Está en la página 1de 8

Thermal Recovery Well Test Design

and Interpretation
A.K. Ambastha, * SPE, and H.J. Ramey Jr., SPE, Stanford U.

Summary. Because a reservoir undergoing a thermal recovery process is typically idealized as a composite reservoir, this study de-
velops new design and interpretation equations by investigating pressure-derivative behavior of a well in a two-zone, radial, infinite or
finite composite reservoir. Accurate design equations help establish the test duration required to observe a particular feature in well test
data and thus the applicability of an interpretation method to determine front radius, or swept volume.
This study shows that the dimensionless time to the end of the first semilog line (deviation time) based on front radius is a constant.
The deviation-time method may be used if wellbore storage does not mask the first semilog line. Design equations for the time of the
beginning of the second semilog line and the time to observe outer-boundary effects show that the intersection-time method is not suitable
for thermal recovery well test analysis. Correlations developed for the end of pseudosteady-state behavior of the swept region should
help select the correct Cartesian line to calculate swept volume. This paper also presents derivative type curves applicable for all front
radii, with mobility and storativity ratios as parameters for infinitely large composite reservoirs. For closed and constant-pressure outer
boundaries, the ratio of outer boundary to front radius is the third parameter.

Introduction
The behavior of composite reservoirs has recently attracted much from pressure analysis. Because the duration of a flow regime
attention, and many studies have appeared on this subject. A com- appears shorter on a derivative graph, care is required for well test
posite system can occur naturally or may be artificially created. analysis by a specialized method to ensure that pressure data are
Aquifers with two different permeabilities forming two regions, in the correct time range. Because of detail enhancement in a
oil and water or gas and oil zones, with different properties in a derivative graph, however, better type-curve matching may be pos-
reservoir, and a finite-thickness skin region are examples of naturally sible with a derivative type curve. To use pressure derivatives,
occurring two-zone composite systems. Secondary or tertiary design equations and type curves based on pressure derivatives for
recovery projects-such as waterflooding, polymer floods, gas in- the system under consideration are necessary. We consider com-
jection, in-situ combustion, steamdrive, and CO 2 miscible posite reservoirs in this study. Fig. 1 is a schematic of a composite
flooding-artificially create conditions wherein the reservoir can reservoir. Inner-and outer-region properties are assumed to be
be viewed as consisting of two zones with different rock and/or different. The distance R is the front radius, which is an important
fluid properties. Geothermal reservoir well tests can often be rep- parameter sought in thermal recovery well test analysis. Strictly
resented by a composite reservoir model. speaking, fronts in thermal recovery usually are not cylindrical
The pressure behavior of composite reservoirs has been con- owing to gravity effects. Thus, the front radius exists only in some
sidered extensively. 1-13 Recently, Brown 14 investigated pressure- average sense. It is perhaps better to speak of the volume of the
derivative behavior of composite reservoirs but limited his study inner region, especially when pseudosteady data are available.
to mobility and storativity ratios on the order of 0.4 to 2.0 and 0.3 "Front radius" is used here, however, as a matter of convenience.
to 30, respectively. Such mobility and storativity ratios are typical This term has been used in most of the previous studies of well
of cases with finite-thickness skin regions around the wellbore. Wat- test analysis appliep to thermal systems.
tenbarger and Ramey 15 modeled a finite-thickness skin region as
a composite system and obtained pressure-transient behavior for Methods To Estimate Front Radius. Different methods have been
such systems using finite-difference techniques. Their solutions cor- proposed to estimate a front radius from pressure-vs.-time data:
respond to a range of mobility ratio from 0.1 to 3.6, which is the (1) deviation-time method, (2) intersection-time method, (3) type-
range Brown used. In this study, we consider mobility and stora- curve-matching methods, and (4) pseudosteady-state method.
tivity ratio contrasts typical of thermal recovery processes and de- The deviation-time method 20 ,21 uses the time at the end of the
velop new design and interpretation equations by investigating semilog pressure-vs.-time line corresponding to the inner region
pressure-derivative behavior. The Eggenschwiler et at. 10 analytical to calculate a front radius on the basis of a theoretical dimension-
solution was used for a two-zone, infinitely large composite less deviation time. The intersection-time method 9 ,22,23 uses the
reservoir. A constant rate was specified at the inner boundary. The intersection time of two semilog lines corresponding to the mobili-
Horne et al. II solution was used for finite composite reservoirs. ties of the inner and outer regions to calculate a front radius, again
The outer boundary for a finite reservoir can be either closed or with a theoretical dimensionless intersection time as the basis. A
at a constant pressure. Wellbore storage was neglected in this study. type-curve-matching method was proposed by Bixel and van
Implications of this study on different methods to calculate front Poollen. 6 Barua and Horne 13 used automated type-curve match-
rapius or inner-zone volume are discussed. ing to analyze thermal recovery well tests. Eggenschwiler et al. 10
proposed a pseudosteady-state method for large mobility and stora-
Pressure-Derivative Approach. Whenever a straight line is sought tivity contrasts. They observed that the swept region could behave
on a graph of pressure vs. a function of time, we seek a constant like a closed system for a short duration after the end of the semi-
slope. Thus, pressure derivatives can be used directly. Also, a log line corresponding to the inner-region mobility. During this time,
pressure derivative can enhance a pressure signal and may be more a pseudosteady Cartesian line may develop with a slope that can
sensitive to disturbances in reservoir conditions. 16-18 In general, be related to the swept volume. This method should be indepen-
a pressure-derivative graph shows the start and end of a given flow dent of the geometry of the swept region and has been applied by
regime later and earlier, respectively, than is indicated on a pressure several investigators to field and simulated cases 24-30 with appar-
graph at a specified accuracy. This is illustrated for simple cases ent success.
in Ref. 19.
The examples in Ref. 19 suggest that there is often more detail
evident in a derivative graph than is apparent in a pressure graph. Infinite Outer· Boundary Results
Times of specific flow events determined from pressure-derivative We now consider pressure-derivative behavior for a well producing
analysis, however, can often be different from those determined (or injecting) at a constant rate in a two-zone composite, infinitely
large reservoir. The derivative of well pressure with respect to the
• Now at the U. of Alberta. base 10 logarithm of time is
Copyright 1989 SOCiety of Petroleum Engineers dPwD/d log tD=tD In(IO)£-I{sPwD}' ................ (1)
SPE Formation Evaluation, June 1989 173
30
Co =0
M =10

Fs = 100
20 Ro

.9

-
OJ)
0
10

-J
-0

10

A
50
100
-0
• 1000

o~~~~--~---L--~
O.ot 0.1 10 100 1000

Fig. 2-Semilog pressure derivatives.

1000
CD =0
..... fw
= 100
R ....0
100 Fs
bJ)

~
--.
0 10
g
I
0.01 0.1 10 100 1000 10000
E

Fig. 1-Two-zone, radial composite reservoir. Fig. 3-Effect of mobility ratio on pressure derivative.

where PwD is the dimensionless wellbore pressure drop in Laplace 2. The curve for RD = 10 is slightly different from the other curves
space, as given by the Eggenschwiler et al. solution. Inversion from for tRD ::50.5. Thus, the correlation is valid for practical purposes.
Laplace space to real space was performed with the Stehfest 31 in- It is likely that well bore storage and other practical matters could
version algorithm. Wellbore storage was neglected to make the tran- affect results for RD <50 and tRD<0.5.
sition between the zones unambiguous. Storage and skin can easily
be added to Eq. 1. Because the pressure derivative is not affected Effect of Mobility and Storativity Ratio. Fig. 3 shows the effect
by the presence of skin as long as wellbore storage is zero, 14 it of mobility ratio on semilog pressure-derivative behavior for a fixed
is not necessary to assume negligible skin to study the pressure- storativity ratio of 100. The semilog pressure-derivative behavior
derivative behavior. for a homogeneous reservoir (M = 1, Fs = 1) is also shown in Fig.
3. The first semilog straight line of slope 1.151 develops on a dimen-
Correlating Parameter. Satman et al. 32 and Tang 33 plotted sionless graph OfPwD vs. 10g(tD)' After the end of the first semilog
pwD-In(RD I500) vs. tRD to correlate pressure responses for all line, the pressure derivative rises for M~ 1. During the transition
front radii to the response for RD =500. R D =500 is an arbitrary period, the pressure derivative goes through a maximum above the
choice, and tRD is given by slope of the second semilog line corresponding to the outer-region
mobility, if mobility ratio, storativity ratio, or both are greater than
tRD = [(0.OOO264k 1)/(t/>IlCt h]( tIR2)=(tD IRB)' ......... (2)
unity. Even in the case of unit mobility ratio, there is a long tran-
Satman et al. and Tang correlated pressure responses, neglecting sition between the two semilog straight lines. The second semilog
wellbore storage or skin. Their approach suggests that a graph of line slope is 1. 151M. For large mobility and storativity ratios, the
dpwDld log tD vs. tRD should apply for all front radii. An example inner region may behave like a closed system for some time during
of such a correlation is shown in Fig. 2, which shows semilog the transition period after the end of the first semilog line.
pressure-derivative behavior for several dimensionless front radii. Pseudosteady-state behavior of the inner region during the tran-
Mobility and storativity ratios are 10 and 100, respectively. The sition was found by Eggenschwiler et al. Thus, during the early
storativity ratio is defined as transition period, a Cartesian graph of pressure vs. time may contain
a straight line, with a slope related to the volume of the inner region.
Fs=(t/>cth/(t/>cth =M IT! . ........................... (3)
From Fig. 3, the following is apparent for a storativity ratio of 100.
Curves for RD=50, 100, and 1,000 appear to form a single 1. The first semilog line ends at tRD -0.18 for any value of mo-
curve for all times. The curve for RD = 10 is also shown in Fig. bility ratio studied.

174 SPE Fonnation Evaluation, June 1989


1000

.9

--J ':[ ~~
100
bI)

l
....0
:@
bl)

:@ --~ 10

FS -1
1
I O,oi 0.1 10 100 1000 10000 le+05
O.oJ 0.1 10 100 1000 10000

Fig. 5-Pressure-derivatlve type-curve for composite


Fig. 4-Effect of storativlty ratio on pressure derivative. reservoirs.

2. There is a long transition period between the end of the first is much larger than the time from pressure analysis, however, which
semilog line and the beginning of the second semilog line. emphasizes an important result of this study. Pressure and pressure
3. The transition period is longer for larger mobility ratios. This derivatives may appear to indicate greatly different event times.
translates to a longer time to the beginning of the second semilog Design equations presented in Eqs. 4 through 7 are accurate to within
line for large mobility ratios. 2 % in pressure derivative. Several investigators 22 ,23,33 have de-
4. The time of the maximum derivative and the magnitude of the veloped criteria for (tRD)end and (tRD)n using pressure data to
maximum derivative are affected by mobility ratio. certain precision. In the following, we compare Eqs. 4 and 6 with
Brown 14 reported a minimum transition time of approximately other design criteria.
two log cycles for composite reservoirs. Long transition periods
are also observed in Wattenbarger and Ramey'sl5 solution for Time to the End of the First Semilog Line. This time, also called
pressure-transient behavior for a single well with wellbore storage the deviation time, has been used to calculate front radius. 20 ,21,23
and a finite skin thickness in an infinitely large reservoir. The skin The appropriate equation in field units to calculate the front radius is
region was treated as the inner region, and the formation as the
outer region. R=V[0.OOO264kl/(<f>l'cth][te~d/(tRD)end]' ............. (8)
Fig. 4 shows the effect of storativity ratio on semilog pressure- where tend = time to the end of the first semilog line on a pressure
derivative behavior for a mobility ratio of 10. For storativity ratios vs. log(time) graph (hours) and (tRD )end = dimensionless deviation
greater than unity, the pressure derivative rises above the value time based on front radius. van Poollen20 used a value for (tRD)end
I. 151M during the transition period and passes through a maximum derived from the radius-of-drainage concept in an earlier paper. 21
slope. Thus, a hump occurs in the pressure-derivative behavior for Merrill et at. 23 derived a value for (tRD )end by generating a wide
mobility and storativity ratios larger than unity. Fig. 4 shows the range of pressure-falloff curves for two-zone, radial composite
following for a mobility ratio of 10. reservoirs using a numerical simulator. They found the dimen-
I. Storativity ratio does not affect the time to the end of the first sionless deviation time to lie between 0.13 and 1.39 by running
semilog line corresponding to the inner-region mobility and mildly several cases. The arithmetic average dimensionless deviation time
affects the time to the beginning of the second semilog line corre- was 0.389. They stated that the range of error with the arithmetic
sponding to the outer-region mobility. The transition time between average value of (tRD )end =0.389 would be
the two semilog lines is approximately three log cycles in duration. 0.58~RestIRact ~ 1.89 .............................. (9)
2. Storativity ratio affects the derivative behavior at intermediate
times. The storativity ratio mildly affects the time and magnitude They felt that the range of error was too large and suggested
of maximum slope. against indiscriminate use of deviation time to calculate the radius
of a fluid bank. Sosa et at. 34 used the average dimensionless devi-
Pressure Derivative Type Curve ation time of 0.389 to analyze simulated falloff tests in water-
and DeSign Equations injection wells. They observed that the front radius using the
deviation-time method was not an accurate estimate for the radius
Fig. 5, a graph of semilog pressure derivative vs. tRD with mo-
of the waterflooded region.
bility and storativity ratios as parameters, is a pressure-derivative
type curve for composite reservoirs in the absence of wellbore Tang 33 approximated (tRD)end to be 0.4 by observing the
pressure response from the Eggenschwiler et at. 10 analytical so-
storage. Analysis of Fig. 5 results in several empirical well test
design equations for composite reservoirs. These design equations lution. Although (tRD )end of 0.18 and 0.4 are not dramatically
are summarized in the following. different, a front radius calculated with (tRD ) end =0.4 will be ap-
The time to the end of the first semilog straight line is proximately 0.67 times a front radius calculated with (tRD )end =
0.18, with all other parameters remaining the same. This is a sig-
(tRD)end=0.18 .................................... (4) nificant difference in answers for front radius, indicating the need
for accurate specification of deviation time to obtain meaningful
The time of the maximum derivative in the transition is
results from the deviation-time method.
(tRD)max=(l.8+0.4log Fs)M . ...................... (5) Use of Eqs. 4 and 8 creates a convenient expression in field units
to calculate R as
The time of start of the second semilog line is
R=V [k l 1(<f>l'cth](tend/681.8) . ...................... (10)
(tRD)n =90(1 + log Fs)M . ........................... (6)
Mobility or storativity ratio or both should be about one order
Eqs. 5 and 6 apply if mobility and storativity ratios are greater
of magnitude away from unity to obtain a deviation time precisely
than unity. In addition, the time to the beginning of the first semilog
and thus reasonable results from the use of Eq. 10. Eq. 10 can be
line corresponding to inner-region mobility is 29
used if the assumptions of the analytical model are reasonably
tD ~ 140, ......................................... (7) satisfied and wellbore storage does not mask the first semilog line
corresponding to the inner-region mobility. Also, a new correlation
which is the same as the time to the beginning of the semilog line
has been developed for the time to the end of storage-dominated
for a finite-radius well with no wellbore storage in an infinitely large
period given by19
homogeneous reservoir. The time in Eq. 7, though correct, is of
little practical importance because of storage and skin. The time (tDICD)""0.04810g(CDe2S)-0.03, ................. (11)

SPE Formation Evaluation, June 1989 175


le+05
Co =0
20
10000 M = 10
Fs = 1000
15 _0.
rco fRo= 10
..P FS =1000 RD bl) 1000
bl)
:@
~
-...
10 reD fR D = 10
50
......0.
.g 100

J 5
• 100 100

• 1000 10
1000

0
0.01 0.1 10 100 1000 10000 I
0.01 0.1 10 100 1000 10000

tRD

Fig. 6-Semllog pressure derivative va. t liD (constant- Fig. 7-Semllog pressure derivative va. fliD (closed outer
pressure outer boundary). boundary).

which describes the time by which the slope of a fog-log graph of Tang 33 obtained
pressure vs. time has decreased by 2 % from the initial value of unity .
(IRD)I1=dOM2-IIFs IFs for MIFs~1 ............... (13a)
Although not investigated in this paper, the geometry of the swept
region is also a critical factor in the application of the deviation- and (tRD)11 =-0.44+(1OM 2IFs) for M1F s< 1. .......... (13b)
time method. If the swept inner region is not cylindrical, then the
Substituting Fs=1 in Eq. 13a results in
deviation time should correspond to the closest discontinuity
affecting the transient response at the well. Thus, deviation time (tRD)11 =- 10M. . .................................. (14)
could correspond to a "minimum" front radius and an underesti-
Thus, Eqs. 12 and 14 produce a time of the beginning of the
mated swept volume. The swept region may not be cylindrical be-
second semilog line in the same range for Fs= 1. Eq. 12 is accurate
cause of (1) gravity override and underride, as in thermal processes,
to within 9%, and Eq. 14 to within 5%. Tang also states, without
(2) viscous fingering, as in the case of unfavorable mobility ratio
derivation, that dimensionless wellbore pressure drop for the second
processes (such as CO 2 flooding), and (3) injection into a fractured
semilog line is
well. '
PwD= ~[M In(2.2458tRD IT/)+ln(R,5)] . ............... (15)
Time of the Beginning of the Second Semilog Line. This time
has been of interest to many investigators. Development of a second Eq' 15 is derived starting from Ramey's approximate solution
semilog line is required for the intersection-time method to determine in the Appendix. Derivation in the Appendix provides a criterion
front radius. Odeh 22 investigated systems with mobility ratios for the time of the beginning of the second semilog line as follows:
equal to diffusivity ratios (i.e., F s= I) that varied from 0.25 to 50
(IRD)I1> lOOMIFs for MIFs~ 1 ................... (16a)
using an analytical solution. He found by graphical methods that
the second semilog line starts at and (tRD)11 > 100 for M IF s S; 1. ..................... (16b)
(tRD)I1=7.7M, for Fs=l. ......................... (12)
Thus, Eqs. 16 establish a lower limit for (tRD)I1' Any design
By comparing the pressure response from the Eggenschwiler el equation presented for (tRD)11 must produce (IRD)l1larger than or
al.1O analytical solution with Ramey's9 approximate solution, equal to those from Eq. 16.

40 le-t07
CD =0 CD :0

M=IO le+06 M=IO

30
Iffi)S

S' .9 10000
0/) 0/)
0
:;; :@
--
d
c
20
--
d
."
c 1000
."

100
10

10

I
0.01 0.1 \0 100 1000 10000 I_S le+()6
10 100 1000 10000 Ic+OS le+06

tRD

Fig. 9-EHect of reD IR D on pressure derivative (closed out-


Fig. 8-EHect of reO IR D on pressure derivative (conatant- er boundary).
pressure outer boundary).

176 SPE Formation Evaluation, June 1989


1000 1000

100 CD =0 100 CD =0
Fs = 100 Fs = 100

10 M =100 10
Cl
...:< j
.." .."
RD
---Cl

~
100
.g
0.1 • 500 0.1

• 1000 M 1

0.01 0.01

0.001 0.001
0.001 0.01 0.1 10 100 1000 10000 0.001 o.oJ 0.1 10 100 1000 10000

Fig . 10-Cartesian derivative . Fig . 11-Effect of mobility ratio on Cartesian derivative .

A comparison of results from Eqs. 13 and 16 showed that re- reD /RD for closed and constant-pressure outer boundaries resulted
sults from Eqs. 13 were poor. Eqs. 13 apply if M, F s, and 1/ are in the following relation for the dimensionless time at which the
all greater than unity. Eq. 6 results in a longer time than Eq. 13. pressure-derivative response for a finite, composite reservoir departs
The difference between times computed from Eqs. 6 and 4 is the from that of an infinitely large composite reservoir:
transition time to reach the second semilog line after the end of
the first semilog line. Even for moderate-mobility-ratio cases, the (tRD)depart = [(reD/RD)2M]/5Fs· .................... (17)
transition time is so long that well tests would seldom be run long Eq. 17 should be applied only to cases where M and Fs 2: 1. The
enough to observe the second semilog line. The second semilog accuracy of Eq. 17 depends on the values of M and Fs. Eq. 17
line may also be masked by outer-boundary effects. It is likely that is best for large values of M and F s compared to unity. It applies
only one semilog line will be evident in most cases. Next, outer- to both closed and constant-pressure outer boundaries. For the
boundary effects are considered to quantify the time when the outer homogeneous-reservoir case (M= 1, Fs= 1), Eq. 17 indicates that
boundary affects pressure-transient behavior. the pressure-derivative response departs from infinite-acting be-
havior at tAD=0.2f7r, where tAD is the dimensionless time based
Outer-Boundary Effects on area A =7rr;. A comparison ofO.2f7r with 0.1 [which is (tAD)pss
For a finite outer boundary, PwD in Eq. 1 is given by the Horne for a well producing at a constant rate in a closed homogeneous
et al. 11 solution. Figs. 6 and 7, which illustrate typical results, reservoir] indicates the results of Eq. 17 when M and F s are close
apply for M=IO, Fs=l,OOO, and reD/R D=IO. Three cases of to unity.
RD=50, 100, and 1,000 are shown on both figures. In addition Eq. 17 quantifies the outer-boundary effects on transient responses
to M and Fs, the group reD/R D is a third correlating parameter in composite reservoirs and is a means to determine whether desired
for finite, composite reservoirs. features will be seen in a pressure-transient test. A comparison of
The reservoir approaches steady-state behavior at late times for Eqs. 17 and 6 provides a limit for reD/R D to observe a second
a constant-pressure outer boundary. On a pressure-derivative graph, semilog on a pressure-transient test as
such as Fig. 6, steady state is indicated by a pressure derivative (reD/R D) >-J450(1 +log Fs)Fs . .................... (18)
of zero. Because large mobility and storativity contrasts imply closed
reservoir behavior, the pressure derivative rises for some time after Similarly, a comparison of Eqs. 17 and 5 provides a limit for
the end of the first semilog line corresponding to the inner-region reD/R D to observe a maximum semilog pressure derivative as
mobility in Fig. 6. But the outer-boundary effects eventually (r eD/RD»-J(9+210g Fs)Fs . ..................... (19)
dominate, and the reservoir approaches steady state after exhibiting
a maximum pressure derivative. Eqs. 18 and 19 show that the limiting value of reDIR D for ob-
A reservoir approaches pseudosteady-state behavior at late times serving a second semilog line or maximum semilog derivative is
for a closed outer boundary produced at a constant rate. Pseu- only a function of the storativity ratio. Eq. 18 shows that for a large
dosteady state is characterized by a linearly increasing pressure storativity ratio, a second semilog line will be masked because of
derivative on either a Cartesian graph or the log-log graph of Fig. outer-boundary effects. The limit on reDIRD posed by Eq. 18
7. The effects of mobility and storativity contrasts and the outer suggests that the intersection-time method is not applicable for
boundary are such that stabilization to a maximum derivative and thermal recovery well test analysis.
bending over of the pressure derivative is not seen in Fig. 7. In-
stead, the reservoir goes directly to pseudosteady state. Type-Curve-Matchlng Method
Figs. 8 and 9 show pressure-derivative behavior for constant- Wellbore storage should be small to use the type-curve presented
pressure and closed outer boundaries, respectively, for several in Fig. 5 for an infinitely large composite reservoir. We recommend
values of reD/RD' Mobility and storativity ratios are 10 and 1,000, well test data collection to a time slightly larger than the time indi-
respectively, for Figs. 8 and 9. Interaction of the effects of mo- cated by Eq. 5 so that a bending over of the semilog pressure deriva-
bility and storativity contrasts and the outer boundary determines tive is observed. An approximate correlation for the maximum serni-
the pressure-derivative behavior at any time. Depending on the size log pressure derivative in an infinitely large composite reservoir is
of the outer region, a second semilog line mayor may not appear.
Figs. 8 and 9 show that reD/R D should be greater than 1,000 for (dPwDld log tD)max=(1.1+log Fs), for M=1 ........ (20a)
the second semilog line to be evident, if M=1O and Fs=I,OOO.
and (dPwDld log tD)max=(0.7+10g Fs)M, for M2: 10... (20b)
Thus, even if one is willing to run a well test long enough, the second
semilog line may be masked by outer-boundary effects. Analysis The accuracy of Eqs. 20 improves as the values of mobility and
of pressure-derivative behavior for several values of M, F s , and storativity ratios increase beyond unity. Eqs. 20 are applicable only

SPE Formation Evaluation, June 1989 177


1000 100 r-------~-------~
CD =0
100 CD =0
Slope within 2% of 21t
M = 100

10
Slope within 5% of 21t
j

....
"0
......
0 1 10 100 10
.g Fs
.i,
0.1
.'.' 1000..'
.. ..
0.01
o
-<
0.001
0.001 0.01 0.1 10 100 1000 10000

,,
.
,. "
Fig. 12-Effect of storatlvlty ratio on Cartesian derivative.

for the cases where M"2 1 and Fs > 1. We should also ensure that
the limit on reDIR D based on Eq. 19 to observe a maximum
semilog pressure derivative is satisfied.
If the conditions mentioned are satisfied, then type-curve matching
can provide the values of M and Fs. The pressure derivative match
0.1 ~~====:l:=
10 100
__ ~1000
point can be used to calculate (klp.)1 by Mobility Ratio

(~) = 141.2qB (dPwDld log tD)match , . . . . . . . . . . . . . . (21) Fig. 13-Correlatlon for the end of pseudosteady-state be-
havior.
P. I h (dpw /d log t)match
and the time match point yields an estimate of front radius, R, if
the inner-region properties are known. An estimate of front radius,
R, is given by In a log-log presentation, a Cartesian derivative would show a
slope of - 1 during infinite-acting radial flow of inner and outer
R= J"-0-.000-=--264-k-I --tm-a-tc-h-. . .................... (22) regions and would be constant at 21r during the pseudosteady-state
period. This is shown in Fig. 10. Dimensionless front radii of 100,
(<!>P.C,)I (tRD) match 500, and 1,000 are presented in Fig. ro. Mobility and storativity
There may be cases where the limit based on Eq. 19 is not satisfied ratios are both 100 in Fig. 10.
and therefore a type curve like Fig. 5 is not appropriate. In such Fig. 11 presents the effect of mobility ratio on the Cartesian
cases, analysis should consider the parameter reDIRD in addition pressure derivative for Fs= 100. Early- and late-time behaviors
to M and Fs. One option is to use automated type-curve matching shown in Fig. 11 follow Eqs. 26 and 27. From Fig. 11, after the
in these cases. If any of the three parameters are known with end of the infinite-acting radial flow corresponding to the inner-
reasonable accuracy by independent means, however, then a type region mobility, a short-duration pseudosteady-state period is
curve can be prepared showing the effects of the other two param- evident, depending on the value of mobility ratio. The larger the
eters, and usual type-curve matching can be performed to estimate value of mobility ratio, the longer the duration of the pseudosteady-
those parameters. state period.
Fig. 12 presents the effect of storativity ratio on the Cartesian
Pseudosteady-State Method pressure derivative for M= 100. Remarks for Fig. 11 apply to Fig.
For large mobility and storativity ratios, the swept inner region may 12 also. For a given mobility ratio, the pseudosteady-state period
behave as a closed reservoir and exhibit pseudosteady-state behavior is longer for larger storativity ratios. Storativity ratio affects the
for a short period of time after the end of the first semilog line. Cartesian pressure derivative at intermediate times. The late-time
Cartesian pressure derivative is independent of the storativity ratio
Pseudosteady-state behavior may be observed when tAD >0.1. Eg-
genschwiler et al. \0 used the slope of a Cartesian straight line to and follows the behavior indicated by Eq. 27.
calculate the inner-zone swept volume through a relation expressed Correlations for the time to the end of pseudosteady-state behavior
in field units as are shown in Fig. 13. From Fig. 13, the time to the end of
pseudosteady-state behavior is larger for larger values of mobility
Vs=(5.615qB)lmcc, . ............................. (23) and storativity ratios. Using the correlation for the slope to be within
During pseudosteady state, the dimensionless pressure for a well 2 % of 21r in Fig. 13, empirically, we observe that pseudosteady-
in a homogeneous reservoir is given by 35 state behavior is likely to appear for cases with MFs"2 104 and
M "210, if pseudosteady-state behavior is desired to last up to
PwD=21rtAD + If21n(Alr;) + Ihln(2.2458/CA). . ......... (24) tAD ""0.2. Correlations presented in Fig. 13 should be of help in
selecting the correct pseudosteady Cartesian line to calculate swept
Differentiating Eq. 24 with respect to tAD results in
volume.
dP wD ldtAD=21r, ................................. (25) The limit on reDIRD to observe pseudosteady-state behavior re-
where tAD is based on area, A =1rR2. The Cartesian pressure sults from comparing tAD=O.1 with Eq. 17:
derivative during infinite-acting (semilog) radial flow for inner and reD IRD >..J1rFs /2M . ............................. (28)
outer regions, respectively, are given by
The limit of Eq. 28 is more likely to be satisfied than the limits
dPwDIdtAD = 1I2tAD ............................... (26)
of Eq. 18 or 19 for typical values of M and Fs encountered in
and dpwDldtAD =M 12t AD . ........................... (27) thermal recovery processes. Also, because the pseudosteady-state
178 SPE Formation Evaluation, June 1989
method is independent of the swept-region geometry, this method tend = deviation time, hours
should yield reasonably correct swept volume and "average" front tRD = dimensionless time based on R,
radius for irregularly swept regions. O.OOO264(kl¢p,crh tlR2
(tRD )depart = dimensionless time for slope response to deviate from
Conclusions infinitely large composite reservoir behavior
1. Pressure-derivative behavior for composite reservoirs with mo- (tRD)end = dimensionless deviation time
bility and storativity contra!'ts typical of thermal recovery processes (tRD)max = dimensionless time for maximum semilog slope
has been studied in the absence of wellbore storage. (tRD)IT = dimensionless time of the start of second semilog line
2. Several new design and interpretation equations related to the Vs = swept volume, ft3 [m 3]
specific flow regimes observed in composite reservoir well tests 1/ = diffusivity ratio, (kl¢p,c rh I(kl¢p,c rh
have been developed. p, = viscosity, cp [Pa' s]
3. The deviation time method may be used if wellbore storage ¢ = porosity, fraction
does not mask the first semilog line. This method is sensitive to
the deviation time, however, and may underestimate swept volume Acknowledgments
in cases of irregularly swept regions.
4. The intersection-time method is not suitable for thermal Financial support was provided by the Stanford Geothermal
recovery well test analysis for two reasons: either a well test would Program, U.S. DOE Contract No. DE-AT02-80SF1l459, SUPRl-
seldom be run long enough to see a second semilog line or outer A, U.S. DOE Contract No. AC03-8ISF1l564, SUPRl-A Indus-
boundary effects would mask the second semilog line. This con- trial Affiliates, and Stanford U.
clusion is in agreement with qualitative observations of previous
investigators. Also, wellbore storage may mask the first semilog References
line, rendering the intersection-time method inapplicable. I. Hurst, W.: "Interference Between Oil Fields," Trans., AIME (1960)
5. Conditions have been established for the applicability of a • 219, 175-92.
derivative type-curve matching method. Guidelines have been 2. Mortada, M.: "Oilfield Interference in Aquifers of Non-Unifonn Prop-
provided for sufficient test data collection to ensure reliable type- erties," JPT (Dec. 1960) 55-57; Trans., AIME, 219.
curve matching. 3. Loucks, T.L. and Guerrero, E.T.: "Pressure Drop in a Composite
6. A correlation has been developed for the time to the end of Reservoir," SPEJ (Sept. 1961) 170-76; Trans., AIME, 222.
4. Jones, P.: "Reservoir Limit Test on Gas Wells," JPT (June 1962)
pseudosteady-state behavior corresponding to the swept inner region
613-19.
for large-mobility- and large-storativity-ratio cases. This correlation 5. Rowan, G. and Clegg, M.W.: "An Approximate Method for Tran-
should be of help in choosing a correct pseudosteady Cartesian line. sient Radial Flow," SPEJ (Sept. 1962) 225-56.
If a pseudosteady Cartesian line develops, the pseudosteady-state 6. Bixel, H.C. and van Poollen, H.K.: "Pressure Drawdown and Buildup
method should yield a correct swept volume and "average" front in the Presence of Radial Discontinuities," SPEJ (Sept. 1967) 301-
radius for irregularly swept regions. Alternatively, a type-curve 09; Trans., AIME, 243.
based on the Cartesian slope may be used to evaluate the size of 7. Larkin, B.K.: "Solutions to the Diffusion Equation for a Region
the inner region. Bounded by a Circular Discontinuity," SPEJ (June 1963) 113-15;
Trans., AIME, 228.
8. Carter, R.D.: "Pressure Behavior of a Limited Circular Composite
Nomenclature Reservoir," SPEJ (Dec. 1966) 328.-34; Trans., AIME, 237.
A = area, 7rR2 or 7rre2, ft2 [m 2] 9. Ramey, H.J. Jr.: "Approximate Solutions for Unsteady I,iquid Flow
in Composite Reservoirs," J. Cdn. Pet. Tech. (Jan.-March 1970) 32-37.
B = FVF, RB/STB [res m 3 /stock-tank m 3 ] 10. Eggenschwiler,.M. et al.: "Interpretation of Injection Well Pressure
Cr = total system compressibility, psi - I [Pa - I] Transient Data in Thennal Oil Recovery, " paper SPE 8908 presented
C = wellbore storage coefficient, bbl/psi [m 3 /Pa] at the 1980 SPE California Regional Meeting, Pasadena, April 9-11.
CA = shape factor II. Horne, R.N., Satman, A., and Grant, M.A.: "Pressure Transient
CD = dimensionless wellbore storage, Analysis of Geothennal Wells with Phase Boundaries," paper SPE 9274
5.6146C 127r(¢crhhr; presented at the 1980 SPE Annual Technical Conference and Exhibition,
Dallas, Sept. 21-24.
Fs = storativity ratio, (¢cr)l/(¢crh 12. Onyekonwu, M.a.: "Interpretation of In-Situ Combustion Thennal
h = formation thickness, ft [m] Recovery Falloff Tests," PhD dissertation, Stanford U., Stanford, CA
k = permeability, md (1985) .
.,c -I = inverse Laplace transform 13. Barna, J. and Horne, R.N.: "Computerized Analysis of Thennal
me = Cartesian line slope, psi/D [PaId] Recovery Well Test Data," SPEFE (Dec. 1987) 560-66.
M = mobility ratio, (kIp.) 1 I(kl p.h 14. Brown, L.P.: "Pressure Transient Behavior of the Composite
Reservoir," paper SPE 14316 presented at the 1985 SPE Annual Tech-
PD = dimensionless pressure drop in inner region, nical Conference and Exhibition, Las Vegas, Sept. 22-25.
k1h(Pi -p)1141.2qBp.1 15. Wattenbarger, R.A. and Ramey, H.J. Jr.: "An Investigation of Wellbore
Pw = wellbore pressure, psi [kPa] Storage and Skin Effects in Unsteady Liquid Flow: II. Finite Difference
PwD = dimensionless wellbore pressure drop, Treatment," SPEJ (Sept. 1970) 291-97; Trans., AIME, 249.
k1h(Pi -Pw)1141.2qBp,j 16. Bourdet, D. et al.: "A New Set of Type Curves Simplifies Well Test
PwD = dimensionless wellbore pressure drop in Laplace
Analysis," World Oil (May 1983) 95-106.
17. Bourdet, D. et al.: "Interpreting Well Tests in Fractured Reservoirs, "
space World Oil (Oct. 1983) 77-87.
q = surface flow rate, STBID [stock-tank m 3 /d] 18. Bourdet, D., Ayoub, J.A., and Pirard, Y.M.: "Use of Pressure
r = radius, ft [m] Derivative in Well Test Interpretation," paper SPE 12777 presented
rD = dimensionless radius, rlrw at the 198.4 SPE California Regional Meeting, Long Beach, April 11-13.
re = outer-boundary radius, ft [m] 19. Ambastha, A.K.: "Pressure Transient Analysis for Composite
reD = dimensionless outer-boundary radius, r elrw Systems," PhD dissertation, Stanford U., Stanford, CA (1988).
20. van Poollen, H.K.: "Transient Tests Find Fire Front in an In-Situ Com-
r w = wellbore radius, ft [m] bustion Project," Oil & Gas J. (Feb. I, 1965) 78-80.
R = discontinuity radius, ft [m] 21. van Poollen, H.K.: "Radius of Drainage and Stabilization Time
R aet = actual discontinuity radius, ft [m] Equations," Oil & Gas J. (Sept. 14, 1964) 138-43.
RD = dimensionless discontinuity radius, Rlr w 22. Odeh, A.S.: "Flow Test Analysis for a Well With Radial Discon-
Rest = discontinuity radius estimated with Eq. 8, ft [m] tinuity," JPT(Feb. 1969) 207-10; Trans., AIME, 246.
s = Laplace parameter 23. Merrill, L.S., Kazemi, H., and Gogarty, W.B.: "Pressure Falloff
Analysis in Reservoirs With Fluid Banks," JPT (July 1974) 809-18;
t = time, hours
Trans., AIME, 257.
tAD = dimensionless time based on area A 24. Walsh, J.W. Jr., Ramey, H.J. Jr., and Brigham, W.E.: "Thennal In-
tD = dimensionless time, O.OOO264(kl¢tJ.Cr)ltlr; jection Well Falloff Testing," paper SPE 10227 presented at the 1981

SPE Fonnation Evaluation, June 1989 179


SPE Annual Technical Conference and Exhibition, San Antonio, Oct. 35. Ramey, H.J. Jr. and Cobb, W.M.: "A General Pressure Buildup Theory
5-7. for a Well in a Closed Drainage Area," JPT(Dec. 1971) 1493-1505;
25. Messner, G.L. and Williams, R.L.: "Application of Pressure Tran- Trans., AIME, 251.
sient Analysis in Steam Injection Wells," paper SPE 10781 presented
at the 1982 SPE California Regional Meeting, San Francisco, March
24-26.
Appendix-Late·Time Drawdown Solution for a
Wen In an Infinitely Large Composite Reservoir
26. Messner, G.L. and Williams, R.L.: "Further Investigation of Pressure
Transient Testing in Steamflood Projects," paper SPE 11087 presented Dimensionless wellbore pressure drop using Ramey's9 approx-
at the 1982 SPE Annual Technical Conference and Exhibition, New imate solution is
Orleans, Sept. 26-29.
27. Onyekonwu, M. O. et al.: "Interpretation of Simulated Falloff Tests, " PwD(tD) = - ~{Ei( -I/4tD)-Ei( -R'jj/4tD)
paper SPE 12746 presented at the 1984 SPE California Regional
Meeting, Long Beach, April 11-13. +Me[(11- 1)RJI /4tDEi( -1]R'jj/4tD)}' .......... (A-I)
28. Onyekonwu, M.O., Ramey, H.J. Jr., and Brigham, W.E.: "Application
of Superposition and Pseudosteady State Concepts to Thermal Recovery All Ei terms can be replaced by their log approximations and the
Well Tests," paper SPE 15536 presented at the 1986 SPE Annual Tech- exponential term will be within 1% of 1.00, if
nical Conference and Exhibition, New Orleans, Oct. 5-8. tRD ~ 100M 1Fs for M IFs~ 1 ..................... (A-2a)
29. Fassihi, M.R.: "Evaluation of an Analytical Technique for Estimating
Swept Volume From Thermal Pressure Falloff Tests in Heterogeneous and tRD~IOO for MIFs :5.1. ....................... (A-2b)
Systems," SPEFE (June 1988) 449-58; Trans., AIME, 285.
30. Da Prat, G., Bockh, A., and Prado, L.: "Use of Pressure Falloff Tests A simplification of Eq. A-I under the conditions of Eq. A-2 re-
to Locate the Burning Front in the Miga Field, Eastern Venezuela," sults in
paper SPE 13667 presented at the 1985 SPE California Regional
Meeting, Bakersfield, March 27-29. PwD(tD)= ~{M In[(2.2458tD)/(1]R'jj)] +In(R'jj)}. ...... (A-3)
31. Stehfest, H.: "Algorithm 368, Numerical Inversion of Laplace Trans-
forms," D-5, Communications of ACM (Jan. 1970) 13, No. 1,49. which represents a late-time drawdown solution for dimensionless
32. Satman, A. et al.: "An Analytical Study of Transient Flow in Systems wellbore pressure drop. Eq. A-3 is the same as that reported by
with Radial Discontinuities," paper SPE 9399 presented at the 1980 Tang. 33
SPE Annual Technical Conference and Exhibition, Dallas, Sept. 21-24.
33. Tang, R.W.-K.: Transient Pressure Analysis in a Composite Reservoir, SPEFE
topical report, U.S. DOE (Aug. 1982).
34. Sosa, A., Raghavan, R., and Limon, T.J.: "Effect of Relative Perme- Original SPE manuscript received for review Sept. 27. 1987. Paper accepted for publication
Nov. 25. 1987. Revised manuscript received July 13. 1988. Paper (SPE 16746) first
ability and Mobility Ratio on Pressure Falloff Behavior," JPT (June presented at the 1987 SPE Annual Technical Conference and Exhibition held in Dallas.
1981) 1125-35. Sept. 27-30.

180 SPE Formation Evaluation, June 1989

También podría gustarte