pw(T's+ ~T )~ mr·251-log X+109(I-wH09bo~;

and the actual curve
all values of wand intermediate values of
~T .. .. .. .. .. .. .. .. .. .. .. (II-5)
INFINITE RESERVOIR
(Jj near portion of curve)
0) Draw- Down, T' > 100 W if A« I or
T' > 100 - t if W « I
FINITE RESERVOIR

0) Draw-Down,T>lOOwR2ifA« I on>ICXY1'2-1:
or T' > 100 (all cases)

1;1mt+(l7"'f-exp tXdw( hoI]})-

t
Og T' +.351 +.435E,.fAT'/WO-W)]} Jl,.(rlO!P-
pw(T)a::P-m
-.435E;[-h/U- w )]-.87S,". 87Sd -. 87m Un R - . 75 - S* + Sd) . . . .... (II-6)

b) Build-Up, ATS> 5w
. . . . .... (II-I)
boT' > IOOwR 2 if X « I
b) Build-Up, AT'S> 3
6 T' > 100w if A « I or
boT' > 100 R2 -i: if w « I

6T' > 100- t

if W « I Ar(rs + 6 r I",p- ,;tnf+ 2
U-;:wI . . pt X6 r/..(I-<UW

. . . .. . .. . . . . .. . . . . . . .... (II-7)

10gr~p'(
t w +~ T' ~=IO"[1..74m(J-w)2l_
~
1'$ .. AR2

(II-2)
-.435 [A~T'/W(J-w)] .............. "" (II-8)
***

SEPTEMBER, 1963 255