Shahamat2015 PDF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

REE167686 DOI: 10.

2118/167686-PA Date: 18-November-15 Stage: Page: 508 Total Pages: 15

A Physics-Based Method To Forecast


Production From Tight and Shale
Petroleum Reservoirs by Use of Succession
of Pseudosteady States
M. S. Shahamat*, University of Calgary; L. Mattar, IHS Global Canada Limited; and R. Aguilera, University of Calgary

Summary sized capacitance/resistance network and the associated control


Analysis of production data from tight and shale reservoirs equipment for simulating four of the most prolific reservoirs in
requires the use of complex models for which the inputs are rarely Saudi Arabia. They used a trial-and-error procedure in which the
known. The same objectives can also be achieved by knowing resistance values, and occasionally capacitance values, were
only the overall (bulk) characteristics of the reservoir, with no adjusted until the voltage history of each controller agreed with
need for all the detailed and rarely known inputs. In this study, we the pressure history of the oil wells it represented. The resulting
introduce the concept of continuous succession of pseudosteady model was then used for forecasting the reservoir and individual-
states as a method to perform the analysis of production data. It well performance.
requires few input data yet is based on rigorous engineering con- Recently, capacitance/resistance models (CRMs) have received
cepts, which works during the transient- as well as the boundary- renewed interest for predicting performance of waterflood and gas-
dominated-flow periods. flood projects and determining interwell connectivities in mature
This method consists of a combination of three simple and reservoirs (Albertoni and Lake 2003; Yousef et al. 2006; Sayar-
well-known equations: material balance, distance of investigation, pour et al. 2009, Weber et al. 2009). Briefly, they based their anal-
and boundary-dominated flow. It is a form of a capacitance/resist- ysis on the assumption of constant productivity index—i.e.,
ance methodology in which the material-balance equation over boundary-dominated-flow (BDF) condition—and treated the injec-
the investigated region represents the capacitance and the bound- tion rates as input signals and production rates as output signals,
ary-dominated-flow equation represents the resistance. The flow and then used the signal-processing and nonlinear multivariate-
regime in the region of investigation (the areal extent of which regression techniques to determine the interwell connectivity and
varies with time during transient flow) is assumed to be pseudos- response delay as their unknown system parameters. Their
teady state. This region is depleted at a rate controlled by the ma- approach is useful for characterizing reservoirs with high perme-
terial-balance equation. abilities but has limited application in unconventional reservoirs
The initial flow rate and flowing pressure are used to define the where the permeability is low, productivity index is not constant,
resistance, and the distance of investigation defines the capacitance. and transient flow plays a significant role for a long time.
The capacitance and resistance are then used in a stepwise proce- In this study, we first elaborate on the similarities between
dure to calculate the depletion and the new rates or flowing pres- electrical and petroleum engineering. On the basis of the similar-
sures. The method was tested, for linear-flow geometry, against ities, we combine the classical reservoir variables such as perme-
analytical solutions for liquids and numerical simulations for gas ability, porosity, total compressibility, and fracture half-length
reservoirs, exhibiting both transient and boundary-dominated flow. into resistance and capacitance terms. Through a simple relation-
Excellent agreement was obtained, thus corroborating the validity ship between rate, pressure, and time and by use of a mechanistic
of the method developed in this study. Two practical examples are model, these terms are used to characterize the reservoir and to
provided to demonstrate the applicability of the methodology to forecast the well’s production. Introducing the concept of continu-
forecast production from tight and shale petroleum reservoirs. ous succession of pseudosteady states, it is shown that the deple-
The proposed method is easy to implement in a spreadsheet tion equations can be used not only for analysis of late-time
application. It indicates that complex systems with complicated production data, where BDF prevails, but also for early-time data
mathematical (e.g., Laplace space) solutions can be represented where transient flow is dominant. We demonstrate the usefulness
adequately by use of simple concepts. The approach offers a new of the approach for long-term production forecasting of low-per-
insight into production analysis of tight and shale reservoirs, by use meability tight and shale reservoirs producing under constant ter-
of familiar and easy-to-understand reservoir-engineering principles. minal rate and constant terminal pressure. Two practical examples
are also provided to reveal the applicability of the methodology
for analysis of field-rate decline data and to forecast petroleum
Introduction production from tight and shale reservoirs. This paper shows that
Fluid flow in petroleum engineering is akin to flow of electricity the developed physics-based model that relies on the depletion
in electrical engineering. The correspondences between petroleum equations is easy to use, is similar to the conventional decline
and electrical engineering have long been used in the petroleum curves, is applicable to both liquid and gas reservoirs, and does
industry for reservoir modeling and simulation. Initial attempts in not require pseudotime calculations.
the use of electrical analogs include those of Bruce (1943), who
presented the theory of the use of electrical capacitance and resist-
ance and built the electrical network to model the flow of fluids in Equivalence Between Electrical and
the reservoir. Wahl et al. (1962) constructed an extremely large- Petroleum Engineering
Flow of fluids in a porous medium is in many respects analogous
to flow of current in a conductor. The electric current (I) through
* Now with IHS.
a conductor (such as a wire) is the result of a potential difference
Copyright V
C 2015 Society of Petroleum Engineers
(DE) across its length. In the electrical world, flow of current is
This paper (SPE 167686) was accepted for presentation at the SPE/EAGE European expressed according to Ohm’s law, I ¼ DE=R, where R is the
Unconventional Resources Conference and Exhibition, Vienna, Austria, 25–27 February
2014, and revised for publication. Original manuscript received for review 4 November 2013. electrical resistance and is a function of cross-sectional area of the
Revised manuscript received for review 13 June 2015. Paper peer approved 3 August 2015. conductor, its length, and the intrinsic properties of the conductor

508 November 2015 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 16:03 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 509 Total Pages: 15

Table 1—Summary of analogy between flow of fluid in a porous medium and flow of electricity in an
electrical conductor.

(resistivity, the inverse of which is conductivity). Similarly, in the the wellbore-storage constant defined by Van Everdingen and
petroleum world there is Darcy’s law, which relates the pressure Hurst (1949). The difference lies in the fact that instead of the
difference to the rate of production. It can be expressed in terms wellbore volume and compressibility, here the volume and com-
of the difference between average reservoir pressure and wellbore pressibility of the reservoir are used. Table 1 provides a sum-
pressure to yield the so-called deliverability equation, q ¼ Dp=R, mary of the analogies between electrical and petroleum
where Dp ¼ pavg  pwf . Expressing Darcy’s law in terms of aver- engineering. It is noted that in the case of the latter, both the ca-
age pressure is important because average pressure and its pacitance and resistance are functions of length through V (reser-
changes in production (time) play a critical role in calculation of voir volume) and L, respectively. It is also noted in Table 1 that
rock and fluid characteristics, and therefore in reservoir perform- f1 and f2 demonstrate different functions operating on the argu-
ance forecasting, economic evaluation, and management. By ments inside the parentheses.
analogy, the resistance (R) is a function of the properties of the The correspondence between electrical and petroleum systems
reservoir, cross-sectional area (Ac ) , and length (L) over which mentioned previously can be used to explain production behavior
the flow takes place. The term resistivity does not have a direct of any reservoir, during both transient-flow and BDF periods.
analog in the petroleum field, but the term  conductivity does. The methodology dependent on this analogy is therefore called
k the capacitance/resistance model (CRM). Explaining the reser-
Conductivity is equivalent to mobility . This means that in voir behavior during BDF is straightforward because the reser-
l
the same way that the conductivity implies the ease of flow of voir has been fully investigated and therefore the capacitance and
electricity, the mobility signifies the ease of flow of fluids in resistance terms are constant. Under this condition, production at
the reservoir. the wellbore means depletion of a tank. As mentioned previously,
In addition, in the electrical analysis there is a capacitance a number of studies have been performed to illustrate the utility
term that is the ability of electrical elements to store energy (elec- of the CRM for waterflood-performance forecasting of reservoirs
trical charge). Capacitance is the ratio of the stored energy to the with high permeabilities (Albertoni and Lake 2003; Yousef et al.
voltage across the terminals (DE). A pure capacitance (C) that 2006; Sayarpour et al. 2009; Weber et al. 2009). Explaining the
experiences a constant current (I) will exhibit a constant rate of reservoir behavior during the transient-flow period is more com-
dE plex. This is because the investigated size of the reservoir
voltage change over time; i.e., I ¼ C . Likewise in petroleum changes with time, which results in corresponding changes in
dt
engineering, capacitance can be defined as the ability of reservoir both capacitance and resistance. As we will demonstrate later,
to supply energy (pressure). It is the ratio of cumulative produced provided that we know (or can guess) the flow regime, we can
fluid to the pressure depletion (difference) resulting from this use the concept of continuous succession of pseudosteady states
production. By use of the compressibility equation, the capaci- for applying the tank-type depletion to the transient-flow period
tance can be expressed as the product of the total system com- as well.
pressibility and the reservoir volume from which the production
is occurring. The definition for capacitance is similar in form to Basic Model
The basic reservoir model used in this work is similar to that of
Wattenbarger (Wattenbarger et al. 1998; Bello and Wattenbarger
2010) and Nobakht (Nobakht and Clarkson 2012a, b; Nobakht
et al. 2012). Depicted in Fig. 1, this geometry is chosen because
production-data analysis of low-permeability tight and shale res-
ervoirs exhibit linear flow for long periods of time. Fig. 1 shows a
rectangular reservoir with a hydraulic fracture in the center. The
fracture completely traverses the reservoir, and therefore its pro-
duction exhibits transient linear flow until the investigated dis-
Hydraulic fracture
tance (yinv ) is equal to the reservoir length in the y-direction (ye );
thereafter, boundary-dominated flow is dominant.
yinv The region that is investigated is a description of the propaga-
tion of the pressure disturbance into the reservoir, and is obtained
ye by use of the distance-of-investigation equation (yinv ), which is
Eq. 1. Eq. 1 states that depending on the nature of the rock, the
fluids that flow toward the wellbore, and the duration of flow,
there is a region that has been considerably affected by (and there-
fore has notably contributed to) the production at the wellbore:
xe = xf sffiffiffiffiffiffiffiffiffi
b2 kt
Fig. 1—Schematic of a hydraulically fractured well in the center yinv ¼ a1 : . . . . . . . . . . . . . . . . . . . . . . . . . . . ð1Þ
/lc
of a rectangular reservoir.

November 2015 SPE Reservoir Evaluation & Engineering 509

ID: jaganm Time: 16:03 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 510 Total Pages: 15

In Eq. 1, k is the permeability in md, l is the viscosity in cp, / which is a reservoir property that determines the speed at which
is the porosity, and time t is in days. Moreover, b2 ¼ 0:00633 and the pressure disturbance travels through the reservoir (Eq. 8). This
a1 is a constant that depends on the criterion used for defining the property is used for obtaining the distance of investigation (Eq. 1).
distance of investigation. There have been numerous values of a1 Second, division of the capacity by the resistivity gives a parame-
presented since the concept of radius/distance of investigation ter, the capacity/resistivity ratio (CRR) (Eq. 9), which is character-
was introduced. All depend on some assumptions; for example, istic of linear flow and can be used for performance prediction. It
the observation of certain pressure drops at that distance or of the is noted that analysis of constant-rate linear flow by Clark (1968)
presence of certain type of flow. Moreover, for linear flow, which and Gringarten et al. (1974) also shows the same combination of
is the focus of this study, it also depends on the type of production parameters for determining the pressure change with time.
at the wellbore. In this study and for constant-rate production, we  
use the Wattenbarger et al. (1998) definition, which gives 1 p b2 k
¼ ; . . . . . . . . . . . . . . ð8Þ
a1 ¼ 1:42, and for the case of constant-pressure production, we Capacity  Resistivity 2 /lc
use Nobakht and Clarkson (2012a), which gives a1 ¼ 2:55. These  
values are used because they result in a better match of the numer- Capacity 8p kc/ xf h 2
CRR ¼ ¼ : . . . . . . . . . ð9Þ
ical-simulation results. It should be mentioned that the definition Resistivity 5:615b1 l B
of yinv in Eq. 1 requires time, which has an upper limit of tBDF ,
which is the time to reach boundary-dominated flow. This upper The capacitance/resistance terminology can be used to explain
limit of time for yinv ensures that the calculated reservoir distance the depletion behavior during BDF of both constant-terminal-rate
in the y-direction does not increase forever with progression of and constant-terminal-pressure production. Knowing that C and R
time, and is limited to the size of the reservoir, ye . are constant during BDF, Eq. 2 shows that constant-terminal-rate
production for equal time intervals results in equal pressure deple-
tion. In addition, Eq. 3 suggests that any pressure drop at the well-
Depletion During Transient and bore caused by constant-rate production results in dropping of
Boundary-Dominated Flow (BDF) average reservoir pressure by the same amount. This means that
Change in the average reservoir pressure with time as a result of in constant-rate production, wellbore pressure tracks the average
production at the wellbore is called depletion. To determine the reservoir pressure during BDF. For constant-terminal-pressure
amount of depletion for any production time interval, the material production, constancy of wellbore pressure means that depending
balance (i.e., the compressibility) equation can be written as on the value of R, declining production rate can be used for deter-
mining the depletion (drop in average pressure).
qDt The situation is different when one applies this reasoning for
Dpdep ¼ ; ............................ ð2Þ describing the behavior during transient flow. To explain this flow
C
behavior, we use the complete solution of the reservoir configura-
where C is the reservoir capacitance in STB/psia and Dpdep is the tion shown in Fig. 1, derived in Appendix A. We use the distance-
pressure depletion as a result of production for a time interval Dt. of-investigation equation (Eq. 1) to calculate pavg as the average
In addition to Eq. 2 and as mentioned previously, the deliver- pressure within the investigated volume. Fig. 2a shows the con-
ability equation is a relation between average reservoir pressure, stant-rate pressure profile at three consecutive times during tran-
the wellbore pressure, and the rate of production: sient flow with their corresponding average-pressure values.
Comparison of the calculated wellbore pressure and average
1 pressures for each time demonstrates that pavg and pwf drop by dif-
q¼ ðpavg  pwf Þ; . . . . . . . . . . . . . . . . . . . . . . . . . . ð3Þ ferent amounts. This is contrary to the constant-rate production
R
during BDF, where these pressure values track each other. This is
where R is equivalent to the inverse of the productivity index because of the effect of the changing size of the reservoir (that is,
(psia/STBD). yinv ) during the transient-flow period.
Capacitance (C) and resistance (R) in Eqs. 2 and 3 depend on Constant-pressure production of the same reservoir shows
the flow regime and can be obtained analytically for simple reser- another story. Here, we use the constant-pressure complete solu-
voir geometries. For flow of a liquid within the reservoir geometry tion derived in Appendix A to obtain the pressure profile and, sim-
shown in Fig. 1, C and R can be easily obtained as ilar to the constant-rate production, calculate the volumetric
average pressure within the investigated region. Fig. 2b depicts
4xf ch/ the pressure profile and associated average pressure for three con-
C¼  yinv ; . . . . . . . . . . . . . . . . . . . . . . . . . . ð4Þ
5:615B secutive times. This plot demonstrates that during transient linear
b Bl flow, the average pressure stays at a constant value that is neither
R¼ 1  ða2  yinv Þ; . . . . . . . . . . . . . . . . . . . . . ð5Þ the initial pressure nor the wellbore pressure (Shahamat 2014;
2pkhxf
Tabatabaie 2014; Behmanesh et al. 2015). As a result, during this
where b1 ¼ 2p  141:2 and a2 is a constant determined from the flow period the pressure difference ðpavg  pwf Þ in the deliverabil-
deliverability equation (Wattenbarger et al. 1998) and is equal to ity equation is constant, but the increasing resistance with time as
p 2 a result of increasing yinv causes the rate to decline.
and for constant-rate and constant-pressure production, This discussion demonstrates that to use the depletion equa-
6 p
respectively. Because of their dependency upon the distance of tion during transient flow, we need to consider the distance of
investigation, yinv , the obtained C and R parameters are subject to investigation in the related calculations. As a result, we introduce
the production constraint, which is constant terminal pressure or the concept of continuous succession of pseudosteady states to
constant terminal rate. It is more convenient to define parameters represent the transient-flow period, for constant-rate and con-
that are only representative of the reservoir characteristics and in- stant-pressure production, by use of simple depletion and resist-
dependent of the production scenario. One can therefore introduce ance equations.
the terms capacity and resistivity through the following equations:

Capacity ¼ C=yinv ; . . . . . . . . . . . . . . . . . . . . . . . . . . ð6Þ Continuous Succession of Pseudosteady


States (SPSS)
Resistivity ¼ R=ða2  yinv Þ: . . . . . . . . . . . . . . . . . . . . ð7Þ Continuous succession of steady states was introduced by Muskat
(1937) as a method that enables use of the simple steady-state
Introducing capacity and resistivity through Eqs. 6 and 7 gives solutions for dealing with time-varying systems where the time
two unique advantages for production forecasting. First, multipli- transient plays a minor role in the physical behavior of the sys-
cation of the capacity and resistivity gives hydraulic diffusivity, tem. He discussed that the methodology is useful for systems

510 November 2015 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 16:03 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 511 Total Pages: 15

5,000 5,000

Δpavg t1
t2 t1

Pressure (psia)
Pressure (psia)
t3 t2
t3

Δpwf

Complete solution Complete solution


pavg – pwf
Average pressure Average pressure
4,000 3,000
0 200 400 0 200 400
Distance (ft) Distance (ft)
(a) (b)

Fig. 2—Pressure profile during transient linear flow: (a) constant-rate and (b) constant-pressure production.

where transmission of the pressure disturbances in the porous these reservoirs, and we compare its results with those of the rig-
medium may be considered as effectively instantaneous. It is orous approaches.
required, therefore, for the reservoir to have small dimensions
and high permeability and for the fluids to be either incompressi- Liquid Reservoir: Constant-Rate Production. Constant-rate
ble or slightly compressible. production of a liquid reservoir from a system shown in Fig. 1
We can build on this concept and introduce the concept of gives a declining wellbore pressure for which the behavior during
SPSS using the electrical analogy discussed previously. With this both transient and boundary-dominated flow (BDF) can be
concept, the transient-flow behavior of any reservoir producing described by use of an analytical solution. The complete solution
under constant-rate or constant-pwf can be reasonably represented involves writing the dimensionless diffusivity equation and asso-
by a continuous succession of the pseudosteady-state equations, ciated boundary and initial conditions for a finite reservoir and
provided that then obtaining the solution and expressing it in the real-time do-
• The main flow regime (e.g., linear or radial) can be reason- main. The complete solution is valid for all time and therefore
ably approximated. applies during both transient flow and BDF. The complete solu-
• The changes in capacitance and resistance are obtained by tion for constant-rate production for the reservoir configuration
use of the distance-of-investigation equation. shown in Fig. 1 is derived in Appendix A and is given as
• Production from one timestep to the other causes depletion
of the associated capacitance, according to the material-bal- 8 2  2 39
  X1 < np =
ance equation. p tD 1 4  tD
pD ¼ þ pyeD 1  e yeD 5 ;
• The resistance and therefore the pressure and/or rate at the 2 yeD : 2 ;
n¼1 ðnpÞ
new timestep are obtained by use of the deliverability equa-
tion, and the reservoir pressure is determined by depletion                    ð10Þ
of the capacitance at the previous timestep.
The procedure consists of the stepwise coupling of the mate-
rial-balance equation for the investigated volume with the deliver- where pD , tD , and yeD are dimensionless pressure, time, and dis-
ability equation. It should be mentioned that irrespective of the tance, respectively, and are defined as shown in Appendix A.
production scenario, the depletion calculations between two con- The complete solution (Eq. 10) gives the correct behavior for
secutive timesteps (tj and tjþ1 ) are performed at a constant rate of constant-rate production of a liquid reservoir with dominant linear
production, hence the name pseudosteady states. flow. As can be seen, this solution involves the summation of a
In the sections that follow, we first generate synthetic rate/ large number of terms (at least 100) to produce acceptable results.
pressure data for liquid and gas reservoirs with dominant linear By use of this equation and the parameters in Table 2, we gener-
flow, by use of rigorous approaches (either analytical or numerical ated a set of wellbore-pressure data for a specified constant rate of
methods). We then demonstrate the implementation and applic- 10 STB/D.
ability of the SPSS methodology to forecast the performance of For using the SPSS, we require only four parameters: first well-
bore-pressure, constant rate of production, tBDF , and capacity/resis-
tivity ratio (CRR). CRR is a parameter that combines a number of
reservoir and fluid properties. It is noted here that irrespective of
the value of resistivity and/or capacity, as long as a reasonable
CRR value is used in the analysis, production forecasting gives
reasonable results. This statement is strictly applicable for cases
where skin is assumed negligible and can be obtained from
the linear-flow theory. Assuming skin is zero, the simple SPSS
procedure is shown in Fig. 3. Knowing the mentioned input
parameters and assuming a value for the resistivity, we use
the CRR to calculate the capacity. Hydraulic diffusivity is
obtained as inverse of multiplication of capacity and resistivity:
1
. Having obtained the hydraulic diffusivity,
Capacity  Resistivity
we can calculate the investigated distance according to Eq. 1 with
the upper limit being obtained by using tBDF . Then we use the
Table 2—Reservoir geometry and properties used for generating the resulting distance of investigation along with the capacity to obtain
complete solution. capacitance and thereby the pressure depletion (Eq. 2). Thereafter,

November 2015 SPE Reservoir Evaluation & Engineering 511

ID: jaganm Time: 16:03 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 512 Total Pages: 15

Inputs
8 h i2 9
1 < ð2n  1Þp =
4 X  tD
q CRR tBDF pwf1 qD ¼ e 2yeD ; . . . . . . . . . . . . . . ð11Þ
pyeD n¼1 : ;
Assume a resistivity
Capacity where qD is the dimensionless rate that is defined in Appendix A.
The complete solution (Eq. 11) gives the correct behavior for con-
stant-pressure production of a liquid reservoir with dominant lin-
C = Capacity × yinv yinv
ear flow. Similar to the constant-rate production, this solution
involves the summation of a large number of terms (at least 100)
j=1 to produce acceptable results during transient flow. During BDF,
at each timestep the summation can be approximated by use of only one term to
qΔt j
yield the familiar exponential decline equation (Fetkovich 1980).
j j j j
Δpdep = pwf +1 = pwf – Δpdep j = j+1 By use of Eq. 11 and the parameters in Table 2, we generated a
Cj
set of declining-rate data for a specified constant wellbore pres-
sure of 500 psia. We then use the first rate and wellbore pressure
to examine the SPSS methodology.
Fig. 3—The procedure for applying SPSS for constant-rate pro-
duction of liquid reservoirs. Again, there are two reservoir parameters of importance for
performing the production analysis: CRR and the time to reach
BDF (tBDF ¼ 360 days). Following a similar procedure, we
the pressure depletion and the first wellbore-pressure are used in a assume a resistivity and then use CRR ¼ 0:158 ðStb=psiaÞ2 =Day
stepwise procedure to calculate the subsequent wellbore pressures. to calculate the capacity. Afterward, resistivity and capacity are
By using the first wellbore pressure from the complete solu- used to calculate the hydraulic diffusivity to be used for determin-
tion, CRR ¼ 0:158 ðStb=psiaÞ2 =D and tBDF ¼ 574 days, and fol- ing the distance of investigation. In addition, the time to reach
lowing the procedure outlined in Fig. 3, we obtained an excellent BDF (tBDF ) is used to set an upper bound for the distance of inves-
match of the complete solution, as shown in Fig. 4. Whereas Fig. tigation. The calculated distance of investigation and capacity and
4a shows the Cartesian plot of wellbore pressure vs. time, Fig. 4b resistivity are used for obtaining the pressure depletion (through
demonstrates the logarithmic plot of the pressure difference the material-balance equation) and also the average pressure
(pi  pwf ) vs. time. The straight lines of slope of one-half and (through the deliverability equation). Thereafter, by knowing the
unity shown in Fig. 4b clearly depict the dominant flow regimes first production rate and following a stepwise procedure, we cal-
to be linear flow and BDF. The obtained match allows us to con- culate the subsequent rates. Fig. 5 demonstrates the procedure.
clude that for constant-rate production, knowing the production Comparison of the results between SPSS and those of the com-
rate and the first measured wellbore pressure and using only two plete solution is shown in Fig. 6. This figure shows Cartesian and
reservoir parameters (CRR and tBDF ), the stepwise procedure ena- log/log comparison of the resulting rates. As is seen, the match is
bles calculation of the wellbore pressure during transient flow and not satisfactory. The reason for this is the assumption that the rate
BDF. Therefore, CRR and tBDF can be used as parameters to be from one timestep (tj ) to the next (tjþ1 ) is constant. Therefore, a
estimated for obtaining an idea about the reservoir characteristics better match can be obtained by use of either smaller timesteps
and predicting its future behavior. It should be mentioned here and/or iteration on the rate for depletion calculation.
that using the analytical solutions (Eqs. 1 and 9) and parameters Fig. 7 shows the match when three iterations on calculated
in Table 2, similar values for CRR and tBDF can be obtained, thus rate were used. The obtained excellent match demonstrates the va-
confirming the validity of the method for the case of constant-rate lidity of the approach for the case of constant-pressure production.
production. It is worth mentioning that use of smaller timesteps produces
results that are similar to those obtained by use of iteration.
The excellent match of the complete-solution-rate data shown
Liquid Reservoir: Constant-Pressure Production. Constant- in this study reveals that by knowing two parameters about the res-
pressure production of a liquid reservoir with a configuration ervoir (CRR and tBDF ), one can follow the stepwise procedure out-
shown in Fig. 1 gives a declining-rate behavior that can be lined in Fig. 5 to calculate the subsequent rates. It is noted here that
described by use of an analytical complete solution. The complete the CRR parameter used for yielding the match for the constant-
solution for constant-pressure production is derived in Appendix pressure case was equal to the value used for obtaining the match
A and is given as for the constant-rate production, the reason being that the same

5,000 1.E+4

Complete solution
4,000 SPSS by use of CRM
Half slope-indicating
linear flow
1.E+3
pi –pwf

3,000
pwf

Unit slope
2,000 indicating BDF
1.E+2

1,000 Complete solution


SPSS by use of CRM
0 1.E+1
0 1,000 2,000 3,000 4,000 5,000 10 100 1,000 10,000

Time (days) Time (days)

(a) (b)

Fig. 4—Comparison of the SPSS and the complete solution for constant-rate production: (a) Cartesian plot of pwf vs. time and (b)
log/log plot of pipwf vs. time.

512 November 2015 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 16:03 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 513 Total Pages: 15

Inputs sure and pseudotime in place of time) or numerical simulation. In


q1 CRR tBDF this section, we use a commercial numerical simulator with prop-
erties shown in Table 3 and wellbore rate of 350 Mscf/D to obtain
Assume a resistivity gas-production data.
Capacity
The discussion about applying depletion and deliverability
equations is valid for liquid reservoirs with constant fluid proper-
yinv ties (viscosity and compressibility). The approach can be used for
the case of gas, provided that we can account for the changes in
C = Capacity × yinv R = Resistivity × α2yinv gas compressibility and viscosity. As is shown in Appendix B, the
j=1 gas-compressibility equation can be written according to Eq. 12,
at each timestep where there is no restriction that the gas compressibility should be
q jΔt j constant. Eq. 12 is the well-known gas-material-balance equation:
q j+1 = q j – j = j+1
C jR j p qg Dt
D ¼ : . . . . . . . . . . . . . . . . . . . . . . . . . ð12Þ
Z dep C
Fig. 5—The procedure for applying SPSS for constant-pressure In addition, for gas reservoirs, we can write the deliverability
production of a liquid resrvoir.
equation in terms of pseudopressure to yield

reservoir and fluid parameters were used for generation of the com- ½Pp;avg  Pp;wf 
qg ¼ ; .................... ð13Þ
plete-solution results. The tBDF parameter, though, was smaller for R
the case of constant-pressure production compared with constant-
rate case. The reason is the difference in the speed of propagation where Pp is the normalized pseudopressure in psia. Similar to the
of the pressure disturbance between these two scenarios. liquid-flow cases discussed previously, for gas reservoirs Eqs. 6
and 7 are used to relate capacitance with capacity and resistance
with resistivity. Again, here the capacity and resistivity terms are
Gas Reservoir: Constant-Rate Production. Gas-production data independent of the production scenario and their multiplication
can be obtained by use of either the complete solutions obtained should give the hydraulic diffusivity. We use the CRR as the pa-
in the previous subsections (with pseudopressure in place of pres- rameter for performance forecasting.

200 1.E+3
Complete solution
SPSS by use of CRM
150 1.E+2
qwb

qwb

100 1.E+1

50 1.E+0
Complete solution
SPSS by use of CRM
0 1.E-1
0 1,000 2,000 3,000 4,000 5,000 10 100 1,000 10,000

Time (days) Time (days)


(a) (b)

Fig. 6—Comparison of the SPSS and the complete solution for constant-pwf production: (a) Cartesian and (b) log/log plot of q vs.
time.

200 1.E+3
Complete solution Half-slope indicating
linear flow
SPSS by use of CRM
150 1.E+2
qwb

qwb

100 1.E+1
Exponential decline
indicating BDF

50 1.E+0
Complete solution
SPSS by use of CRM
0 1.E–1
0 1,000 2,000 3,000 4,000 5,000 10 100 1,000 10,000

Time (days) Time (days)


(a) (b)

Fig. 7—Comparison of the SPSS (with three iterations on rates) and the complete solution for constant-pwf production: (a) Carte-
sian and (b) log/log plot of q vs. time.

November 2015 SPE Reservoir Evaluation & Engineering 513

ID: jaganm Time: 16:03 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 514 Total Pages: 15

equation. By using this and the depletion value for this timestep,
we can calculate the new p/Z, which can be converted back to
pseudopressure through another table lookup. The new pseudo-
pressure is then used for the next round of calculations. Applying
this procedure, we obtained a perfect match of wellbore pressure,
shown in Figs. 9a and 9b using CRR ¼ 79:5 ðMScf=psiaÞ2 =D
and tBDF ¼ 422 days.

Gas Reservoir: Constant-Pressure Production. In this subsec-


tion, the reservoir described in the previous subsection with prop-
erties in Table 3 is produced under constant wellbore pressure of
500 psia. The declining synthetic rates (from numerical simula-
tion) are used as the basis to compare the applicability of the pro-
posed approach for making gas-rate forecasting. The procedure
that is shown in Fig. 10 requires the first rate and first pseudopres-
sure and estimation of two reservoir parameters: CRR and tBDF .
Table 3—Reservoir geometry and fluid properties used for numerical
By using CRR and assuming a value for resistivity, we calcu-
simulation of a gas reservoir.
late capacity. Capacity and resistivity define the hydraulic diffu-
sivity, which is used for determining yinv , and tBDF sets an upper
limit for yinv . Combining capacity and resistivity with obtained
It should be noted that Eq. 12 gives the depletion in terms of p/ distance of investigation through Eqs. 6 and 7 gives capacitance
Z, whereas Eq. 13 requires pseudopressure for its calculation. To and resistance that would allow the calculation of depletion and
convert pressure to pseudopressure and to p/Z, one can obtain a ta- deliverability equations by applying Eqs. 12 and 13. Through the
ble of gas properties and perform linear interpolation between its table lookup, one can calculate the p/Z corresponding to the aver-
entries to calculate m(p) and p/Z for any value of pressure. The age pseudopressure from the deliverability equation. By use of
procedure for performing the SPSS analysis for constant-rate pro- this and the depletion value for this timestep, we can calculate the
duction from a gas reservoir is shown in Fig. 8. The blue “T” in new p/Z, which can be converted back to pseudopressure through
Fig. 8 is the abbreviation for a table lookup (using the table of another table lookup. The new pseudopressure is then used in the
pressure, z-factor, viscosity, and pseudopressure for interpolation). deliverability equation to calculate a new production rate and
Akin to the liquid case, the analysis procedure requires the first therefore a next round of calculations. Applying this procedure
rate and first pseudopressure (which can be easily obtained by a with small timesteps, we obtained a close match of the declining
simple table lookup) and the estimation of two reservoir parame- simulated rates by using CRR ¼ 79.5(MScf/psia)2 and tBDF ¼ 375
ters: CRR and tBDF . By use of CRR and assuming a value for resis- days. This is shown in Figs. 11a and 11b.
tivity, we can obtain capacity. These two parameters define the The results obtained in this study for liquid and gas cases dem-
hydraulic diffusivity, which is used for determining yinv , and tBDF onstrate that depletion equations can be used for describing the
sets an upper limit for yinv . Combining capacity and resistivity transient flow and BDF, provided that the distance-of-investiga-
with the obtained distance of investigation through Eqs. 6 and 7 tion concept is used in the analysis. It has been illustrated that the
gives capacitance and resistance that would allow the calculation approach is applicable for both oil and gas reservoirs producing
of depletion and deliverability equations by applying Eqs. 12 and under constant rate or constant wellbore pressure. Results for the
13. Through the table lookup, one can calculate the p/Z corre- gas reservoir confirm that application of the simple gas-material-
sponding to the average pseudopressure from the deliverability balance equation through a stepwise procedure eliminates the

Inputs
1
q CRR tBDF ppwf

Assume a
Capacity
resistivity

yinv

C = Capacity × yinv R = Resistivity × α 2 yinv

at each timestep j=1


p
j qΔt j
Δ = ppavg j = qR j+ppwf j
z dep Cj
T
j+1 j j j
p p p p
= – Δ
z avg z avg z dep z avg
T
ppavg j+1 Δ ppdepj = ppavg j – ppavg j+1

T
ppwf j +1 = ppwf j – Δ ppdepj ppwf j+1

j = j+1

Fig. 8—The procedure for applying SPSS for constant-rate production of a gas reservoir.

514 November 2015 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 16:03 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 515 Total Pages: 15

5,000 10,000
Simulation
SPSS by use of CRM
4,000

3,000

pi –pwf
pwf
1,000
2,000

1,000
Simulation
SPSS by use of CRM
0 100
0 1,000 2,000 3,000 4,000 5,000 10 100 1,000 10,000

Time (days) Time (days)

(a) (b)

Fig. 9—Comparison of the SPSS and the numerical simulation for constant-rate production of a gas reservoir: (a) Cartesian plot of
pwf vs. time and (b) log/log plot of pipwf vs. time.

Inputs
1
q CRR tBDF ppwf

Assume a
Capacity
resistivity

yinv

C = Capacity × yinv R = Resistivity × α 2 yinv

at each timestep j=1


j jΔt j
p q j j
Δ = ppavg = q jR j+ppwf
z dep Cj
T
j +1 j j
p p p p j
= – Δ
z avg z avg z dep z avg

T
j +1
j+1 ppavg – ppwf
ppavg q j +1 =
Rj

j = j+1

Fig. 10—The procedure for applying SPSS for constant-pressure production of a gas reservoir.

4,000 10,000
Simulation
SPSS by use of CRM
3,000
1,000

2,000
q

100
1,000
Simulation
SPSS by use of CRM
0 10
0 1,000 2,000 3,000 4,000 5,000 50 500 5,000
Time (days) Time (days)
(a) (b)

Fig. 11—Comparison of the results of the SPSS (with small timesteps) and the numerical simulation for constant-pressure produc-
tion of a gas reservoir: (a) Cartesian and (b) log/log plots of q vs. time.

November 2015 SPE Reservoir Evaluation & Engineering 515

ID: jaganm Time: 16:03 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 516 Total Pages: 15

Without Skin Without Skin


With Skin With Skin

PD

1/qD
Sqrt(tD) Sqrt(tD)

(a) (b)

Fig. 12—Effect of skin on the specialized-linear-flow plot for (a) constant-rate production and (b) constant-pressure production.

need for using iterative pseudotime calculations for production Because the CRM approach involves the use of an initial pres-
forecasting. This means that the effects of changing gas viscosity sure value, it has the skin effect embedded in it. Therefore, the
and compressibility have been properly embedded in gas-mate- skin effect for constant-rate production is implicitly included in
rial-balance and -deliverability equations. the CRM analysis, and the constant-rate approaches mentioned in
the previous sections are also applicable to situations where there
Effect of Skin is skin.
For constant-pressure production, however, the situation is dif-
The analysis methods mentioned in the previous sections are ap-
ferent. As demonstrated by Bello and Wattenbarger (2010), the
plicable to cases where there is no skin (s). In this section, a brief
additional pressure drop because of skin causes a diminishing
description of skin is given, followed by an explanation of skin
effect on the production rate. Unlike the constant-rate case, skin is
effects on capacitance/resistance-methodology (CRM) analysis.
not simply additive in constant-pressure production (Nobakht and
There are different factors in, and in the vicinity of, the well-
Mattar 2012). This is because the additional pressure drop (Dps )
bore that can alter the pressure measured at the well. In tight and
depends on the wellbore rate, which declines with time and in
shale wells, the additional pressure drop can be caused by multi-
turn has a decreasing effect. Bello and Wattenbarger
pffiffiffiffiffi (2010) used
phase flow, liquid loading, incomplete recovery of the injected
the linear-flow-specialized plot (1=qD vs. tD ) to illustrate this
fracturing fluid, damage on the fracture face, near-fracture satura-
behavior (Fig. 12b).
tion changes or emulsions, pressure-dependent fluid and/or rock
In this situation, the CRM approaches presented in the previ-
properties, finite fracture conductivity, or adsorption/diffusion of
ous sections are not applicable because the skin effect should be
gas near the wellbore region (where the pressure is lower). Com-
included in the resistance (R). Skin causes additional resistance to
binations of these and other effects often complicate analysis of
flow, and therefore its effects can be considered in the CRM by
production data. There are different ways to account for these
modifying the resistance term through the use of an additional
effects in well-testing and production-data analysis. Van Everdin-
term (s ):
gen (1953) defined skin (s) as a dimensionless additional pressure
drop at the wellbore. As a result, he was able to consider damage/
R ¼ R   ða2  yinv Þ þ s ; . . . . . . . . . . . . . . . . . . ð14Þ
stimulation effects mathematically in well-testing (constant-rate)
solutions. This means that the classical way to include skin effect Dps
where s is the resistance caused by skin and is given by s ¼
is through addition of s to the dimensionless pressure solution at q
the wellbore. Thus, all the pressure values from the start to the (in which Dps is additional pressure drop caused by skin) and

end of production exhibit the same additional pressure drop where R is the resistivity.
caused by skin (Dps) (Fig. 12a).
Thus, a simple addition of s in the resistance term allows for
inclusion of skin effect in the CRM analysis. This means that by
calculating resistance with Eq. 14, constant-pressure approaches
40 developed in the previous sections can be used for cases where
Complete solution with skin skin is significant, and its effects can be considered in the analysis.
CRM by use of SPSS Fig. 13 shows the application of CRM for a reservoir with skin
30 effect exhibiting linear flow. The complete solution was obtained
for a rectangular reservoir of ye , xe , and xf equal to 500 ft; perme-
ability of 0.005 md; and s ¼ 2. By use of the CRM approach and
Eq. 14, the initial rate of 29.6 STB/D, CRR ¼ 0:15 ðStb=
qwb

20 psiaÞ2 =D, and s ¼ 5:8 psia=STB=D, excellent results are ob-


tained. Fig. 13 shows that the CRM can be used for analysis of
cases with skin effect.
10 Notwithstanding the behavior of idealized reservoirs produc-
ing at constant pwf that were discussed previously, Nobakht and
Mattar (2012) argue that the practical observation from numerous
shale-gas wells essentially producing at constant flowing pressure
0 indicate that they yield a “straight-line-with-intercept” rather than
0 200 400 600 800 1,000 the “curve-with-intercept” shown in Fig. 12b. This behavior is
Time (days) contrary to the theoretical work by Bello and Wattenbarger
(2010) and is used to present a simple method of removing the
Fig. 13—Comparison of CRM and the complete solution with (apparent) skin from the data. The modified production data then
skin effect for constant-pressure production. can be analyzed by use of linear-flow-specialized plots. Along the

516 November 2015 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 16:03 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 517 Total Pages: 15

10,000 3,000 10,000


Gas rate Gas rate
BHP Synthetic

Gas Rates (Mscf/D)


Gas Rates (Mscf/D)
1,000 2,000 1,000

100 1,000 100

10 0 10
0 500 1,000 1,500 2,000 2,500 3,000 0 500 1,000 1,500 2,000 2,500 3,000
Time (days) Time (days)
(a) (b)

Fig. 14—(a) Cartesian plot of historical BHP and rate for the dry-gas well in the western Canadian sedimentary basin (Yu et al.
2013); (b) comparison of the generated synthetic rates with field rates (this study).

same lines, it can be stated that when analyzing the constant-pwf create the synthetic rate-data set has an original gas in place of
production data with apparent skin, one can remove the skin 1,805 MMScf.
effects from all data by the method that Nobakht and Mattar The synthetic rate data were then analyzed by use of the capac-
(2012) proposed. Having accomplished that step, the modified itance/resistance methodology (CRM) proposed in this paper.
rate data are analyzable by using the CRM approach without the Examining the pressure plot, it was assumed that the bottomhole
need for use of an additional parameter, such as resistance caused pressure is constant and equal to 150 psia. Gas gravity of 0.621
by skin (s ), in the analysis. was used to generate the table of pressure/pseudopressure p/Z.
The required parameters for obtaining a reasonable match of the
Applications to Field and Synthetic Data declining-rate data were determined to be q1 ¼ 10; 690 MSc=D,
capacity/resistivity ratio (CRR) ¼ 6; 800 ðMScf=psiaÞ2 =D, and
In this section, two field examples are presented. In the first
tBDF ¼ 475 days. Although the former parameter was determined
example, the methodology is applied for analysis of synthetic by use of the first-available production rate, the two latter parame-
variable-rate/-pressure data. The results of this case validate the ters were determined through a history-matching process. This
ability of the proposed methodology to reasonably forecast field- means that the CRR and tBDF values were altered in such a way as
production rates. This is then followed by another field example to improve the match quality of the available production history.
with more than 40 years of high-quality production data. The
The result of examining these parameters in CRM is shown in
rate/time data are analyzed, future rates are forecasted, and con- Fig. 15. It can be seen that the CRM reasonably history matches
sistent results are obtained. and forecasts the rate decline.
By using an abandonment rate of 15 MScf/D, the expected ulti-
Dry-Gas Multifractured Well in Western Canadian Sedimentary mate recovery (EUR) on the basis of the CRM was determined to
Basin (Yu et al. 2013). This example well is a Cadomin dry-gas be 1,783 MMScf. This is in close agreement with the original gas
openhole horizontal well that is hydraulically fractured in multi- in place used for construction of the synthetic rates (1,805
ple stages, with pertinent details given by Yu et al. (2013). The MMScf). Evaluation of the data by use of Arps method with
historical pressure and rate data for this well are shown in Fig. b ¼ 2.0 during transient flow and then b ¼ 0.5 for the rest of the
14a. With the obtained bottomhole-pressure (BHP) data and by production period yielded EUR equal to 1,919 MMScf. Use of
use of a commercial software (Fekete Harmony 2013) for model- b ¼ 1.0 during the boundary-dominated flow (BDF) yielded 2,408
ing a horizontal well with multistage fractures, a synthetic set of MMScf for the EUR. These results demonstrate the overestimation
rate data was created. A comparison plot of the synthetic rate data of EUR by use of the Arps method. Fig. 15 shows a comparison of
with the actual field data is shown in Fig. 14b. The model used to these methods. It is noted that as time progresses, the separation of
the calculated forecasts by use of the CRM and the Arps method
10,000
increases. This occurs although both methods obtained reasonable
Gas rates match of the available production history.
CRM
Arps method, b = 0.5
Gas Rate (Mscf/D)

Mexico Tight Gas Well (Amini et al. 2007). This example is a


1,000 tight gas reservoir with a permeability of less than 0.001 md. The
hydraulically fractured vertical well of interest in this reservoir
has more than 40 years of high-quality production data, and as a
result is considered a “near-textbook” example which can be used
100 for production-data analysis (Fig. 16). Amini et al. (2007) devel-
oped elliptical models to analyze the declining production rate of
this case.
Similar to the previous field case, the same steps were taken to
10 analyze the rate/time data from this well. Following the CRM
0 2,000 4,000 6,000 8,000 10,000 analysis procedure, the parameters were determined to be CRR ¼
Time (days) 730 ðMScf=psiaÞ2 =Day and tBDF ¼ 4; 100 days. A gas gravity of
0.65 was used for generating the results. The skin, and therefore
Fig. 15—Comparison of rate prediction for dry-gas well in the resistance caused by skin, were considered to be negligible
western Canadian sedimentary basin by use of two different (S  ¼ 0). As can be seen in Fig. 16, the CRM closely matched
methods. the rate-decline history. Fig. 16 depicts the obtained results in

November 2015 SPE Reservoir Evaluation & Engineering 517

ID: jaganm Time: 16:04 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 518 Total Pages: 15

50,000 appearance of BDF in the production data, which can demonstrate


Gas rates bigger reservoir size for the same reservoir permeability. The pa-
Gas Rate (Mscf/D) CRM rameter CRR reflects the combined effect of different rock and
Arps Method, b = 0.5 fluid properties. In fact, it is a bulk reservoir parameter that can be
5,000 used for describing the overall behavior of the entire system. The
higher the CRR, the larger the relative effect of reservoir capacity
to resistivity, and therefore the possibility for better well perform-
ance. In short, the CRM approach involves meaningfully combin-
ing as much of the reservoir unknowns as possible to obtain a
500 match of the available production data. As a result, there are fewer
parameters in the CRM approach than the other analytical meth-
ods and the degrees of freedom are smaller; therefore, the non-
uniqueness issue is believed to be mitigated.
50 The analysis cases considered in this study only include con-
0 10,000 20,000 30,000 40,000 stant-rate and constant-pressure production. These operating con-
Time (days) ditions are difficult to maintain in the field. As a result, it is
essential to apply the introduced approach for analysis of cases
Fig. 16—Analysis of Mexico tight-gas-rate data by use of the where both the production rate and the pressure vary with time.
CRM. This is the subject of a future study, because further research is
required to extend the applicability of the approach for such cases.
semilogarithmic coordinates. The time for start of BDF (tBDF ) and
the obtained rate decline are consistent with the results presented Conclusions
in the literature for this case (Amini et al. 2007; Ilk 2010). This study presents a new methodology for predicting future
By using an abandonment rate of 15 MScf/D, the EUR derived rate or pressure of oil and gas reservoirs with linear-flow geome-
from the CRM was determined to be 19,220 MMScf. Evaluation try. The methodology consists of a simple coupling of equations
of the data by using the Arps method with b ¼ 2.0 during transient for material balance, distance of investigation, and boundary-
flow and then b ¼ 0.5 for the rest of the production period yielded dominated flow (BDF) by using the concept of continuous suc-
EUR equal to 19,533 MMScf. Fig. 16 shows a comparison of cession of pseudosteady states. Tested against analytical solu-
these two methods. It demonstrates that for this case, CRM results tions for liquids and numerical simulations for gas reservoirs
are very close to the Arps-method results, and their entire rate during both transient flow and BDF, the approach showed excel-
forecasts closely overly each other for the time frame shown in lent agreement, which confirmed the ability of the method for
Fig. 16. reliably forecasting production. Key conclusions from this work
are the following:
Discussion 1. The approach requires very little input data yet is derived from
Forecasting production and estimating reserves have been among rigorous engineering concepts, and works during the transient-
the most-challenging tasks for petroleum engineers. It becomes a flow and BDF periods.
more-daunting task in the case of tight and shale reservoirs with 2. The proposed method is easy to implement in a spreadsheet
extended transient flow. Lack of understanding of the basic reser- application. When applied to gas reservoirs, it does not require
voir properties, such as the permeability and the areal/vertical calculating pseudotime.
extent, often forces the engineers to resort to easy-to-use methods 3. Effect of skin on the capacitance/resistance-methodology
(such as the Arps decline equations), which are mostly empirical results can be accounted for by use of an additional term in the
in nature. Considerable research has been conducted to provide resistance value: a resistance caused by skin. The higher the
physical justification for these methods, including the seminal value of skin, the higher is the resistance caused by skin.
work of Fetkovich (1980) on Arps decline equations. The problem 4. The depletion equations can be used not only for analysis of
with these methods, however, is that their involved parameters late-time production data, where BDF prevails, but also for
lack physical meaning from a reservoir-engineering standpoint, early-time data where transient flow is dominant. This indi-
and therefore they suffer from arbitrariness. cates that complex systems with complicated mathematical
The capacitance/resistance methodology (CRM) presented in (e.g., Laplace space) solutions can be represented adequately
this work is based on simple coupling of material-balance, dis- by use of simple concepts.
tance-of-investigation, and boundary-dominated-flow (BDF)
equations. Similar to the Arps method, it requires minimal data, is
very easy to use, and can be used to quickly obtain a reasonable Nomenclature
match of the history data. Contrary to the Arps method, its Ac ¼ cross-sectional area, ft2
involved parameters are the result of fusing different reservoir B ¼ liquid formation volume factor, bbl/STB
factors affecting production at the wellbore. These parameters Bgi ¼ gas formation volume factor, ft3/scf
include an initial data point for performing the calculation, cg ¼ gas compressibility, 1/psia
capacity/resistivity ratio (CRR), and tBDF . Although the depend- cti ¼ initial total compressibility (for gas cases), 1/psia
ency on the value of the initial point (q, pwf) opens the method to c ¼ total compressibility, 1/psia
noise or user bias, it should be kept in mind that this method pro- C ¼ electrical capacitance, farad
vides just a good approximation. In some cases, a very-good ini- C ¼ fluid-flow capacitance, STB/psia
tial guess for the initial point can be obtained by using the first- CRR ¼ capacity/resistivity ratio, ðSTB=psiaÞ2 =D for oil and
available production data. There are other cases where the initial ðMScf=psiaÞ2 =D for gas
available production data are affected by noise and operating con- G ¼ initial gas in place, Mscf
ditions, and therefore the initial production data are not represen- H ¼ net-pay, ft
tative of the reservoir response. As a result, a value different from I ¼ electric current, A
the first data point must be used for analysis. Because of data K ¼ permeability, md
uncertainty and quality issues, one can make use of an approach L ¼ length, ft
similar to that of Duong (2011) to determine the first reliable rate/ mðpÞ ¼ normalized pseudopressure, psia
pressure and the corresponding time (i.e., using the best 1-month pavg ¼ average reservoir pressure within the investigated region,
or 3-month average rate instead of the first measured rate). The psia
parameter tBDF is self-explanatory; larger tBDF indicates later pD ¼ dimensionless pressure

518 November 2015 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 16:04 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 519 Total Pages: 15

pD ¼ dimensionless pressure in Laplace domain Fetkovich, M. J. 1980. Decline Curve Analysis Using Type Curves. J Pet
pi ¼ initial reservoir pressure, psi Technol 32 (6): 1065–1077. SPE-4629-PA. http://dx.doi.org/10.2118/
pwf ¼ wellbore flowing pressure, psia 4629-PA.
q ¼ well-production rate, STB/D Gringarten, A. C., Ramey, H. J. and Raghavan, R. 1974. Unsteady-State
qD ¼ dimensionless rate Pressure Distributions Created by a Well With a Single Infinite-Con-
qg ¼ gas-well-production rate, Mscf/D ductivity Vertical Fracture. SPE J. 14 (4): 347–360. SPE-4051-PA.
R ¼ electrical resistance, X http://dx.doi.org/10.2118/4051-PA.
R ¼ flow resistance, psia/(STB/D) Ilk, D. 2010. Well Performance Analysis for Low to Ultra-Low Permeabil-
t ¼ time, days ity Reservoir Systems. PhD dissertation, Texas A&M University, Col-
tBDF ¼ time to reach BDF, days lege Station, Texas.
tD ¼ dimensionless time Muskat, M. 1937. The Flow of Homogeneous Fluids Through Porous
T ¼ reservoir temperature, 8F Media. New York City: McGraw-Hill.
u ¼ Laplace-space variable Nobakht, M. and Clarkson, C. R. 2012a. A New Analytical Method for
V ¼ reservoir volume, ft3 Analyzing Linear Flow in Tight/Shale Gas Reservoirs: Constant-Flow-
xe ¼ reservoir length in x-direction, ft ing-Pressure Boundary Condition. SPE Res Eval & Eng 15 (3):
xf ¼ fracture half-length, ft 370–384. SPE-143989-PA. http://dx.doi.org/10.2118/143989-PA.
yD ¼ dimensionless distance in y-direction Nobakht, M. and Clarkson, C. R. 2012b. A New Analytical Method for
ye ¼ reservoir length in y-direction, ft Analyzing Linear Flow in Tight/Shale Gas Reservoirs: Constant-Rate
yeD ¼ dimensionless reservoir length in y-direction Boundary Condition. SPE Res Eval & Eng 15 (1): 51–59. SPE-
yinv ¼ distance of investigation, ft 143990-PA. http://dx.doi.org/10.2118/143990-PA.
Z ¼ gas-compressibility factor, fraction Nobakht, M. and Mattar, L. 2012. Analyzing Production Data From
a1 ¼ constant used for defining the distance of investigation; Unconventional Gas Reservoirs With Linear Flow and Apparent Skin.
a1 ¼ 1:42 for constant-rate and 2.55 for constant pres- J Can Pet Technol 51 (1): 52–59. SPE-137454-PA. http://dx.doi.org/
sure production 10.2118/137454-PA.
a2 ¼ constant used for determining liquid productivity index; Nobakht, M., Mattar, L., Moghadam, S., et al. 2012. Simplified Forecasting
p 2 of Tight/Shale-Gas Production in Linear Flow. J Can Pet Technol 51
a2 ¼ for constant-rate and for constant-pressure (6): 476–486. SPE-133615-PA. http://dx.doi.org/10.2118/133615-PA.
6 p
production Sayarpour, M., Zuluaga, E., Kabir, C. S., et al. 2009. The Use of
b1 ¼ constant used for determining dimensionless pressure Capacitance–Resistance Models for Rapid Estimation of Waterflood
and rate; b1 ¼ 2p  141:2 in field units Performance and Optimization. J. Pet. Sci. Eng. 69 (3–4): 227–238.
b2 ¼ constant used for determining dimensionless time; b2 ¼ http://dx.doi.org/10.1016/j.petrol.2009.09.006.
0:00633 in field units Shahamat, M. S. 2014. Production Data Analysis of Tight and Shale Res-
cg ¼ gas gravity, fraction ervoirs. PhD dissertation, University of Calgary, Calgary.
DE ¼ potential difference, V Tabatabaie, S. H. 2014. Unconventional Reservoirs: Mathematical Model-
Dp ¼ pressure difference, psia ing of Some Non-Linear Problems. PhD dissertation, University of
Dpdep ¼ pressure depletion, psia Calgary, Calgary.
Dt ¼ time interval, days Van Everdingen, A. F. 1953. The Skin Effect and Its Influence on the Pro-
l ¼ viscosity, cp ductive Capacity of a Well. J Pet Technol 5 (6): 171–176. SPE-203-G.
/ ¼ porosity, fraction http://dx.doi.org/10.2118/203-G.
Van Everdingen, A. F. and Hurst, W. 1949. The Application of the Lap-
lace Transformation to Flow Problems in Reservoirs. J Pet Technol 1
References (12): 305–324. SPE-949305-G. http://dx.doi.org/10.2118/949305-G.
Albertoni, A. and Lake, L. W. 2003. Inferring Interwell Connectivity Only Wahl, W. L., Mullins, L. D., Barham, R. H., et al. 1962. Matching the Perform-
From Well-Rate Fluctuations in Waterfloods. SPE Res Eval & Eng 6 ance of Saudi Arabian Oil Fields With an Electrical Model. J Pet Technol
(1): 6–16. SPE- 83381-PA. http://dx.doi.org/10.2118/83381-PA. 14 (11): 1275–1282. SPE-414-PA. http://dx.doi.org/10.2118/414-PA.
Amini, S., Ilk, D. and Blasingame, T. A. 2007. Evaluation of the Elliptical Wattenbarger, R. A., El-Banbi, A. H., Villegas, M. E., et al. 1998. Produc-
Flow Period for Hydraulically-Fractured Wells in Tight Gas Sands– tion Analysis of Linear Flow Into Fractured Tight Gas Wells. Pre-
Theoretical Aspects and Practical Considerations. Presented at the sented at the SPE Rocky Mountain Regional/Low-Permeability
SPE Hydraulic Fracturing Technology Conference, College Station, Reservoirs Symposium, Denver, 5–8 April. SPE-39931-MS. http://
Texas, 29–31 January. SPE-106308-MS. http://dx.doi.org/10.2118/ dx.doi.org/10.2118/39931-MS.
106308-MS. Weber, D., Edgar, T. F., Lake, L. W., et al. 2009. Improvements in Capaci-
Arps, J. J. 1945. Analysis of Decline Curves. Petroleum Development and tance-Resistive Modeling and Optimization of Large Scale Reservoirs.
Technology 160 (1): 228–247. SPE-945228-G. New York: Transac- Presented at the SPE Western Regional Meeting, San Jose, California,
tions of the Americal Institute of Mining and Metallurgical Engineers, 24–26 March. SPE-121299-MS. http://dx.doi.org/10.2118/121299-MS.
AIME. http://dx.doi.org/10.2118/945228-G. Yousef, A. A., Gentil, P. H., Jensen, J. L., et al. 2006. A Capacitance
Behmanesh, H., Hamdi, H. and Clarkson, C. 2015. Analysis of Transient Model to Infer Interwell Connectivity From Production- and Injection-
Linear Flow Associated with Hydraulically-Fractured Tight Oil Wells Rate Fluctuations. SPE Res Eval & Eng 9 (6): 630–646. SPE-95322-
Exhibiting Multi-Phase Flow. Presented at the SPE Middle East Uncon- PA. http://dx.doi.org/10.2118/95322-PA.
ventional Resources Conference and Exhibition, Muscat, Oman, 26–28 Yu, S., Lee, W. J., Miocevic, D. J., et al. 2013. Estimating Proved
January. SPE-172928-MS. http://dx.doi.org/10.2118/172928-MS. Reserves in Tight/Shale Wells Using the Modified SEPD Method. Pre-
Bello, R. O. and Wattenbarger, R. A. 2010. Modelling and Analysis of sented at the SPE Annual Technical Conference and Exhibition, New
Shale Gas Production With a Skin Effect. J Can Pet Technol 49 (12): Orleans, 30 September–2 October. SPE-166198-MS. http://dx.doi.org/
37–48. SPE-143229-PA. http://dx.doi.org/10.2118/143229-PA. 10.2118/166198-MS.
Bruce, W. A. 1943. An Electrical Device for Analyzing Oil-Reservoir
Behavior. Trans. AIME 151 (1): 112–124. SPE-943112-G.
Clark, K. K. 1968. Transient Pressure Testing of Fractured Water Injection Appendix A—Complete Solution for Linear Flow
Wells. J Pet Technol 20 (6): 639–643. SPE-1821-PA. http://
The general linear flow solution can be determined by solving the
dx.doi.org/10.2118/1821-PA.
following diffusivity equation:
Duong, A. N. 2011. Rate-Decline Analysis for Fracture-Dominated Shale
Reservoirs. SPE Res Eval & Eng 14 (3): 377–387. SPE-137748-PA.
@ 2 p /lc @p
http://dx.doi.org/10.2118/137748-PA. ¼ : . . . . . . . . . . . . . . . . . . . . . . . . . . ðA-1Þ
Fekete Harmony. 2013. Calgary: Fekete Associates Inc. @y2 b2 k @t

November 2015 SPE Reservoir Evaluation & Engineering 519

ID: jaganm Time: 16:04 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 520 Total Pages: 15

In Eq. A-2, b1 ¼ 2p  141:2. From dimensionless definitions


for either of the production scenarios, the diffusivity equation to
be solved will take on the form
@ 2 pD @pD
¼ : . . . . . . . . . . . . . . . . . . . . . . . . . . . ðA-3Þ
@y2D @tD

Note that the initial condition and the boundary conditions are
defined in Table A-2.
We solve the diffusivity equation with associated initial and
boundary conditions in Laplace space. Writing the conditions in
Table A-2 in Laplace domain, we get Table A-3.
The solution of the diffusivity equation gives
pffiffi pffiffi
pD ¼ AeyD u þ BeyD u : . . . . . . . . . . . . . . . . . . . . ðA-4Þ
Table A-1—Dimensionless parameters for linear flow.
Satisfying the initial and boundary conditions in Table A-3
and Eq. A-4, we obtain the coefficients A and B for each of the
production scenarios, which are listed in Table A-4.
We define the dimensionless parameters for constant-rate- and Therefore, the final complete solutions in the Laplace domain
constant-pressure-production scenarios in Table A-1, where g is are given in Table A-5.
b k These solutions can be inverted into the real-time domain by
the hydraulic diffusivity and equal to g ¼ 2 . Note that from use of numerical-inversion algorithms. Alternatively, we can use
/lct
Darcy’s law, we have an entirely different approach to solve the diffusivity equation in
real time. By use of the separation-of-variables method to obtain
khxf @p complete solutions for a reservoir with linear-flow geometry, one
qðy; tÞ ¼ : . . . . . . . . . . . . . . . . . . . . . . . . ðA-2Þ can obtain the following equation for constant-rate production:
b1 Bl @y

Table A-2—Linear flow, initial and boundary conditions.

Table A-3—Linear flow, initial and boundary conditions in Laplace domain.

Table A-4—Linear flow, initial and boundary conditions in Laplace domain.

520 November 2015 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 16:04 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 521 Total Pages: 15

Table A-5—Linear-flow solutions in Laplace domain.

8    9 where p means average pressure (pavg ). We note in Eq. B-1 that V


> yD 2  2 3>
  X1 <>cos np np >
= and dV are in ft3. Replacing the former by GBgi and the latter by
p tD yeD 4  tD
5 :
pD ¼ þ pyeD 1  e yeD qBg dt and then substituting them into the Eq. B-1 compressibility
2 yeD >
n¼1 > ðnpÞ2 >
> equation, we get
: ;

                   ðA-5Þ 1 qBg dt
cg ¼  : . . . . . . . . . . . . . . . . . . . . . . . . ðB-2Þ
GBgi dp
For constant-pressure production, Eq. A-6 is obtained:
8 9
X1 <   hð2n  1Þpi2 =  From the definition
 of gas formation volume factor (FVF)
4 ð2n  1Þp  tD ZT
pD ¼ 1  sin yD e 2yeD : Bg ¼ 0:0283 , we can write the following relation between
n¼1
:ð2n  1Þp 2yeD ; p
Bg and Bgi :
                   ðA-6Þ
Bg ðZ=pÞ
Eqs. A-5 and A-6 are the pressure profiles for constant-rate ¼ : . . . . . . . . . . . . . . . . . . . . . . . . . . . ðB-3Þ
Bgi ðZ=pÞi
and constant-pressure production and are real-time equivalents of
the complete solutions in Table A-5. Substitution of yD ¼ 0 in Eq.
A-5 yields the wellbore pressure for the constant-rate scenario: Its substitution in Eq. B-2 gives
8 2  2 39
  X1 < np = ðZ=pÞ 1 qdt
p tD 1 4  tD
5 : cg ¼  : . . . . . . . . . . . . . . . . . . . . . . ðB-4Þ
pD ¼ þ pyeD 1  e yeD ðZ=pÞi G dp
2 yeD :
n¼1 ðnpÞ
2 ;

                   ðA-7Þ There is an alternative definition for compressibility, on the


basis of FVF, through Eq. B-5:
We can also obtain the rate profiles within the reservoir for ei-
ther of the production scenarios by differentiating the pressure 1 dBg
profiles in Eqs. A-5 and A-6. The rate profile for constant-rate cg ¼  : . . . . . . . . . . . . . . . . . . . . . . . . . . ðB-5Þ
Bg dp
production is obtained according to Eq. A-8:
8    9 Upon substituting the equation for gas FVF into Eq. B-5 and
> yD 2  2 3>
X 1 >
<sin np np >
= taking the derivative with respect to pressure, we get
yeD 4  tD
5
qD ¼ 2 1  e yeD . . . . ðA-8Þ
1 1 dZ
>
n¼1 >
np >
>
: ; cg ¼  : . . . . . . . . . . . . . . . . . . . . . . . . . . ðB-6Þ
p Z dp
For the case of constant-pressure production, we obtain Equating the right-hand sides of Eqs B-4 and B-6 and rear-
8 9 ranging the obtained equation yield:
1 <   hð2n  1Þpi2 =
4 X ð2n  1Þp  tD    p  1 qdt
qD ¼ cos yD e 2yeD : 1 p dZ
pyeD n¼1 : 2yeD ;  2 ¼ : . . . . . . . . . . . . . . . . . ðB-7Þ
Z Z dp Z i G dp
                   ðA-9Þ dðp=ZÞ
The left-hand side of Eq. B-7 is equal to . Upon its sub-
stitution, we obtain dp
Substitution of yD ¼ 0 in Eq. A-9 yields the wellbore rate for
the constant-pressure scenario:  p  1 qdt
8 h dðp=ZÞ
i2 9 ¼ : . . . . . . . . . . . . . . . . . . . . . ðB-8Þ
1 < ð2n  1Þp =
4 X  tD dp Z i G dp
qD ¼ e 2yeD : . . . . . . . . . . . . ðA-10Þ
pyeD n¼1 : ; Multiplying through by dp, we can simply obtain the following
relation:

Appendix B—The Depletion Equation for Gas p  p  q dt


g
d ¼ : ..................... ðB-9Þ
The depletion equation for a liquid is according to the compressi- Z Z i G
bility equation. Its calculation is easy because the liquid compres-
sibility can be assumed constant. For gas flow, however, the Integrating Eq. B-9 over the entire production period, we get
compressibility changes substantially with production, and there-
fore depletion calculation according to simple liquid-flow equa- p p  Gp

tions is not possible. Here, we show that the depletion for a gas ¼ 1 : . . . . . . . . . . . . . . . . . . . . . ðB-10Þ
Z Z i G
reservoir can be calculated easily according to the familiar p/Z
(material-balance) equation. As is evident, we obtained Eq. B-10 from the gas-compressibil-
We start with the definition of the compressibility: ity equation. This means that for flow of gases, one can use the p/Z
equation in place of the compressibility equation to calculate deple-
1 dV
cg ¼  ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . ðB-1Þ tion. By use of p/Z, therefore, we do not have to be concerned
V dp about the changes in the gas compressibility with production.

November 2015 SPE Reservoir Evaluation & Engineering 521

ID: jaganm Time: 16:04 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035


REE167686 DOI: 10.2118/167686-PA Date: 18-November-15 Stage: Page: 522 Total Pages: 15

Mohammad Sadeq Shahamat is a reservoir engineer and prin- data analysis. He is a Distinguished Member of SPE and was an
cipal technical analyst with IHS Global Canada Limited. Within SPE Distinguished Lecturer in well testing in 2003. In 2006, Mat-
IHS, he is primarily involved in analysis and interpretation of tar received the SPE International Reservoir Description and
unconventional wells spanning major shale plays in Canada Dynamics Award. He holds a master’s degree in petroleum
and the US. Shahamat’s primary areas of interest include engineering and has authored more than 70 technical
mathematical modeling, production forecasting, reserves esti- publications.
mation, and well-spacing/completions optimization of con-
Roberto Aguilera is professor and CNOOC Nexen Chair in Tight
ventional and unconventional oil and gas reservoirs. He has
Oil and Unconventional Gas at the University of Calgary. He
authored several technical papers in various journals and con-
ferences. Shahamat holds a bachelor’s degree in petroleum has authored or coauthored more than 200 papers. Aguilera
engineering from the Petroleum University of Technology is the recipient of the 2011 SPE Canada Regional Distinguished
Achievement Award for Petroleum Engineering Faculty and is
(PUT), Iran; dual master’s degrees in reservoir engineering from
past executive editor of the Journal of Canadian Petroleum
the University of Calgary/PUT; and a PhD degree in petroleum
engineering from the University of Calgary. Technology. He holds a petroleum-engineering degree from
the Universidad de America in Bogota, Colombia, and mas-
Louis Mattar is a principal reservoir adviser with IHS Global Can- ter’s and PhD degrees in petroleum engineering from the Col-
ada Limited. He specializes in well-testing and production- orado School of Mines.

522 November 2015 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 16:04 I Path: S:/REE#/Vol00000/150035/Comp/APPFile/SA-REE#150035

You might also like