SPE 100384

Diagnosis of Reservoir Behavior From Measured Pressure/Rate Data

C.S. Kabir, SPE, Chevron ETC, and B. Izgec, SPE, Texas A&M U.

Copyright 2006, Society of Petroleum Engineers Inc.

rate. Most of these methods apply to single wells in volumetric
This paper was prepared for presentation at the SPE Gas Technology Symposium held in reservoirs producing single-phase fluids from a fixed drainage
Calgary, Alberta, Canada, 14-17 May 2006.
boundary. Mattar and Anderson (2003) provide a
This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as
comprehensive treatment of the pertinent methods. Analytic
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to methods (Marhaendrajana 2005; Marhaendrajana and
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at Blasingame 2001) have also been proposed to handle well
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
interference in multiwell reservoirs. Gringarten (2005) showed
for commercial purposes without the written consent of the Society of Petroleum Engineers is that the reservoir-compartmentalization question can be
prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300
words; illustrations may not be copied. The proposal must contain conspicuous addressed by deconvolving simultaneously measured
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
pressure/rate data for wells across a perceived fault barrier.
Multidisciplinary approach has also been reported to address
the compartmentalization question (Bigno et al. 1998).
Abstract Changes in well performance may often be attributed to
This paper presents a simple diagnostic tool to identify condensate banking, reservoir subsidence, fines migration
reservoir flow behavior from a Cartesian pressure/rate graph. precipitating changing skin, and a host of completion and/or
Some of the benefits of the proposed tool are its simplicity wellbore-lift issues, besides depletion. Our challenge is to
without requiring any calculations, leading to understanding of decipher the real reason for premature production decline. In
reservoir compartmentalization and application of an this regard, Anderson and Mattar (2004) offer a few diagnostic
appropriate material-balance technique. clues about wellbore loading and changing skin, changing well
Data diagnosis entails graphing pressure with rate and productivity, and identifying external pressure support or
discerning trends; positive slope signifies the pseudosteady- interference.
state (PSS) flow period, whereas the negative slope implies This study offers a simple methodology to diagnose long-
infinite-acting (IA) flow. Constant-rate production exhibits term well performance, especially those that are influenced by
infinite slope whereas constant-pressure production produces outer boundaries. In particular, whether wells belong to the
zero slope. Mathematical justifications for these diagnostic same or different compartments become quite evident.
signatures are presented. During PSS flow, wells belonging to Because we are solving an inverse problem, independent
the same container will exhibit the same slope. methods must be used to eliminate potential reservoir,
Differences in slope are an indication of reservoir wellbore, and surface flowline network issues before reaching
compartmentalization, lateral or vertical. Equally important reasonable conclusions. Mathematical proofs are presented in
we provide mathematical proof of why different wells in a support of the contentions presented in this study.
multiwell reservoir system should have the same slope. Field
examples from multiple gas and gas/condensate systems show Theoretical Considerations
how the proposed tool works in practice. When production is initiated in a well, various flow regimes
are encountered as transition from IA to PSS flow, with
Introduction possible intervening transitional flow, occurs. Fig. 1
With increasing usage of permanent downhole and/or surface schematically depicts such a scenario on a Cartesian pwf–q
sensing, the need for simple diagnostics becomes imperative graph. Of course, the size of the connected-pore volume
so that actions can be taken just in time for reservoir (CPV) within a well’s drainage boundary and the rate of fluid
management. Studies have shown that motivations for real- withdrawal dictate the decline rate during PSS flow in a closed
time sensing revolve around on-time action to maximize system.
benefits. One complicating factor during the boundary-dominated
Various analysis techniques exist to analyze production flow is a well’s ever-changing outer boundaries precipitated
rate data for estimating in-place fluid volume and remaining by changing rates of neighboring wells, drilling infill wells,
reserves. These methods entail from traditional decline curve injecting fluids, and encroaching aquifer, to name a few. For a
analysis, such as those offered by Arps (1945) and Fetkovich perspective, Fig. 1 is intended as a practical diagnostic tool in
(1980) to more sophisticated techniques (Agarwal et al. 1999; a closed system for reservoirs with significant mobility
Blasingame et al. 1991; Blasingame et al. 1989; Mattar and producing gas or oil, and is not intended for tight-gas
McNeil 1997) involving both flowing bottomhole pressure and reservoirs.
2 SPE 100384

observation is that wells can be grouped to reveal

Infinite-acting Flow compartmentalization, if any, in a given reservoir.
pwf According to Appendix B, the slope of the p–q graph can
be represented by the following relation for a well in a
Plateau Period
multiwell reservoir
Constant-pwf (constant rate or
production tubing limited) dp wf ,k (t ) 1 dN p q k 70.6 Bμ dqT
= + (3)
dq k (t ) Nc t dq k qT kh dq k
Pseudosteady-State Flow
Note that the slope given by Eq. 3 is positive because dNp/dqk
term is always negative in PSS flow, thereby making all terms
positive on the right-hand side. As discussed in Appendix B,
the slope is essentially inversely proportional to the size of the
q container, N. Therefore, Eq. 3 serves as a powerful diagnostic
Fig. 1 – Diagnostic pwf–q graph for a closed system. tool for discerning compartmentalization.

As shown in Appendix A, the slope of the IA flow period Proof of Concept

during variable-rate production in a liquid system is given by Gas Flow in a Single-Well Reservoir. Let us consider a case
where dry-gas production occurs from a single-well reservoir.
⎧ n q −q ⎫ ⎧⎪ ⎛ ⎞ ⎫ This example is simulated with a commercial analytic
⎪ j −1 ⎪ dq n k ⎟ − 3.23 + 0.869 s ⎪⎬
∑
j
− m⎨ ⎬−m ⎨log⎜⎜ ⎟
package. Fig. 2 depicts a series of ccnstant-pressure-step input
t − t dt n ⎪⎩ ⎝ φμct rw
2
dp wfn ⎪⎩ j =1 n j −1 ⎪
⎭ ⎠ ⎭⎪ resulting in the corresponding rate profile.
=
dq n

pwf, psia
dq n 70 5,500
q, MMscf/D
dt n
60
(1) 4,500
50
In Eq. 1, the numerator remains negative, suggesting negative 3,500
slope for the pwf–q graph during the IA flow period. However, 40
,
for the constant-rate case, we have dqn/dtn = 0, thereby leading 30
dpwfn/dqn to infinity or the vertical line in Fig. 1. Obviously, 2,500
when a well is produced at constant-pwf, the slope attains a 20
zero value, as indicated by the dashed line. 1,500
10
In contrast to IA flow, the slope of the pwf–q graph, as
shown in Appendix A, can be written for variable-rate 0 500
production during pseudosteady-state flow (closed outer 0 3,000 6,000 9,000 12,000
boundaries) in a liquid system as
t, hr
0.2339 B
− qn Fig. 2 – Production history of the synthetic gas example.
dp wfn φhct A
= (2)
dq n dq In analogy to pressure-transient analysis, one needs to
− n identify flow regimes, infinite-acting, transitional, and PSS,
dt n
before embarking on decline-curve analysis. To this end, the
so-called inverted-Blasingame log-log diagnosis has been in
Inspection of Eq. 2 suggests that the slope is positive and is use for sometime. This diagnostic tool is constructed by
dominated by the reciprocal of CPV, φhA. Note that rate considering rate-normalized pressure integral and its
declines exponentially with time in a volumetric system. This derivative, as shown in Fig. 3. The graph’s name originates
fact implies that the dqn/dtn term in the denominator directly from the fact that instead of doing the rate-integral as in
influences the slope on the pwf–q graph because the numerator traditional Blasingame (Blasingame et al. 1989), pressure-
is a linear term. We also observe that the slope retains the integral is chosen so that it can be cast in terms of the familiar
same value so long the drainage volume remains contiguous. pressure analog. As shown in Fig. 3, the late-time unit-slope
In other words, whether one or ten wells drain the same CPV, line indicates the PSS flow period, whereas the early-time
the slope should remain the same in a closed volumetric plateau suggests infinite-acting period.
system. Experience shows that identifying plateau is very difficult
If reservoir compartmentalization occurs, wells draining while dealing with field data. Even syntheric data do not easily
from compartments of different CPV will exhibit different lend themselves to this response, as Fig. 3 indicates. Finding
slopes. Put another way, production performance of a well in a this plateau allows one to anchor the curve for type-curve
large drainage volume will exhibit a shallower slope than matching by minimizing the degrees of freedom, especially for
those in small drainage volumes. A corollary of this
SPE 100384 3

those cases where the PSS flow (unit-slope line) is not well
0.00016
developed and/or plagued by data noise.
Following the diagnostic step, one can interpret this dataset 0.00014

qg /{m(pi )-m(pwf )}, Mscf-cp/D-psi2

in the traditional decline-curve analysis mold with the
Blasingame type-curves, as shown in Fig. 4. This analysis 0.00012
allowed computation of original-gas-in-place at 22.3 Bscf, in
0.00010 y = -6E-06x + 0.0001
perfect harmony with the input value. Fig. 5 shows that one
can also use the Agarwal et al. (1999) approach to arrive at the R2 = 0.9825
0.00008
same result.
1.E+09 0.00006
Integral & Derivative, psia 2/cp

0.00004
Rate-Normalized Pressure

G i = 22.3 Bscf
0.00002
1.E+08
0.00000
Pseudosteady-state 0 5 10 15 20 25
G i {m(p i )-m(p )}/{m(p i )-m(p wf )}, Bscf
1.E+07 Infinite-acting
Fig. 5 – Estimating in-place gas with Agarwal et al. method.
5,000

1.E+06
4,000
1.E+01 1.E+02 1.E+03 1.E+04 1.E+05
t e , hr pwf, psia
3,000
Fig. 3 – Flow regime diagnosis with inverted-Blasingame plot.

2,000
1.E-07
1,000
PI, PI-Integral, PI-Integral

1.E-08
PI-Integral
PI 0
Derivative

0 20 40 60 80
1.E-09
PI-Integral q g , MMscf/D
Derivative
Fig. 6 – p–q diagnosis, synthetic gas example.
1.E-10
1600 4,500
o Δ X Well 1
1400
qo , STB/D or Rgo , scf/STB

1.E-11 p wf
10 102
103
10 4
105 — Well 2
1200 3,500
t e , hr
1000 R go

pwf, psia
Fig. 4 – Data analysis with Blasingame type curve.
800 2,500
In contrast to Fig. 3, the p–q graph shows the infinite- 600
acting period transitioning into the PSS flow period at late qo
times without having to do any calculations. Fig. 6 presents 400 1,500
the p–q diagnostic graph. Note that stair-steps are the 200
reflection of series of constant-pwf steps used to generate the
rate profile, as shown in Fig. 2. 0 500
Gas/Oil Flow in a Two-Well Reservoir. Let us consider a 0 15 30 45 60 75 90
case where two-phase gas/oil flow occurs in a two-well Producing Time, days
reservoir. Fig. 7 presents performance of both wells produced
Fig. 7 – Production performance with equal pwf, gas/oil example.
under the same pwf schedule in this homogeneous reservoir.
Model details are in Appendix C.
As expected, the two solutions overlap and the results are
presented in Fig. 8. Both the early-time IA and the late-time
PSS periods are readily apparent. When Well 2 is produced at
300 psi larger drawdown than Well 1, separation of production
performance occurs, as Fig. 9 testifies. Fig. 10, presenting the
4 SPE 100384

corresponding p–q graph, also indicates separation, but with Case Studies
the same slope. Note that changing pwf tantamount to In this section we present five case studies, three from
introducing heterogeneity or changing kh in Eq. 3. gas/condensate reservoirs, one from a dry-gas reservoir, and
one from an oil reservoir. Generally speaking, low
2,900 permeability with moderate condensate yield is common to
first three examples, located in very diverse geologic settings.
2,400 o Well 1 Offshore Gas/Condensate Reservoir, Field Example–1.
Δ Well 2 This example is drawn from a North Sea reservoir, as
pwf, psia

1,900 discussed by Ayyalasomayajula et al. (2005). Wells are

grouped in two areas, with a supposed barrier between the
1,400 two.
1,800
900
1,500
400
1,200
100 300 500 700 900

pwh, psig
q o , STB/D 900

Fig. 8 – Diagnostic p–q graph with equal pwf, gas/oil example. 600 m = 3 psi-D/MMscf

300 o C1, Δ C7, + C21, x C23

1,600 4,500
o Δ X Well 1
1,400 0
qo , STB/D; Rgo, scf/STB

— Well 2 0 20 40 60 80
1,200 p wf 3,500
q, MMscf/D
1,000 R go
pwf, psia

Fig. 11 – Four wells indicate same drainage volume in Area C ,

800 2,500 FE-1.
600
qo Fig. 11 shows that four wells from Area C essentially
400 1,500 replicate the PSS flow behavior amid scatter. Some of the data
200 scatter is contributed by shut-in pressures. In contrast, Fig. 12
exhibits that the two wells in Area B drain from another
0 500 reservoir compartment with higher pressures than those
0 15 30 45 60 75 90 located in Area C, as shown in Fig. 11. Equally important, six-
Producing Time, days fold increase in the slope in Fig. 12 than that in Fig. 11
suggests different drainage volumes. In other words, Area A
Fig. 9 – Production performance with unequal pwf, gas/oil
example. appears smaller than Area C because the larger slope indicates
higher rate of depletion.
2,900 Note that only wellhead pressures are available for these
wells, but they still underscore the point of connected-pore
2,400 o Well 1 volume. That is because pwf has one-to-one correlation with
Δ Well 2 pwh (Kabir and Hasan 2006) and should not ordinarily interfere
with diagnosis. Of course, the use of pwf removes any
pwf, psia

1,900
ambiguity. Perhaps Fig. 13 showing the average-reservoir
1,400 pressure in Area A lends credence to our contention of
connected-pore volume with unified depletion.
900

400
100 300 500 700 900
q o , STB/D

Fig. 10 – Diagnostic p–q graph with unequal pwf, gas/oil example.

SPE 100384 5

3,000 12,000
10,000
A1 8,000

kh, md-ft
2,000
pwh , psig

6,000

m = 17.76 psi-D/MMscf 4,000

1,000
A6 2,000
0
0 7/24/98 8/28/99 10/1/00 11/5/01 12/10/02
0 20 40 60 80 Date
q, MMscf/D Fig. 14 – Degradation of kh owing to subsidence, FE-2.
Fig. 12 – Two wells in A area indicate same but different drainage
volume than the wells located in Area C, FE-1.
6,000 5,000
Well #4
4,000
5,000

pwf, psig
3,000
p, psia

Well #3
4,000
2,000 0.0191q
pwf = 911.23e
2
3,000 1,000 Well #1 R = 0.8658

0
2,000
0 20 40 60 80
3/11/97 7/24/98 12/6/99 4/19/01 9/1/02 1/14/04
q g , MMscf/D
Date
Fig. 15 – Behavior of wells showing compartmentalization, FE-2.
Fig. 13 – Average reservoir pressure behavior of FE-1 wells in the
Area A.

Although condensate banking continues to occur with Net

time, the banking generally manifests in terms of skin at early Thickness
stages of depletion, resulting in reduced productivity. At late scale
times, production is dominated by the outer-boudary effects
Possible flow
when PSS flow sets in. In other words, depletion effect will barrier
obscure any near-wellbore issues, as indicated by steep decline
of average drainage-area pressure beyond 1998, as Fig. 13
testifies.
This fault
Deepwater Gas/Condensate Reservoir, Field Example–2. probably
seals, but #4
This reservoir experienced subsidence during its producing only to this #3
life. Fig. 14 makes this point with periodic buildup data, point??
Shale-filled
recorded with permanent downhole sensors in this three-well #1
channel
reservoir. This data set also suggests that changes in average-
reservoir pressure,⎯p, dwarfed the declining productivity This fault
probably
index, precipitated by subsidence. That is so because the wells doesn’t Distance
were in small compartments; the effect of extend past
scale: the N
this point
compartmentalization is underscored by different slopes for face of this
Wells 3 and 4 in Fig. 15. The figure exhibits that Well 4 and 1 block is 3
miles across
are in the same compartment, but Well 3 is in a smaller fault
block owing to its steeper slope than the other two. Fig. 16 Fig. 16 – Seismic A/B map showing barriers and reservoir
thickness, FE-2.
showing the seismic map corroborates the notion of
compartmentalization.
6 SPE 100384

Onshore Dry-Gas Reservoir, Field Example–3. This

6,000
example is taken from an onshore field with significant in-
place volume. Performance of only four wells is graphed to 5,000
illustrate reservoir compartmentalization. Fig. 17 shows that
Wells A-13 and A-8 decline at a rate faster than Wells A-9 4,000

pwh , psig
and A-12, following long infinite-acting flow duration,
exhibited by the negative slope lines. Note that the plateau 3,000
period, signified by the near-vertical line, in Wells A-9 and A- 2,000
12 lasts longer than the other two, thereby suggesting a larger
drainage volume. Of course, the lower absolute value of the 1,000
slope during PSS flow period confirms this notion.
When the traditional material-balance analysis ( p / z vs. 0
cumulative production) was attempted, care was taken to 0 50 100 150
consider those data that belong to the PSS period, meaning q g , MMscf/D
data with positive slope on the p–q graph. Note that inclusion
Fig. 18 – Five wells appear in the same compartment, FE-4.
of IA data leads to pessimistic estimation of the in-place
volume.

2,200

2,000

1,800
pwh , psig

+ A-13
Δ A-8
1,600
x A-9
o A-12
1,400
m = 21.3 psi-D/MMscf
m = 16.5 psi-D/MMscf
1,200
0 5 10 15 20 25
q g , MMscf/D
Fig. 19 – Discerning material-balance trend in Arun field, FE-4
Fig. 17 – Well performance suggests compartmentalization, FE-3. (after Marhaendrajana 2000).

Arun Field, Field Example–4. Many papers have appeared in Attaka Field, Field Example–5. This well is located in the
the literature on Arun field, located in Sumatra, Indonesia. Attaka oil field in East Kalimantan, Indonesia. Fig. 20
Some relevant papers include transient-test analysis displays the production profile, wheras Fig. 21 shows the
(Marhaendrajana et al. 1999), decline-curve analysis corresponding p–q graph. A gentle rate decline at early times
(Marhaendrajana and Blasingame 2001), and compositional- is represented by the near-vertical response, whereas the late-
reservoir simulation (Sutan-Assin et al. 1989). time data is characterized by the positive slope, signifying the
We digitized production data of five out of 11 wells from PSS flow period. Note that no negative slope or the IA flow is
Marhaendrajana’s dissertation (2000) and graphed them on apparent here. Also shown on Fig. 20 is the material-balance
Fig. 18. Long-duration IA period followed by the PSS flow line indicating a decent fit with the data.
period is readily apparent. Confluence of these wells amid Fig. 22 presents the same data on the inverted-Blasingame
scatter is quite striking. In fact, Fig. 19 shows that all 11 wells plot. Solid lines show an excellent visual fit with the material-
follow the same trend, particularly in the PSS period, as balance model on the log-log graph. Note that the dashed line
identified by the negative unit-slope line. While constructing signifies the unit-slope line for the entire dataset. We think
the pressure-drop-normalized rate/total-material-balance time that data diagnosis with the p–q graph is assuring in that one
graph, Marhaendrajana implicitly assumed that all the 11 wells can avoid seeking the plateau on the log-log graph that does
are in the same compartment. This observation is based on the not develop here.
fact that the total material-balance time presupposes the total
gas in the connected-pore volume. We think that the prudent
approach is to dignose the CPV with the p–q graph first before
doing any analysis to avoid bias.
SPE 100384 7

DCA on individual wells to obtain the total in-place volume.

2,500
This is so because sum of individual wells make up the total
pwf, psia or qo , STB/D

2,000 volume as shown in Appendix B and elsewhere

p wf (Marhaendrajana 2005).
1,500 Wells equipped with permanent sensors are ideal
qo candidates for the p–q diagnosis because no calculations are
1,000 involved. Experience suggests that most benefit of real-time
data is derived by discerning information from it so that rapid
500 business decisions can be made, proactively. In this respect,
our experience suggests that data frequency is a nonissue for
0 diagnosis. In other words, data frequency ranging from high
0 1,000 2,000 3,000 4,000 5,000 (minutes) to low (months) has no bearing on our ability to
Time, hr establish trends, provided sufficient number of data points are
available. The key ingredient is data synergy, however. Let us
Fig. 20 – Production performance of an Attaka well, FE-5. explore this point.
Because the p–q diagnosis depends entirely upon synergy
2,400 of two independent measurements, data incoherence manifests
itself in terms of noncompliance with the trend suggested by
2,200 Fig. 1. This issue is exemplified by a gas-well example, as
2,000 shown in Fig. 23. Here, the late-time pressure data shows
pwf, psia

unusual behavior, perhaps owing to liquid loading or other

1,800 wellbore-related effects. When this data are diagnosed, the
late-time data, identified by the lighter color, depart from the
1,600
recognized PSS norm. Fig. 24 exhibits the p–q diagnosis. In
1,400 this situation, the prudent approach is to ignore the data in
data calculated pink.
1,200
2,500
400 900 1,400 1,900
q o , STB/D
2,000
q, Mscf/D or pwh, psig

Fig. 21 – p–q diagnosis and material-balance calculation, FE-5.

1,500 qg
1000
Rate-Norm Pressure Integral

1,000
& Derivative, psia

p wh
100 500

0
10
1 10 100 1,000 10,000
Producing Time, days
1 Fig. 23 – Production history of a gas well.
10 100 1000 10000
t e , hr This example illustrates the value of p–q diagnosis before
embarking upon any analysis. Experience suggests that the
Fig. 22 – Data diagnosis with inverted-Blasingame plot, FE-5. p–q diagnosis on a Cartesian graph reveals any weakness in
data, because the scale compression associated with a log-log
Discussion graph is avoided. Nonetheless, Fig. 25 shows that the basic
The proposed diagnostic tool involving graphing of features, presented in the Cartesian graph, are retained on the
pressure/rate data on a Cartesian graph can provide a number log-log diagnosis.
of clues about reservoir behavior, including reservoir
compartmentalization. Therefore, we think this tool should be
precursor to any production-data analysis.
By identifying various compartments in a reservoir, one
can do decline-curve analysis (DCA) on a single well in each
compartment to obtain the total in-place gas. Alternatively,
when wells are within a CPV, one can use the total-material-
balance analysis along the lines shown in Fig. 19, or else do
8 SPE 100384

Acknowledgment
500
We are indebted to various operating units for providing field
examples and Chevron management for permission to publish
400 this work. One of us (CSK) greatly benefited from discussion
with Chih Cheng of Kappa Engineering concerning the
pwh , psig

300 inverted-Blasingame plot and providing the last two examples.

200 Nomenclature
A = drainage area, ft2
100 Bo = formation-volume factor, RB/STB
CA = Dietz shape factor for drainage area,
0 dimensionless
ct = total system compressibility, 1/psi
0 500 1,000 1,500 2,000 h = formation thickness, ft
q g , Mscf/D J = productivity index, MMscf/D-psi
k = permeability, md
Fig. 24 – p–q diagnosis of the gas-well example. m = semilog slope (=162.6 Bμ/kh), psi/log-cycle
m(p) = pseudopressure, psi2/cp
nw = number of wells
N = cumulative production of a well, STB
Np= cumulative production of the field, STB
pD= {kh(pi-p(x,y,t)}/141.2qrefBμ
100 pi = initial pressure, psia
pwh , psig

pwf = flowing bottomhole pressure, psia

pwh = flowing wellhead pressure, psig
⎯p = average drainage-area pressure, psia
q = oil rate, STB/D
qD = q(t)/qref
qg = gas rate, MMscf/D
qk = well rate, STB/D
1
qT = total field rate, STB/D
100 1,000 10,000 Q(t) = instantaneous cumulative production, MMscf
q g , Mscf/D reD= re/rw, drimensionless drainage area
rw = wellbore radius, ft
Fig. 25 – p–q diagnosis of the gas-well example on log-log rwa = apparent wellbore radius, ft
graph. Rgo= producing gas/oil ratio, scf/STB
Rs = solution gas/oil ratio, scf/STB
Conclusions
s = mechanical skin, dimensionless
1. A simple diagnostic tool in terms of pressure/rate
t = producing time, hr
graph is presented to discern reservoir flow behavior
Sgi = initial gas saturation, fraction of pore space
en route to understanding compartmentalization. The
Soi = initial oil saturation, fraction of pore space
negative and positive slopes on the p–q Cartesian
Swi = initial water saturation, fraction of pore space
graph signify infinite-acting and pseudosteady-state
te = equivalent producing time {=Q(t)/q(t)}, hr
flow periods, respectively. Analytic proof is
ttot = total field material-balance time, hr
presented in support of the observed behavior.
2. Multiple wells in the same field will exhibit the same tDA = 0.00633kt/φμctA, dimensionless
slope on the p–q graph, suggesting reservoir ttot = material-balance time, hr
connectivity. Dissimilar slopes suggest Tsc = temperature at standard conditions, oF
compartmentalization. An analytic expression is x = rectangular coordinate distance, ft
derived in support of this notion. Issues with y = rectangular coordinate distance, ft
condensate banking, reservoir subsidence, and areal xD = x/√A, dimensionless
heterogeneity, not considered in the analytic yD = y/√A, dimensionless
formulation, do not appear to have much impact on μ = viscosity, cp
understanding reservoir compartmentalization. γ = 0.577216, Euler’s constant
3. Both synthetic and multiple field examples verify the φ = porosity, fraction
notion presented in this study and the underlying τ = dummy integration variable
physical principles.
subscripts
i = constant rate
i = well number index
SPE 100384 9

j = timestep index Marhaendrajana, T. 2000. Modeling and Analysis of Flow Behavior

k = well position index in Single and Multiwell Bounded Reservoirs, PhD Dissertation,
n = nth flow period Texas A&M University, College Station, Texas.
0 = initial condition Mattar, L. and McNeil, R. 1997. The ‘Flowing’ Gas Material
Balance. J. Cdn. Pet. Tech. () 52.
1, 2, 3 = indices of flow period
Mattar, L. and Anderson, D.M. 2003. A Systematic and
Comprehensive Methodology for Advanced Analysis of
References Production Data. paper SPE 84472 prepared for presentation at
Agarwal, R.G., Gardner, D.C., Kleinsteiber, S.W., and Fussell, D.D. the SPE Annual Technical Conference and Exhibition, Denver,
1999. Analyzing Well Production Data Using Combined-Type- CO, 5-8 October.
Curve and Decline-Curve Analysis Concepts. SPEREE 2(5): 478– Onur, M. and Serra, K.V. 1988. Analysis of Pressure-Buildup Data
486. (SPE 57916-PA) from a Well in a Multiwell System. paper SPE 18123 prepared
Anderson, D.M. and Mattar, L. 2004. Practical Diagnostics Using for presentation at the SPE Annual Technical Conference and
Production Data and Flowing Pressures. paper SPE 89939 Exhibition, Houston, Texas, 2-5 October.
prepared for presentation at the SPE Annual Technical Rodriguez, F. and Cinco-Ley, H. 1993. A New Model for Production
Conference and Exhibition, Houston, Texas, 26-29 September. Decline. paper SPE 25480 prepared for presentation at the SPE
Arps, J.J. 1945. Analysis of Decline Curves. Trans., AIME 160, 228. Production Operations Symposium, Oklahoma City, OK, 21-23
Ayyalasomayajula, P.S., Silpngarmlers, N. and Kamath, J. 2005. March.
Well Deliverability Predictions for a North Sea Gas/Condensate Sutan-Assin, T., Rastogi, S.C., Abdullah, M., Hidayat, D., Bette, S.,
Reservoir. paper SPE 95529 prepared for presentation at the SPE and Heineman, R.F. 1989. Use of Compositional Simulation in
Annual Technical Conference and Exhibition, Dallas, Texas, 8-12 the Management of Arun Gas/Condensate Reservoir. paper SPE
October. 17683 prepared for presentation at the SPE Offshore South East
Bigno, Y., Baillie, J.M., and Coombes, T.P. 1998. The Interpretation Asia Show, 2-5 February, Singapore.
of Reservoir Pressure Data in the Dunbar Field (UKCS). SPEREE
1(5): 384–391. Appendix A – Slopes of Infinite-Acting and
Blasingame, T.A., McCray, T.L., and Lee, W.J. 1991. Decline Curve Pseudosteady-state Flow Periods
Analysis for Variable Pressure Drop/Variable Flow Rate Systems.
paper SPE 21513 prepared for presentation at the SPE Gas
Infinite-Acting Flow Period. During infinite-acting flow
Technology Symposium, Houston, Texas, January 22-23. period, the generalized multirate superposition equation for
Blasingame, T.A., Johnston, J.L., and Lee, W.J. 1989. Type-Curve liquids is given by (Earlougher 1977)
Analysis Using the Pressure Integral Method. paper SPE 18799
μB ⎧⎪ ⎫
prepared for presentation at the SPE California Regional Meeting, n
pi − pwf q j − q j −1
Bakersfield, CA, 5-7 April.
qn
= 162.6 ⎨ ∑
kh ⎪ j =1 qn
( )⎪
log tn − t j −1 ⎬
Blasingame, T.A. and Lee, W.J. 1986. Variable-Rate Reservoir ⎪⎭
Limits Testing. paper SPE 15028 prepared for presentation at the
⎩ (A-1)
SPE Permian Basin Oil and Gas Recovery Conference, Midland, μB ⎧⎪ ⎛⎜ k ⎞⎟ ⎫⎪
Texas, 13-14 March. + 162.6 ⎨log⎜ − 3 .23 + 0 . 869 s ⎬
kh ⎪ ⎝ φμct rw2 ⎟⎠ ⎪⎭
Dietz, D.N. 1965. Determination of Average Reservoir Pressure ⎩
From Build-Up Surveys. JPT 27(8): 955–959; Trans., AIME,
234. Setting m = 162.6μB/kh, and rearranging Eq. A-1, we obtain
Earlougher, R.C., Jr. 1977. Advances in Well Test Analysis. SPE
Monograph Series, 5, 31, Richardson, TX.
⎧ n ⎫
∑( ) ( )
Fetkovich, M.J. 1980. Decline Curve Analysis Using Type Curves. ⎪ ⎪
JPT (June 1980) 1065. pwf = pi − m⎨ q j − q j −1 log tn − t j −1 ⎬
Gringarten, A.C. 2005. Analysis of an Extended Well Test to Identify ⎪⎩ j =1 ⎪⎭
(A-2)
Connectivity Between Adjacent Compartments in a North Sea
⎧⎪ ⎛ k ⎞ ⎫
Reservoir. paper SPE 93988 prepared for presentation at the SPE − mqn ⎨log⎜ ⎟ − 3.23 + 0.869s ⎪⎬
⎪⎩ ⎜⎝ φμct rw ⎟⎠
Europec/EAGE Annual Conference, Madrid, Spain, 13-16 June. 2
⎪⎭
Kabir, C.S. and Hasan, A.R. 2006. Simplified Wellbore Flow
Modeling in Gas/Condensate Systems. SPEPO 21(1): 89–97.
Marhaendrajana, T. 2005. A Novel Approach for the Evaluation of Using the chain rule of differentiation, one can write
Oil and Gas Well Performances in Multiwell Reservoir Systems.
paper SPE 93222 prepared for presentation at the SPE Asia dp wf dp wf dq
Pacific Oil and & Gas Conference and Exhibition, Jakarta, = (A-3)
Indonesia, 5-7 April. dt dq dt
Marhaendrajana, T. and Blasingame, T.A. 2001. Decline Curve
Analysis Using Type Curves―Evaluation of Well Performance Rearranging Eq. A-3, we have
Behavior in a Multiwell Reservoir System. paper SPE 71517
prepared for presentation at the SPE Annual Technical
Conference and Exhibition, New Orleans, Louisiana, 30 dp wf dp wf / dt
= (A-4)
September–3 October. dq dq / dt
Marhaendrajana, T., Kaczorowski, N.J., and Blasingame, T.A. 1999.
Analysis and Interpretation of Well test Performance at Arun
Field, Indonesia. paper SPE 56487 prepared for presentation at Differentiating Eq. A-2 with respect to t1, t2, and t3, we obtain
the SPE Annual Technical Conference and Exhibition, Houston,
Texas, 3-6 October.
10 SPE 100384

d dp wf
{(q1 − q 0 ) log(t1 − t 0 )} = 1 (q1 − q 0 ) (A-5) =−
0.2339 B
qn (A-12)
dt1 (t1 − t 0 ) dt n φhct A

d Differentiating Eq. A-11 with respect to t1, t2, etc., we can

{(q1 − q 0 ) log(t 2 − t 0 ) + (q 2 − q1 ) log(t 2 − t1 )} write
dt 2
(A-6)
1
= (q1 − q 0 ) + 1 (q 2 − q1 ) d{q1 (t1 − t 0 )}
= q1
(t 2 − t 0 ) (t 2 − t1 ) (A-13)
dt1

d
{(q1 − q 0 ) log(t 3 − t 0 ) + (q 2 − q1 ) log(t 3 − t1 ) for timestep 1, and
dt 3
+ (q3 − q 2 ) log(t 3 − t 2 )} (A-7) d{q1 (t1 − t 0 ) + q 2 (t 2 − t1 )}
= q2 (A-14)
(q − q0 ) (q 2 − q1 ) (q3 − q2 ) dt 2
= 1 + +
(t 3 − t 0 ) (t 3 − t1 ) (t 3 − t 2 )
for timestep 2, and so on.
Therefore, the general form of time derivative can be written
Therefore, the general form of time-derivative can be written
as
as
⎧ n ⎫ n
d ⎪
∑ ( )⎪⎬ = ∑ qt j −− tq j −1 ⎧ n ⎫
⎨ (q j − q j −1 ) log t n − t j −1
∑ ( )⎪⎬ = q n
(A-8) d ⎪
dt n ⎪ ⎪⎭ n j −1 ⎨ q j t j − t j −1 (A-15)
⎩ j =1 j =1 dt ⎪ ⎪⎭
⎩ j =1
Combining Eqs. A-4 and A-8, we have
In other words,
⎧ n q −q ⎫ ⎧⎪ ⎛ ⎞ ⎫
⎪ j −1 ⎪ dq n k ⎟ − 3.23 + 0.869 s ⎪⎬ dq q 2 − q1
∑
j
− m⎨ ⎬−m ⎨log⎜⎜ ⎟ = (A-16)
dp wfn t
⎪⎩ j =1 n − t j −1 ⎪ dt n ⎪⎩ ⎝ φμct rw
2
⎠ ⎪⎭ dt t 2 − t1
= ⎭
dq n dq n
dt n Eq. 15 is negative because the numerator is negative, while the
(A-9) denominator is always positive. Therefore, combining
Eqs. A-4, A-12, and A-16, we obtain
For the constant-rate case, we have dqn/dtn = 0, thereby
leading dpwfn/dqn to infinity or the vertical line on the pwf/q 0.2339 B
− qn
graph. Otherwise, the numerator remains negative in Eq. A-9, dp wfn φhct A
suggesting negative slope for the pwf/q graph during infinite- = (A-17)
dq n dq
acting flow. Obviously, when a well is produced at constant- − n
pwf, the slope attains a zero value. dt n
PSS Flow Period. During PSS flow, the general governing
flow equation for variable-rate liquid flow is given by Eq. A-17 shows that the positive slope will be exhibited when
(Blasingame and Lee, 1986) pwf is graphed against q.

Appendix B – Derivation of p-q Slope for Individual

Bμ ⎛⎜ ⎞ 0.2339 B
pi − p wf n q j (t j − t j −1 )
4A
qn
= 70.6 ln
kh ⎜ e γ C A rw 2
⎟+
⎟ φhct A ∑ qn
Wells in a Multiwell Reservoir
The mathematical model describing the pressure behavior in a
⎝ ⎠ j =1
(A-10) bounded rectangular reservoir with multiple wells can be
derived analytically. In this formulation, each well produces at
Rearranging Eq. A-10, one obtains an arbitrary, constant rate and wells are located at arbitrary
positions in the reservoir. The model is given by
Bμq n ⎛⎜ 4A ⎞ 0.2339 B n
p wf = pi − 70.6 ln ⎟−
∑
q j (t j − t j −1 ) d2 p d2 p
nw
φμct dp
∑ Ahi(k / μ )δ (x − x w,i , y − y w,i ) =
kh ⎜ γ 2 ⎟ φhct A j =1 q (t ) B
⎝ e C A rw ⎠ + −
2 2 k dt
dx dy i =1
(A-11) (B-1)

Differentiating Eq. A-11 with respect to time, one obtains Implicit in Eq. B-1 are the usual assumptions involving
closed rectangular boundaries, fully-penetrating vertical wells,
SPE 100384 11

and flow of slightly compressible single-phase liquid in a pi − p wf ,k (t ) 1 1 1

homogeneous reservoir. Writing Eq. B-1 in dimensionless = t tot +
q k (t ) Nct q k (t ) Nct
form yields (Onur and Serra 1991; Rodriguez and Cinco-Ley
1993) t nw

∫∑ q(τ ) F {(x w,k + ε ), ( y w,k + ε ), (t − τ ), x wi , y wi }dτ

2 2 nw 0i =1
∑ q D ,i (t DA )δ (x D − xwD,i , y D − y wD,i )
d pD d pD
+ + 2π 141.2 Bμ
dx D 2 dy D 2 i =1 + s
kh
dp D
= (B-7)
dt DA
(B-2) where

This work follows the solutions of Eq. B-2 given by ttot is the material-balance time and is given by
Marhaendrajana (2000), Marhaendrajana and Blasingame
(2001), and Marhaendrajana (2005). t n
1 N p ,tot
Using Duhamel’s principle for variable rate and variable
pressure in a closed system results in the following expression
t tot ,k =
q k (t ) ∫∑
i =1
qi (τ )dτ =
q k (t )
(B-8)
0

nw t DA
⎛ x D , y D , t DA − ⎞
p D ( x D , y D , t DA ) = 2π ∑ ∫ q D,i (τ )ψ i ⎜⎜⎝τ , xwD ,i , y wD ,i ⎟⎟⎠dτ and Np,tot represents the cumulative oil production from the
i =1 0 entire field. Simplifying Eq. B-7, we obtain
(B-3)
pi − p wf ,k (t ) 1
where ψ i ( x D , y D , t DA , x wD , y wD ) is an instantaneous q k (t )
=
Nct
t tot + f (t ) (B-9)
line-source solution. For a single well, Eq. B-3 transforms into
In Eq. B-9, the variable f(t) is time dependent. However, this
t DA
f(t) variable becomes constant during boundary-dominated
p D ,cr ( x D , y D , t DA ) = 2π ∫ ψ (x i D, y D , t DA − τ , x wD , y wD )dτ flow. The reciprocal of Eq. B-9 is the general formulation of
0 Arps’ (1945) harmonic decline. Note that Eq. B-9 takes into
(B-4) account the complexity of production schedule; that is,
constant pressure, constant rate, or variable pressure and rate.
Defining ξ=tDA- τ , and taking the derivative of Eq. B-4 Recasting Eq. B-9 in dimensionless variables and by type-
with respect to tDA, one can obtain convolution integral curve matching field data, one can estimate the skin factor.
formulation of the pressure response at any well location in a Let us recall the solution of boundary-dominated flow for a
multiwell reservoir system as follows single well

nw t DA q (t ) 1
dp D,cr ⎛ x D , y D , t DA ⎞ =
∑ ∫ ⎜ ⎟dτ (B-10)
p D ( x D , y D , t DA ) = q D ,i (τ ) ⎜ ⎟ p i − p wf ,k (t ) t
dt DA ⎝ − τ , x , y
wD,i wD,i ⎠ + b pss
i =1 0 Nct
(B-5)
where
The pressure solution for well k, including the near-
wellbore skin factor is given by
Bμ ⎧⎪ 1 ⎛ 4 A ⎞⎫⎪
b pss = 141.2 ⎨ ln⎜⎜ γ ⎟⎬
{
p D ( x wD,k + ε ), ( y wD,k + ε ), t DA = } kh ⎪⎩ 2 ⎝ e C A rwa
2 ⎟⎪
⎠⎭
(B-11)

nw t DA
⎧ dp , (t ,τ ) ⎫
∑ ∫ q D,i (τ )⎨ D cr DA ⎬ dτ + q Dk (t DA ) s,k
⎩ dτ ⎭ k,i
For the multiwell case, the Dietz (1965) shape factor, CA, is
determined by both reservoir shape and well position, as well
i =1 0
(B-6) as by the state of other wells; that is, their number, position,
rate, and pressure. The apparent drainage area, rwa, of a well in
In Eq. B-6, the solution inside the integral is computed for a a multiwell system is the ratio of its producing rate to total
particular well that includes the effects of other wells in the field withdrawal rate. Defining this ratio as qk/qtot, Eq. B-9 can
reservoir. Eq. B-6 can be written as be written as
12 SPE 100384

q k (t ) 1 dp wf ,k (t ) 1
= (B-12) =− (B-21)
pi − p wf ,k (t ) 1 dN p Nc t
t tot + b pss
Nct
For PSS flow, we have
where bpss for a multiwell system is given by
dp wf 1 dN p C q k
=− +
141.2 Bμ ⎛⎜ 1⎞ ⎛ 1⎞
reD reD (B-22)
b pss = ln − ⎟ = C ⎜ ln − ⎟ dqT Nc t dqT 2 qT
kh ⎜ qT / q k 2 ⎟⎠ ⎜ qT / q k 2 ⎟⎠
⎝ ⎝
Using the chain rule of differentiation,
(B-13)
dqT dqT dq k
= (B-23)
Note that Eq. B-12 is the form given by Fetkovich (1980) for a dt dq k dt
single-well reservoir, whereas Eq. B-13 represents its
extension to multiwell reservoirs (Marhaendrajana 2000). where
Using the final form of the multiwell solution, we develop the
slope of the p–q plot, which allows diagnosis of reservoir’s dqT dqT / dt
connectivity. Rearranging Eq. B-12 and noting that ttot = = (B-24)
dq k dq k / dt
Np/qk,
Combining Eqs. B-19 through B-24, we can write
Np
p wf ,k (t ) = p i − − q k b pss (B-14)
Nc t dp wf ,k (t ) ⎛ 1 dN p C q k ⎞ dq T 1 dN p
= ⎜− + ⎟ −
d q k (t ) ⎜ Nc dq 2 qT ⎟ dq
or, ⎝ t T ⎠ k Nct dq k
(B-25)
Np ⎛ reD 1⎞
p wf , k (t ) = p i − − q k C ⎜ ln − ⎟ (B-15) Substituting for C, we have the final form given by
Nc t ⎜ qT / q k 2 ⎟⎠
⎝
dpwf , k (t ) ⎧ 1 dN p qk 1 141.2 Bμ ⎫ dqT
= ⎨− + ⎬
We note that dqk (t ) ⎩ Nct dqT qT 2 kh ⎭ dqk
(B-26)
1 dN p
p wf ,k (t ) = f (q k , qT , N p ) (B-16) −
Nct dqk

where
Combining Eqs. B-26 and B-20, and rearranging, we obtain
N p = f ( qT ) (B-17)
dp wf ,k (t ) 1 dN p q k 70.6 Bμ dqT
= + (B-27)
and d q k (t ) Nc t dq k qT kh dq k

qT = f ( q k ) (B-18) Eq. B-27 presupposes that no reservoir compartmentalization

occurs. Nonetheless, in a compartmentalized reservoir,
Eq. B-27 can be used for each compartment. Suppose in a
Thorefore, we can write
reservoir of two compartments with one well in each, we
observe the following in each compartment:
dp wf ,k (t ) dp wf dqT dp wf dN p 1. In the second term on the right side of Eq. B-27, we
= + (B-19)
dq k (t ) dqT dq k dN p dq k have, dqT/dqk = 1 and qk/qT = 1,
2. With the fluid and rock properties being the same, the
and term 70.6Bμ/kh will remain constant,
3. In the first term on the right side of Eq. B-27, the
changes in cumulative production with rate; that is,
dNp dN p dqT dNp/dqk will be small during PSS flow.
= (B-20)
dq k dqT dq k In other words, Eq. B-27 suggests that the slope on the p–q
graph will be dominated by the reciprocal of in-place volume,
Note that N. That is why a well in a smaller fault block exhibits a steeper
slope than that in a larger compartment.
SPE 100384 13

Perhaps another way of looking at Eq. B-27 is by

considering constant-rate production of a single well in closed
system. Differentiating the PSS solution (Earlougher 1977) for
liquids, we can write

dp wf 0.234 B 70.6 Bμ ⎡⎛⎜ A ⎞ ⎤

⎟ + ln⎛⎜ 2.2458 ⎞⎟ + 2s ⎥
= + ⎢ ln
dq φc t hA kh ⎢⎜ rw 2 ⎟ ⎜ ⎟ ⎥
⎣⎝ ⎠ ⎝ CA ⎠ ⎦
(B-28)

Inspection of Eq. B-28 suggests that the slope is dominated by

the reciprocal of connected-pore volume, φhA .

Appendix C– Model Parameters for the Two-Phase

Flow Example
The Cartesian grids of 5×5×5 describe the model. The areal
grids are 100×100 ft with the vertical grid of 20 ft each.
Porosity is 20% and horizontal and vertical permeability
values are 5 md and 0.5 md, respectively. Saturations at the
endpoints are Soi = 0.8, Swi = 0.2, and Sgi = 0.0. Fig. C–1
presents the oil properties for the system.

600 1.4
500
1.3
400
Bo , RB/STB
Rs, scf/STB

300 1.2
200
1.1
100
0 1
0 1,000 2,000 3,000 4,000 5,000
Pressure, psia

Fig. C–1 – Oil properties for the two-phase flow problem.