Professional Documents
Culture Documents
Copyright 2008, Society of Petroleum Engineers Inc. dominant flow regime, we can now decide which of
This paper was prepared for presentation at the 32nd Annual SPE these techniques to be used.
International Technical Conference and Exhibition in Abuja, Nigeria,
4–6 August 2008.
Finally, the concepts were applied to field rate-time
This paper was selected for presentation by an SPE Program data.
Committee following review of information contained in an abstract
submitted by the author(s). Contents of the paper, as presented, have Introduction
not been reviewed by the Society of Petroleum Engineers and are
subject to correction by the author(s). The material, as presented, does Decline curve analysis (DCA) as one of the methods of
not necessarily reflect any position of the Society of Petroleum production rate-time data analysis has advanced
Engineers, its officers, or members. Papers presented at SPE meetings significantly, over the past years.
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 for commercial purposes without This paper starts with the basic theories of Decline
the written consent of the Society of Petroleum Engineers is Curve Analysis based on the assumption that “whatever
prohibited. Permission to reproduce in print is restricted to an abstract causes controlled the trend of a curve in the past will
of not more than 300 words; illustrations may not be copied. The
abstract must contain conspicuous acknowledgement of where and by
continue to govern its trend in the future in a uniform
whom the paper was presented. Write Librarian, SPE, P.O. Box manner” and proceed to detailed explanation of the
833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435. DCA associated models available in most production
decline surveillance tools. There is clarification on
when and where to reinitialise time to zero on the data
Abstract points and also a discussion on how to determine the
Decline curve analysis is a graphical procedure used dominant flow regime from the production data. The
for analyzing declining production rates, and sequential and systematic application of these concepts
forecasting future performance of oil and gas wells. in a field case gives consistent and reliable results.
This is achieved by curve fitting the past production
performance using the rate-time data and extrapolating This is evident in the results obtained from different
it to predict future performance, with the primary aim techniques (including dynamic simulation) when
of estimating reservoir remaining reserves and/or compared with production information.
remaining productive life.
Basic Theory of DCA
The effective use of the forecast techniques: Empirical, There are two kinds of DCA techniques namely:
Fetkovich, Locke & Sawyer, and Analytical Transient
solutions for oil and gas wells/Reservoirs, using a • Empirical /Traditional (Arps) decline analysis
production surveillance tool that manages multi-wells technique - the classical curve fit of historical
effectively in less time is discussed. The last three production data.
techniques-type-curves require good understanding of • Type curve matching technique.
the reservoir model as well as the parameters
controlling the well behaviour. To effectively use any of
these techniques for forecasting future production, it is 1.2.1 Empirical Decline Curve Analysis
imperative to determine the dominant flow regime from
Rate-time decline curve extrapolation is one of the
the production data.
oldest and most often used tools of the petroleum
engineer. The various methods used always have been
This paper focuses on concepts that enable engineers
regarded as strictly empirical and generally not
determine the dominant flow regime from diagnostic
scientific1
plots generated from rate-time data and/or numerical
iterative method. Once we are able to establish the
2 Aderemi S and Akpara K SPE 119732
Effective use of Production Surveillance Tool in Forecasting Future Production
Nearly all conventional DCA is based on empirical The importance of dimensionless variables is that they
relationship of production rate versus time given by simplify the reservoir models by embodying the
Arps2 (1945) as: reservoir parameters, thereby reducing the total number
qi of unknowns. They have the additional advantage of
qt = Eqn. 1 providing model solutions that are independent of any
(1 + bDi t ) b
1
particular unit system; a specific example is shown in
The mathematical description of these production Fetcovich Type Curve.4,5
decline curves is greatly simplified with the use of the
instantaneous (nominal) decline rate. Fetkovich Type Curve
d (ln q ) 1 dq Fetkovich generated two different sets of decline type
D=− =− Eqn. 2 curves, the transient stems and the boundary-dominated
dt q dt stems. The transition from the former to the later occurs
The minus sign has been added since dq and dt have at tDd = 0.3.
opposite sign and it is convenient to have the declined
rate, D always positive. Transient flow Boundary-dominated
“early-time” flow
Arps equation can be conveniently expressed in the
following three (3) forms. (Table 1 and Appendix I)
(q )
b
Hyperbolic 0<b qi Q=
qi 1− b
− qt
1− b
1.2.2.1 Boundary Dominated Region
qt = (1 − b )Di i
<1 (1 + bDi t ) b
1
In the well testing domain, an empirical solution for a 1.2.3 Noise Effect
well producing from the centre of a cylinder reservoir
at constant sandface flowing pressure in terms of the Production data are often noisy, thus difficult to
dimensionless flow rate, qD and the dimensionless time, analyze. To reduce the effect of this noise, the
tD are defined as follows: cumulative production can be used. It is a much
smoother curve than production data, and can make the
141.2qBμ analysis more reliable.
qD = Eqn. 7
kh( Pi − Pwf ) Fraim and Lee9,10,11 developed cumulative type curves
of QDd versus tDd, where QDd is the dimensionless
0.006329kt cumulative production, defined as the ratio of
tD =
φμC r rwa2 cumulative production to the ultimate movable fluid,
and tDd is the dimensionless time defined in type curve-
Eqn. 8
Arps.
Oil
Fetkovich investigated production decline and
BN p1.787 Np Eqn. 14
generated the transient and boundary dominated data QDd = =
for qD vs. tD used for both early-time and late-time ( )
Ct hφ re2 − rw2 (Pi − Pwf ) N net
curves. The solution was first studied by Van Gas
Everdingen and Hurst, 7 who provided the solution in 637.8 Psc TG p Gp Eqn. 15
Q Dd =
Laplace space, as follows: ( )
Tsc hφμ ct re2 − rw2 (ψ i − ψ wf )= G pnet
s
( ) ( )
⎡ I s RD K1 s − K1 s RD I1 s ⎤
qD = L−1 ⎢ 1
( )( ) Eqn. 9 t Dd =
tD Eqn. 16
( )( ) ⎥
( ) 1 ⎡⎛ re ⎞ ⎤⎡ ⎛ r ⎞ 3 ⎤
2
⎣ K1 s RD I 0 s − I1 ( s RD ) K 0 s ⎦ ⎢⎜ ⎟ − 1⎥ ⎢ln⎜⎜ e ⎟⎟ − ⎥
2 ⎢⎜⎝ rw ⎟⎠ ⎥⎦ ⎣⎢ ⎝ rw ⎠ 4 ⎦⎥
⎣
For tD < 200, the transient portion of these flow rate
equations can be represented by numerical curve fit
equations, reported by Edwardson et al.8 (1962) as Decline Curve Models of interest.
The production surveillance tool used for analyses in
26.7544+ 43.5537tD + 13.3813tD + 0.492949tD
0.5 1.5
qD = Eqn. 10 this paper captures not only the empirical models, but
47.421tD0.5 + 35.5372tD + 2.60967t1D.5
also the scientific models/techniques of evaluating
DCA. It captures three industry standard TCAs and the
For tD > 200, dimensionless rate becomes workflow for each is fairly similar.12
3.90086 + 2.02623t D (ln t D − 1) •
qD = Eqn. 11 Fetkovich
t D (ln t D ) •
2
Analytical Transient
• Locke & Sawyer
In a bid to simplify application of eqns. 7 through 11,
Fetcovich applied transformations on the dimensionless This tool also allows other common types of
variables (rate and time) as shown in Eqn 12 and Eqn forecasting approaches using other variables such as
13. These dependent parameters are called decline water cut, cumulative production, and P/Z etc.
curve dimensionless rate, qdD and time, tdD The TCA require a better understanding of the reservoir
model and the parameters controlling the well
⎡ ⎛ r ⎞ 1⎤ Eqn. 12 behaviour. Unwanted results most times are due to
qdD = qD ⎢ln⎜⎜ e ⎟⎟ − ⎥ incorrectly specified reservoir and well parameters.
⎣⎢ ⎝ rw ⎠ 2 ⎦⎥
TCA does work because is founded on the basics of
fluid flow principles (same used in pressure transient
tD analysis).
t dD =
1 ⎡⎛ re ⎞ ⎤⎡ ⎛ r ⎞ 1 ⎤
2
⎢⎜ ⎟ − 1⎥ ⎢ln⎜⎜ e ⎟⎟ − ⎥
2 ⎢⎜⎝ rw ⎟⎠
Table 2: Summary of associated applications to
⎣ ⎥⎦ ⎢⎣ ⎝ rw ⎠ 2 ⎥⎦
well/reservoir types
4 Aderemi S and Akpara K SPE 119732
Effective use of Production Surveillance Tool in Forecasting Future Production
Fetkovich Analytical Transient Locke & (iii) Transient data for Fetkovich type curves, lie at
Sawyer least Td D <= 0.3
• Conventional • Dual Porosity • Fractured
reservoirs • Fracture wells injector Effective Forecast Workflow
• Radial • Horizontal wells wells.
Reservoir • Rectangular finite • Radial Any decline curve is characterized by three factors:
• Finite system systems or infinite systems
• Single • Linear pressure • Initial production rate qi
porosity variation • Curvature of the decline, b
• More detailed than • Rate of decline, D
Fetkovich.
Each value of qi, Di and b will produce its own unique
Constant Rate Period curve. It is therefore important to estimate these
Reinitialize time to zero in this parameters before embarking on production forecast
region using DCA concept. Initial Rate of Decline is the
easiest to determine from the rate time data.
Decline
Log
rate
Methodology
Build up
Methods of Estimating Curvature of Decline, b
Log Time
Figure 2: Ideal Rate-Time plot on a log-log scale. 3.1.1 Graphical Method
Fetkovich(1980) pointed out that that there are several The graphical presentation of this type of decline curve
obvious situations where rate-time data must be indicates a linear relationship when plots of:
reinitialised for reasons that include: • Flow rate (qt) is plotted against time (t) on a
semilog scale.
• The prevailing drive or production mechanism
has changed. ln (qt ) = ln (q i ) − Di t Eqn. 17
• An abrupt change in the number of wells on a • Flow rate (qt) versus cumulative production
lease or a field due to infill drilling. (Np or Gp) is plotted on a Cartesian scale.
• Changing the size of tubing would change qi q t = q i − Di N p ( t ) Eqn. 18
and also the decline exponent b. This is the easiest, simplest method to apply and
perhaps the most conservative method of DCA to use.
• Others are changes in the reservoir
This is the linearized exponential decline method.
characteristics, saturation changes, relative
permeability, choke setting and workovers.
3.1.1.2 Case 2, b = 1
The graphical presentation of production recovery
It is good practice to ‘re-initialize” data to the first performance of a hydrocarbon system that follows a
point, or close to where the rates start declining (Figure harmonic decline will give a straight line when a plot
2). of:
• 1 vs t on a Cartesian scale with a slope of
2.1.2 Rule of Thumb to suspect transient Data qt
⎛ Di ⎞
⎜ q ⎟
Flow regime is suspected to be in the transient state ⎝ i ⎠
Eqn. 25
Starting with an initial value of b=0.5
i.e. b = 0.5 , the method will usually converge
k
Figure 1: Graph of q-b versus time (hyperbolic 6. Solve for Di by solving Eqn for Di and using
decline). the calculated value of b from step 5 and the
coordinates of a point on the smooth graph, i.e
If the correct value of b that yields the best straight line
(t 2 , q 2 ) to give:
lies between 0 and 1 (0<b<1), it shows that the
boundary dominated flow is dominant (i.e. b<0<1) else (qi q2 ) − 1
b
The following iterative method is designed to Gentry14 (1972) developed a graphical method for the
determine Di and b from the historical production data3. coefficients b and Di as shown in Figure 4 and 5.
6 Aderemi S and Akpara K SPE 119732
Effective use of Production Surveillance Tool in Forecasting Future Production
Case 1
Forecast starts from BOH and ends at 2 years before
EOH. There is agreement between production forecast
and 2 year history prior to EOH (See Figure 11)
Curvature of decline is estimated by applying
procedures in sections 3.1.2 and 3.1.1.3.
Case 2
Forecast start from EOH through 60 months (See
The plots of gas rate vs cumulative production and gas
Figure 12).
rate vs. time on semi-log scale show neither
exponential nor harmonic decline as both plots yield no
4) Well_3 produced for thirteen (13) year period. Same
straight line (Figure 6). To check the exactness of
procedure as in examples 1 and 2 are applied to
estimated values of b and Di, 0.51952 and 0.3667/year
generate correct decline curvature, b ≈ 0.6. Comparison
respectively using methodology in section 3.1.2, we
in the result of production forecast for the hyperbolic
apply section 3.1.1.3 for values of b> 0.51952, b=
decline and Fetcovich model is made on this data since
0.51952 and b < 0.51952. The correct value of b is
both models are inherent in boundary dominated
confirmed as shown n figure 7. This tells us that the
regime (See figures 13 through 15 and Table 5).
field case is that of boundary dominated flow with a
decline curvature of 0.51952. This paves way for
5) Field has 12 production wells with 8 months
accurate forecast of knowing that the principal
production period. Estimation of decline curvature
parameters are correctly defined.
indicates transient flow regime is dominant ( i.e b>1).
Dynamic simulation based prediction is used to
2) This case example shows limited data for a well
compare with Analytical transient type curve. This
under transient flow (Figure 8 and Table 4). For this
forecast estimate agrees pretty well for twenty years
case, forecast is assisted by including reservoir data
(Figure 16).
(storativity and rock and fluid compressibility) in
analytical transient and forecast is made for 120
months.
Conclusion
Figure 4: Relationship between production rate and cumulative production (After Gentry, 1972)
Figure 5: Relationship between production rate and time (After Gentry, 1972)
9 Aderemi S and Akpara K SPE 119732
Effective use of Production Surveillance Tool in Forecasting Future Production
NOMENCLATURE
μ= viscosity (cp)
k = permeability (md)
μ = Viscosity, cp
APPENDICES
q
− Dt = ln bDi t Di t 1 1
qi
b
= q −b − qib = −
qi qi qi qt
q = q i * e − Dt
q = q i (1 + bDi t ) q = qi (1 + Di t )
−1 −1
b
Rate t t t t t t
Q = ∫ q * dt = ∫ qi * e − Dt * dt Q = ∫ q * dt = ∫ q i (1 + bDi t ) * dt Q = ∫ q * dt = ∫ q i (1 + Di t ) * dt
−1 −1
b
Cumulative
0 0 0 0 0 0
qi − qi * e − Dt qi ⎡( )
b −1
⎤
Q= Q=
(1 − b )Di ⎢⎣
1 + bDi t b −1
⎥⎦ Q=
qi
[ln(1 + Di t )]
D D
Substitute from Rate-time Substitute from Rate-time
equation: equation: Substitute from Rate-time
b
Figure 6: Plots of gas rate vs Gp and gas rate vs time on semi-log scale.
q-b
q-b
q-b
Figure 9: Results of 10years forecast using Analytical Transient Type Curve Approach for well_1
14 Aderemi S and Akpara K SPE 119732
Effective use of Production Surveillance Tool in Forecasting Future Production
Figure 10: Plots of gas rate vs Gp and gas rate vs time on semi-log scale
Figure 13: Plots of gas rate vs Gp and gas rate vs time on semi-log scale
Figure 14: Forecast Results for Well_14 using Hyperbolic Decline Analysis
Figure 15: Forecast Results for Well_14 using Fetkovich’s Type Curve Analysis
16 Aderemi S and Akpara K SPE 119732
Effective use of Production Surveillance Tool in Forecasting Future Production
Cum. Cum.
Rate Rate Difference
Production Production
Date bbl/d Mbbl bbl/d Mbbl Mbbl
1 2000/07 8.13 0.252 7.79 0.242 0.011
2 2000/08 8.061 0.502 7.715 0.481 0.021
3 2000/09 7.994 0.742 7.642 0.71 0.032
4 2000/10 7.928 0.987 7.57 0.945 0.043
5 2000/11 7.863 1.223 7.499 1.17 0.054
6 2000/12 7.799 1.465 7.429 1.4 0.065
7 2001/01 7.735 1.705 7.359 1.628 0.077
8 2001/02 7.675 1.92 7.293 1.832 0.088
9 2001/03 7.616 2.156 7.229 2.056 0.1
10 2001/04 7.556 2.383 7.163 2.271 0.112
11 2001/05 7.497 2.615 7.098 2.491 0.124
12 2001/06 7.438 2.838 7.034 2.702 0.136
13 2001/07 7.38 3.067 6.97 2.918 0.149
14 2001/08 7.322 3.294 6.907 3.132 0.162
15 2001/09 7.266 3.512 6.846 3.338 0.174
16 2001/10 7.211 3.735 6.785 3.548 0.187
17 2001/11 7.156 3.95 6.726 3.75 0.2
18 2001/12 7.102 4.17 6.667 3.957 0.214
19 2002/01 7.048 4.389 6.608 4.161 0.227
20 2002/02 6.997 4.585 6.553 4.345 0.24
21 2002/03 6.947 4.8 6.498 4.546 0.254
22 2002/04 6.896 5.007 6.443 4.74 0.267
23 2002/05 6.846 5.219 6.388 4.938 0.282
24 2002/06 6.796 5.423 6.334 5.128 0.295
25 2002/07 6.747 5.632 6.28 5.322 0.31
26 2002/08 6.697 5.84 6.227 5.515 0.324
27 2002/09 6.65 6.039 6.175 5.701 0.339
28 2002/10 6.602 6.244 6.124 5.89 0.354
29 2002/11 6.556 6.441 6.073 6.073 0.368
30 2002/12 6.51 6.642 6.023 6.259 0.383
31 2003/01 6.463 6.843 5.973 6.445 0.398
32 2003/02 6.42 7.023 5.926 6.61 0.412
33 2003/03 6.377 7.22 5.88 6.793 0.428
34 2003/04 6.333 7.41 5.832 6.968 0.443
35 2003/05 6.29 7.605 5.785 7.147 0.458
36 2003/06 6.247 7.793 5.739 7.319 0.473
37 2003/07 6.205 7.985 5.693 7.496 0.489
38 2003/08 6.162 8.176 5.648 7.671 0.505
39 2003/09 6.121 8.36 5.603 7.839 0.521
40 2003/10 6.081 8.548 5.559 8.011 0.537
41 2003/11 6.04 8.729 5.516 8.177 0.553
42 2003/12 6.001 8.915 5.473 8.346 0.569
43 2004/01 5.961 9.1 5.43 8.515 0.585
44 2004/02 5.922 9.272 5.389 8.671 0.601
45 2004/03 5.885 9.454 5.349 8.837 0.618
46 2004/04 5.847 9.63 5.308 8.996 0.634
47 2004/05 5.809 9.81 5.267 9.159 0.651
48 2004/06 5.772 9.983 5.228 9.316 0.667
49 2004/07 5.735 10.161 5.188 9.477 0.684
50 2004/08 5.699 10.337 5.149 9.637 0.701
51 2004/09 5.663 10.507 5.111 9.79 0.717
52 2004/10 5.627 10.682 5.073 9.947 0.735
53 2004/11 5.592 10.85 5.035 10.098 0.751
54 2004/12 5.558 11.022 4.998 10.253 0.769
55 2005/01 5.523 11.193 4.961 10.407 0.786
56 2005/02 5.49 11.347 4.926 10.545 0.802
57 2005/03 5.458 11.516 4.891 10.697 0.819
58 2005/04 5.425 11.679 4.856 10.842 0.837
59 2005/05 5.392 11.846 4.821 10.992 0.854
60 2005/06 5.36 12.007 4.787 11.135 0.871