Vous êtes sur la page 1sur 18

SPE-181691-MS

A Transient Coupled Wellbore-Reservoir Model Using a Dynamic IPR


Function

A. Posenato Garcia, The University of Texas; P. Cavalcanti de Sousa, Texas A&M University; P. J. Waltrich,
Louisiana State University

Copyright 2016, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Dubai, UAE, 26-28 September 2016.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
Conventional IPR models are used in coupled wellbore-reservoir transient simulations, even if bottomhole
pressure conditions are assumed to be constant on the derivation of such IPR models. The dynamic IPR
model proposed in this paper not only captures the relevant reservoir dynamics from the well perspective,
but is also computationally more efficient than discretized models using hundreds of gridblocks to simulate
the near wellbore region.
Any well-behaved function can be decomposed in a combination of sinusoidal functions. Hence, a
sinusoidal bottomhole flowing pressure with a frequency varying from a given value down to zero is
assumed as a boundary condition to the hydraulic diffusivity equation. The zero frequency corresponds
exactly to the conventional IPR model since it is the steady-state condition. The analytical solution that
calculates the flow rate for the sinusoidal pressures for each one of the frequencies is presented. Finally,
the Dynamic IPR is defined as the low-order, linear ordinary differential equation that best approximates
this solution.
It is a well known fact that the reservoir dynamics is relevant to the calculation of the flow rate in
certain conditions, such as drawdown and pressure buildup well tests. A comparison between the Dynamic
IPR function, the conventional IPR models, and the analytical solution is presented for these cases. The
conventional IPR model gives unrealistic results, since it allows the flow rate to change instantaneously
to the next equilibrium condition, as opposed to the Dynamic IPR which can capture the fully transient
response. The Dynamic IPR showed excellent agreement with the true solution. Two different examples are
presented in this paper showing the application of the Dynamic IPR: a reservoir well testing and a severe
slugging well. With a given measured bottomhole flowing pressure, a comparison between the Dynamic
IPR, conventional IPR and a reservoir simulation is presented. Once again, the conventional IPR model
provides unsatisfactory results when compared to the Dynamic IPR. These results showed that neglecting
reservoir dynamics in transient simulations might result in inadequate production forecasts, inefficient
designs and eventually unsafe operations while using conventional IPR models.
This work points out some limitations in one of the common practices in the industry, which is the use of
steady state IPR for coupled wellbore-reservoir transient simulations. The alternative Dynamic IPR herein
2 SPE-181691-MS

proposed is computationally inexpensive when compared to discretized reservoir models while still being
able to capture the transient well behavior and the steady-state solution.

Introduction
Traditionally, well deliverability is obtained by combining Inflow Performance Relationship (IPR) and
Vertical Flow Performance (VFP) curves (Economides et al., 2011). This method was first discussed in 1954
(Beggs, 1991), and it provides snapshots of the average bottomhole flowing pressure (Pwf) and average oil
flow rate (qo) relationships at given times in the life of the well. It fails, however, in accurately portraying
the transient behavior of these variables. Transient relationships are important in the design and analysis
of pressure-transient test, design of production tubing and artificial systems, reservoir management, and
estimating flow rates from multiple producing zones (Izgec et al., 2006). The development of reservoir
simulators and wellbore flow models intended to overcome this gap, but research in both fields progressed
separately until it was realized that the interaction between Pwf and qo was of great importance in some
applications (Bahonar et al., 2011). For example, when modeling production instabilities (an event which
may kill a well), wellbore dynamics strongly influence reservoir dynamics, and vice-versa (Nennie et al.,
2007). Other cases where the dynamic interactions between bottomhole flowing pressure and oil flow rate
are important, such as severe slugging, coning of gas and water, and casing heading, can be found in Sagen
et al. (2007).
Capturing the real behavior of these variables requires coupling the reservoir and the wellbore models.
General-use coupled models rely on running reservoir simulations (accounting for near wellbore effects),
and using the output of that run as the input to the wellbore model, which in turn will retrofit the reservoir
simulator for a new run. This can be costly and time consuming, and it is not always successful, or accurate.
On the other hand, coupled models developed for special cases, while still relying upon simulations, are
faster and provide reliable results, but have no general application (Hu et al., 2007; Bahonar et al., 2011).
Fig. 1, based on the work of Pagan et al. (2016), shows an overview of the nodal analysis method, and
highlights where coupled reservoir-wellbore models are effective.

Figure 1—Conventional IPR and reservoir-wellbore simulation cannot simulate all


transient and steady-state events happening throughout the lifetime of the well. The new
Dynamic IPR is proposed to overcome this problem - modified from Pagan et al. (2016).
SPE-181691-MS 3

This paper introduces a new technique of coupling wellbore and reservoir models where the simulation
of the reservoir response to changes in the bottomhole flowing pressure are obtained not by numerical
methods, but rather by solving the diffusivity equation using the Fourier transform. This mathematical tool
generates time-dependent equations - the Dynamic IPR -, which is able to provide the reservoir response to
any pressure variation, no matter how fast or slow this change might be.

Methods
Conventional IPR models are algebraic equations correlating the bottomhole flowing pressure and the flow
rate through the production zone completion. Economides et al. (2011) provide several IPR models for
different well geometries and types of hydrocarbons being produced. The general formula of IPR models
of undersaturated reservoirs is given by Eq. 1:

(1)

where PI is the productivity index, and ΔP is the pressure drawdown, ΔP = Pe - Pwf, with Pe being the
reservoir pressure, and Pwf, the bottomhole flowing pressure.
Since most of these models rely on correlations and experimental data fit, it is common practice in
the industry to perform well tests periodically to correct the IPR models according to the most updated
information on the fluids and reservoir parameters.
The focus of this work are homogeneous, isotropic, circular-shaped, undersaturated oil reservoirs with
a finite diameter vertical well in its center. We are, thus, interested in the IPRs that correspond to these
same conditions. These IPRs are defined for three different flowing times: transient, pseudosteady-state,
and steady-state, given respectively by Eqs. 2 through 4 below:

(2)

(3)

(4)

where Ei is the integral exponential function, k is the permeability, h is the net pay, μ is the viscosity of the
oil, Bo is oil formation volume factor, t is time, ϕ is the reservoir porosity, ct is the total compressibility, qo is
the oil flow rate, re is the effective radius of the reservoir, rw is the wellbore radius, Pi is the initial reservoir
pressure, is the average reservoir pressure, and Pe is the pressure at the effective radius.
The main assumption in these models is that the bottomhole flowing pressure is constant, therefore, it
is common practice in the oil industry to utilize these equations in classical nodal analysis for example.
However, it is equally common to use these IPRs to generate the reservoir response in time-dependent
wellbore models, where the bottomhole flowing pressure will vary significantly with time. In this scenario,
most IPR models are no longer valid and generate unrealistic reservoir responses to changes in wellbore
conditions. Presently, coupling reservoir and wellbore simulations is the only alternative to account for
4 SPE-181691-MS

transient effects in both the reservoir and wellbore, but this can be computationally costly, and have limited
application in the lifetime of a well. This paper introduces a new alternative: the Dynamic IPR.

Dynamic IPR
IPR models give flow rate for a fixed bottomhole pressure, or vice-versa, and are not able to capture reservoir
dynamics, i.e., the interactions between the Pwf and qo. The transient interaction between two interconnected
subsystems take place in several other fields of engineering. Based on the research of Vepa (1976) in
aerospace engineering, we postulate that it is possible to represent the dynamic relationship between Pwf
and qo by taking into account their time derivatives as in Eq. 5:

(5)

where qo(n) and ΔP(n) are the n-th order time-derivative of the oil flowrate and the pressure drawdown
respectively, and ak and bk are coefficients that depend on the reservoir and fluid properties.
For simplicity, we estimate that Eq. 5 can be approximated by truncating both summations after the third
term. From this, we define our Dynamic IPR as a finite order approximation such as Eq. 6:

(6)

where qo′ and qo″ are the first- and second-order time derivate of the oil flowrate, and ΔP′ and ΔP″ are the
first- and second-order time derivative of the pressure drawdown.
Contrary to the conventional IPR models that rely on algebraic models, the Dynamic IPR defined in Eq.
6 is a linear ordinary differential equation designed to capture all aspects of the reservoir behavior, be it
steady-state, or time-dependent in large or small time-scales. In fact, it is easy to prove that, in steady- state,
the Dynamic IPR (Eq. 6) can be written in the same format shown in Eq. 1. In this case, since the time
derivatives are equal to zero, we have Eq. 7:

(7)

where the coefficients ao and bo are directly linked to reservoir and fluid properties through the productivity
index. Consequently, these coefficients are not only meaningless mathematical constants, but they must
have a significant physical meaning.
Additionally, one can generalize Eq. 6 to gas, or two-phase reservoirs by replacing the drawdown with a
generic function of the reservoir pressure and the bottomhole flowing pressure as in Eq. 8 below:

(8)

In addition, as opposed to conventional reservoir simulation systems, the Dynamic IPR is not a partial
differential equation (no explicit spatial dependence); hence, it is less computationally expensive than the
discretization of the hydraulic diffusivity equation in both time and space. Nevertheless, the information
required to determine the parameters of the Dynamic IPR still lie within the hydraulic diffusivity.
SPE-181691-MS 5

Validation and Application of the Dynamic IPR model


In this section, we solve the hydraulic diffusivity equation for a specific transient problem, and show how
to calculate the coefficients of Eq. 6 from this analytical solution. In addition, we show how the Dynamic
IPR model compares to conventional IPR equations, and discuss the applicability of the Dynamic IPR.

Solving the Hydraulic Diffusivity Equation


Let us assume again that we have a homogeneous, isotropic, circular-shaped, undersaturated oil reservoir
with a finite diameter vertical well in its center. The hydraulic diffusivity equation in radial coordinates is
written as Eq. 9:

(9)

The solution of Eq. 9 requires one initial condition and two boundary conditions. If the reservoir was
initially closed, or not flowing for a very long time, the initial condition is given by Eq. 10:

(10)

which means that the entire reservoir is at the same pressure. The first boundary condition is given at the
effective reservoir radius, as shown in Eq. 11:

(11)

which represents a reservoir subject to a water influx mechanism or a region of the reservoir surrounded
by injector wells.
The second boundary condition is given by Eq. 12 below:

(12)

where the bottomhole flowing pressure can be any integrable function of time. It is theoretically possible to
obtain a solution for any given time-dependent bottomhole flowing pressure. The following section shows
the analytical solution for a simple, sinusoidal bottomhole flowing pressure.
Analytical solution. We define the pressure drawdown as having a sinusoidal form with amplitude AΔP,
and angular frequency ω, given by Eq. 13:

(13)

Garcia et al. (2014) have shown that the hydraulic diffusivity equation has an analytical solution in the
time domain for this specific case in the form of Eq. 14:

(14)

where F and G are calculated in terms of Kelvin functions (Barron and Barron, 2012).
6 SPE-181691-MS

A comparison between the analytical solution and the steady-state IPR (Eq. 4) is presented in Fig. 2. The
parameters used to generate these results were taken from Hu (2004) and are presented in Table 1.

Figure 2—Comparison between analytical solution, and steady-state IPR.

Table 1—Parameters of hydraulic diffusivity simulation (Hu, 2004).

Parameter Description Value Unit

h Reservoir height 10 m
Reservoir static pressure
Pe 10×107 Pa
and initial pressure

Pwf Bottomhole flowing pressure 9.0×106 - 0.50×106 sin(ωt) Pa

rw Radius of the well 0.10 m


re Outer radius of the reservoir 25 m

Ct Total compressibility 1.0×10 -8


Pa-1
ϕ Porosity 0.2

μ Oil viscosity 0.001 Pa.s

f = ω/2π Pwf oscillation frequency 1×10-5 to 1×10-2 Hz

k Permeability 0.02 Darcy

The flowrate provided by the steady-state IPR significantly differs from the analytical solution, showing
the inadequacy of a conventional IPR for transient simulations.
Bottomhole flowing pressures are never sinusoidal, however. Therefore, the Dynamic IPR was developed
to be a tool that is both as simple as an analytical solution, and as flexible as numerical simulations. The
following sections will provide the mathematical basis of the Dynamic IPR.

Calculation of Dynamic IPR Coefficients


A Fourier series decomposes any periodic function into a sum of sinusoidal functions (Haberman, 2004).
Thus, if we assume that the drawdown (ΔP) is periodic, one can decompose ΔP into a sum of different sine
functions similar to that introduced by Eq. 13. Furthermore, because the hydraulic diffusivity equation is
linear, one can apply the principle of superposition, that is, any linear combination of the solutions is also a
SPE-181691-MS 7

solution to the problem. In other words, the solution to the hydraulic diffusivity equation with an unknown
periodic ΔP is given by the sum of solutions of the form seen in Eq. 14.
The generalization of the Fourier series is the Fourier transform. This mathematical tool is not limited
to periodic functions, but can be applied to any integrable function, decomposing it into an infinite sum
of complex, sinusoidal functions at any given frequency. For this reason, the Fourier transform is a
representation of a time-dependent function in the frequency domain.
Using the Fourier transform, Garcia et al. (2014) derived the Productivity Index (PI) for a sinusoidal
drawdown (ΔP) of frequency ω (Eq. 13) in the frequency-domain. The PI is introduced by Eq. 1 and is
defined as the ratio between the flow rate and the bottomhole flowing pressure. In the frequency-domain,
the PI is a complex number and can be expressed by its modulus (i.e., its magnitude), and its phase. The
phase is a frequency-domain representation of the delay between the changes in flow rate due to a variation
in the bottomhole pressure. The modulus (Eq. 15) and phase (Eq. 16) are, respectively (Garcia et al., 2014):

(15)

and

(16)

where qo(ω) is the Fourier transform of qo(t), ΔP(ω) is the Fourier transform of the drawdown, ΔP(t).
One of Fourier transforms properties states that the Fourier transform (FT) of the derivative of a given
function h = h(t) is equal to the FT of the original function times the product iro, where i is the imaginary
number. Mathematically, we can write: FT[dh/dt] = iω FT[h]. Similarly, FT[d2h/dt2] = (iω)2 FT[h]. If we
use this property and apply the Fourier transform on both sides of Eq. 5, we obtain Eq. 17:

(17)

where the coefficients ak and bk are the same as those introduced by Eq. 6.
Using the approximation given by Eq. 6, we have Eq. 18:

(18)

This means that the complex-valued PI can be approximated as the ratio of two complex polynomials.
Since we know the modulus (Eq. 15) and phase (Eq. 16) of the complex-valued PI, a curve fitting in the
frequency-domain allows us to obtain the coefficients ak and bk in Eq. 6. From that, the Dynamic IPR (Eq.
6) is fully determined. Fig. 3 illustrates this curve fitting in the frequency domain, where the magnitude
(modulus) and the delay (phase) of the complex PI are plotted against a range of frequencies. This range
represents the frequencies of bottomhole pressure and flow rate oscillations in a coupled reservoir- wellbore
8 SPE-181691-MS

system. The approximated solution to the complex-valued PI(ω) is the Dynamic IPR in the frequency
domain (Eq. 18).

Figure 3—Comparison of the frequency response of the reservoir dynamic provided by the
analytical solution against the Dynamic IPR with respect to the (a) magnitude, and (b) phase.

Finally, if the coefficients of the Dynamic IPR (Eq. 6) are found in such a way that the bottomhole flowing
pressure-flow rate relationship matches the one obtained from the analytical solution, then the Dynamic
IPR will capture the exact reservoir dynamics for any arbitrary input at any time. As an alternative, it is
theoretically possible to obtain PI(t) by computing the inverse Fourier transform of PI(ω).
Example of Dynamic IPR Applicability. So far, we presented the analytical solution for the hydraulic
diffusivity equation with a sinusoidal boundary condition at the wellbore, we showed how this solution is
translated into the frequency domain (Fig. 3), and we introduced a way to obtain the coefficients of the
Dynamic IPR presented in Eq. 6. In this section, we propose a way to determine when the reservoir dynamics
is relevant in transient simulation of wells, that is, when conventional IPRs are not reliable.
First, we assume that the bottomhole flowing pressure is sinusoidal with a frequency of 10-5 Hz, which
corresponds approximately to one cycle per day. This frequency is shown in Fig. 4a by the vertical bold line.
At this frequency, the magnitude of PI (Eq. 15) has the same value as the magnitude of PI at frequency zero
(i.e.; steady-state condition). This means that the oscillation amplitude estimated through the conventional
IPR will be correct. On the other hand, the phase at the frequency of 10-5 Hz is close to 2°. This means that the
flowrate calculated with a conventional IPR will be 2° out-of-phase with respect to the analytical solution.
The flowrate estimated through the conventional IPR is compared against the analytical solution in Fig. 4b:
the amplitudes are the same, and the 2° phase difference is hardly noticeable. This means that, for practical
purposes, the result obtained from the conventional IPR is virtually the same as the analytical solution.
SPE-181691-MS 9

Figure 4—A comparison of the reservoir dynamics predicted by the analytic solution
and the linear IPR, in (a) the frequency domain, (b) in the time domain for a frequency of
-5 -2
oscillation of 10 Hz, and (b) in the time domain for a frequency of oscillation of 10 Hz.
10 SPE-181691-MS

On the other hand, the results for a frequency of 10-2 Hz (approximately one oscillation for every couple
of minutes), are also presented in Fig. 4a. Fig. 4c shows that the Magnitude of the PI function is around
2.3 times larger than the steady-state value. The IPR solution oscillates between 145 and 55 m3/d, with an
amplitude of 90 m3/d. The analytical solution oscillates between 210 and -10 m3/d, with an amplitude of
220 m3/d. Fig. 4a shows that the dephasing between the flowrate calculated via the conventional IPR and
the actual flowrate should be around 17°. This is still a low value, but the delay can be clearly observed
in Fig. 4c.
The values obtained in this section are functions of the reservoir and fluid parameters, as it can be
observed in Eqs. 15 and 16. However, the Dynamic IPR approach can be applied to other reservoirs.

Case Studies
In this section, we analyze a few classic cases to illustrate the versatility of the Dynamic IPR. The idea behind
this procedure is to prove that once the Dynamic IPR for a particular reservoir is obtained, it can be used
to determine flow rate and bottomhole flowing pressure relationships in any scenario as long as fluid and
reservoir properties remain the same. As mentioned before, the cases presented here are for homogeneous,
isotropic, circular-shaped, undersaturated oil reservoirs, with a finite diameter vertical well in its center.
Cases 1 and 2 correspond to a simple Drawdown and a Drawdown-and-Build-Up respectively. These two
cases involve non-periodic functions, and have simple analytical solutions - at least before the pressure drop
reaches the external boundaries of the reservoir. These two cases show that, even though periodic functions
were used to determine the coefficients of the Dynamic IPR, it can by equally applied to conditions in which
non-periodic dynamics are observed.
Case 3 consists of a casing-heading scenario. Here, the superposition of the sinusoidal solutions at
different frequencies yields the analytical solution; however, the input bottom-hole flowing pressure first
has to be decomposed into a sum of sinusoidal functions. Even though this is perfectly feasible, we compare
the results with a conventional finite-differences simulation of the reservoir.
Finally, case 4 couples the Dynamic IPR with a simple, transient wellbore model to evaluate the impact
of the reservoir dynamics into measurable quantities such as production flowrate, wellbore and wellhead
pressures. With this case study we intend to show the Dynamic IPR is able to correctly characterize the
wellbore-reservoir dynamics with minimal computational effort, and also that the steady-state IPR presents
imprecise results.

Case 1: Drawdown
This case corresponds to a constant, non-zero flowrate through the perforations starting at the initial time.
Before this, the entire reservoir was at a pressure Pe. Eq. 2 is the analytical solution for this case if one wants
to know the flow rates as a function of a fixed ΔP. Here, we are interested in knowing the pressure as a
function of a fixed flow rate. Dake (1983) derived this as Eq. 19:

(19)

Eq. 19 is only valid throughout the duration of the transient regime, which means that it is valid while the
pressure drop is still propagating through the reservoir towards the external boundary. Since the pressure at
the external boundary is assumed to be Pe, the long-time solution converges to the steady-state IPR (Eq. 4).
Dake (1983) shows that the time required for the pressure drop to reach the external boundary is (Eq. 20):
SPE-181691-MS 11

(20)

Fig. 5 shows the transient IPR (Eq. 19) and the steady-state IPR (Eq. 4), which, at early times, responds
to a non-zero flowrate as if there was no transient behavior. The vertical line in Fig. 5 corresponds to time
t = te. After this time, the transient IPR is not valid anymore. This can be observed in Fig. 5: the analytical
solution falls below the steady-state IPR. Hence, a single analytical solution cannot simultaneously capture
the transient and the steady-state behavior.

Figure 5—Comparison of the bottomhole flowing pressure calculated with


steady-state, transient, and Dynamic IPRs for a pressure drawdown test.

The Dynamic IPR is also plotted in Fig. 5. The early-time response matches well with the transient IPR,
but the curves diverge shortly at times 0.25te < t < te. This is due to the low order nature of the Dynamic
IPR here introduced as Eq. 6. If one chooses to work with higher order derivatives for both ΔP and qo, this
difference should decrease; however, by doing this, the method becomes computationally more expensive.
Despite that, it continues to be less expensive than solving the hydraulic diffusivity equation as long as
the highest order derivative in the Dynamic IPR is smaller than the number of nodes used to discretize the
reservoir. For t > te, the Dynamic IPR converges to the steady-state IPR solution. Thus, the Dynamic IPR
represents the reservoir behavior at all times with one single equation.

Case 2: Drawdown and Buildup


The Drawdown followed by a Buildup corresponds to a constant nonzero flowrate through the perforations
starting at the initial time and then a zero flow rate after time t0. At time t = t0, we assume a well shut in at
the perforations, thus no well storage effects are taken into account.
Before time to, the pressure behaves like in the previous case. However, after time to a new term has to
be included in Eq. 19. Dake (1983) derived this as Eq. 21 below:

(21)

In Fig. 6 we plotted the responses generated by the steady-state IPR (Eq. 4), the transient analytical IPR
(Eq. 21), and the Dynamic IPR. Just like in the previous case, the use of the steady-state IPR results in
12 SPE-181691-MS

an unrealistic response of the bottomhole flowing pressure to sudden changes in flow rates. The transient
analytical solution and the Dynamic IPR appear to coincide at most times, except for some separation zone
that is soon overcome as both curves tend asymptotically to the maximum pressure value given by the
steady-state IPR. Similarly to what occurred in Case 1, Eq. 21 is not valid once the pressure drop reaches
the external boundary of the reservoir. The Dynamic IPR, on the other hand, converges to the steady-state
solution provided by the conventional IPR.

Figure 6—Comparison of the bottomhole flowing pressure calculated with steady-


state, transient, and Dynamic IPRs for a pressure drawdown followed by a buildup test.

Case 3: Severe Slugging


A typical example for which transient simulations of wells are required is the case of production oscillations.
These oscillations might be of different natures, such as severe slugging, casing heading, oscillations caused
by coupling of different wells in the production separator, density waves, to name a few. Both casing heading
and severe slugging will present a similar bottomhole flowing pressure signature as presented in Fig. 7a.

Figure 7—Gas Lift well subject to a casing heading oscillation, (a) bottomhole
flowing pressure, (b) comparison of flowrates calculated with different methods.
SPE-181691-MS 13

Simple expressions like the one from Eq. 19 are not available for this case, but if Pwf is decomposed
into a sum of sinusoidal functions, the time-domain analytical solution presented by Garcia et al. (2014)
(Eq. 14) could be applied separately to each sinusoidal function. The final analytical solution would be the
sum of the flow rates calculated for each sinusoidal bottomhole flowing pressure. However conceptually
interesting, this is approach is not useful for practical purposes and thus the reference flowrate for this case
is considered to be the one calculated from the finite difference method. This method has been extensively
applied in the oil industry and is familiar to all petroleum engineers.
Fig. 7b compares the flowrates calculated with the steady-state IPR, the finite differences method and
the Dynamic IPR. The Dynamic IPR and the finite difference solution present very similar results, but the
oscillation amplitude is close to 50% larger than the one predicted by the conventional IPR. It is important
to observed that the same Dynamic IPR equation was used to solve Cases 1 to 3, which means that a single
equation can approximate the reservoir dynamics in any condition.

Case 4: Coupled Well-bore Reservoir


Cases 1 through 3 were conceived to validate the Dynamic IPR against conventional solutions, while Case
4 focuses on illustrating the impact of the reservoir dynamics into the transient simulation of a well. For this
purpose, we consider a simplified model of a gas lift well to present the coupling of the well dynamics with
the Dynamic IPR. This model was developed by Eikrem et al. (2008). It is based solely on the conservation
of mass principle and it has the capability to qualitatively capture the main dynamics of casing-heading.
Consider the well model represented in Fig. 8, where the liquid in the tubing comes from the reservoir and
the gas origins from both the reservoir and from the annulus. Surface injection choke keeps the mass flow
rate constant and controls the gas flow rate into the annulus. The production choke regulates the outflow
to the production facilities (i.e. Production Separator).

Figure 8—Sketch of gas-lift oil well - modified from Eikrem et al.


(2008) - illustrating the elements of the simplified transient model.
14 SPE-181691-MS

This model consists of three state variables: the mass of gas in the annulus, the mass of gas in the tubing,
and the mass of gas in the tubing above the injection point. All non-constant flows are modeled as simplified
valve equations, including the reservoir. Here, however, we modified the model to use the steady-state IPR
and the Dynamic IPR.
The steady-state IPR represents the base case, with the bottomhole flowing pressure as the input and the
flowrate as the output. The usual justification for using the a steady-state IPR in transient simulation of wells
is that the reservoir parameters have very slow variations in comparison to the production dynamics and
therefore are treated as constants. This case shows that, contrary to that assumption, the reservoir dynamics
affects the behavior of the well and proving the importance of the coupled wellbore- reservoir simulations
to improve the reliability of transient simulations.
For comparison purposes, we coupled the Dynamic IPR (Eq. 6) with the transient well model introduced
by Eikrem et al. (2008). The well properties for the Eikrem et al. (2008) model are presented in Table 2.
The same reservoir properties from Table 1 were used to estimate the coefficients of the Dynamic IPR (Eq.
6). The gas injection rate was reduced until a casing-heading could be observed.

Table 2—Parameters of the simulation model.

Parameter Value Unit

M 0.029 kg/mol
R 8.31 J/Kmol

g 9.81 m/s2
Ta 293 K
La 2,400 m

Va 37.68 m3

ρo 1000 kg/m3

Ps 1×105 Pa
Wgc 0.024 kg/s

pr 2.2×10 7
Pa
Tw 400 K
Lw 2,400 m
Lr 150 m

Aw 0.316×10 -3
m2

Ar 0.316×10-3 m2

Civ 2×10-3 m2

Cpc 2×10-3 m2
rgo 0.001 –

Fig. 9 shows a comparison of the bottomhole pressure obtained from a transient simulation using a
conventional IPR against a transient simulation using a Dynamic IPR. The period of oscillation observed
in Figs. 9a and 9b for the results obtained using the conventional IPR is close to 50 minutes (that
is, approximately 3×10-4 Hz). According to the magnitude plot presented in Fig. 4a, at this frequency
the reservoir dynamics start becoming relevant. Once the Dynamic IPR is introduced in the transient
simulation of the well, the period of oscillation turns to 70 minutes: a 40% increase compared to using the
conventional IPR. The amplitude of oscillation of the bottomhole flowing pressure increases by 15% when
SPE-181691-MS 15

the reservoir dynamics is taken into account. The wellhead pressure peak also increases by about 20% with
the incorporation of the Dynamic IPR.

Figure 9—Comparison of application of a conventional IPR against the Dynamic IPR for an oscillatory
well, where (a) shows the bottomhole flowing pressure, and (b) shows the wellhead pressure.

The relative differences between the results obtained by considering the reservoir dynamics in relation
to the ones obtained by simply using the conventional IPR will increase as the frequency of oscillation
increases.
Next, we divided all the linear dimensions (lengths and radii) of the well (Table 2) by a factor of two
and we ran the transient simulations again. As seen in Fig. 10, the results obtained with the conventional
IPR indicated a stable well, with constant bottomhole flowing pressure and wellhead pressure. However,
the results obtained with the Dynamic IPR indicate that the well oscillates with a period of 2 hours and 40
minutes. These results indicate that the reservoir dynamics might have an influence on the stability of the
wells. It might be impossible to simulate the true dynamics of the coupled reservoir-wellbore system without
incorporating the reservoir dynamics. Garcia (2013) showed that an oscillatory behavior causes production
losses in a well, and that neglecting it can also compromise the process system's efficiency and safety.
16 SPE-181691-MS

Figure 10—Comparison of application of a conventional IPR against the Dynamic IPR for an oscillatory
well, where (a) shows the bottomhole flowing pressure, and (b) shows the wellhead pressure.

These numbers and the simplified wellbore model are used here for illustration purposes only, not as
a general guidance for production operations. They show that simulated results can be off when using
conventional IPR models in transient simulations, but how off depends on each system. The main point
we make here is that there are conditions in which the reservoir dynamics become relevant for adequate
transient simulations of wells.

Conclusion
This work presented classic transient solutions to the hydraulic diffusivity equations that are also relevant
to transient simulations of wells. The use of conventional steady-state IPR models is demonstrated to be
inadequate in these cases. By implicitly assuming an infinite transmissibility, the steady-state IPR models
simply cannot capture the transient behavior of the reservoir. Even though it is a common practice in the
industry, the use of simplified reservoir model in transient well simulations using commercial simulators
can have a significant influence in the reliability of the estimated results.
An alternative to the conventional IPRs is the direct coupling of a transient reservoir model with a
transient well model. This approach is likely more precise but it is not only computationally more expensive
but it also increases the license cost of the software. In this work, we present an alternative approach through
a dynamic "correlation" that can capture any transient behavior of the reservoir.
We verified that the Dynamic IPR could reliably estimate the reservoir response in classical cases such
as Buildup and Drawdown, where the maximum relative error was lower than 0.5%. We also applied it in
cases of practical relevance to transient simulations such as Severe Slugging and Casing Heading, where
the amplitude of the flowrate oscillation predicted by the conventional IPR was almost 50% lower than the
actual amplitude of oscillation.
We also demonstrated that the Dynamic IPR can be easily coupled with transient wellbore models
and give reliable results, especially when compared against results obtained through the application of
conventional steady-state IPR models. We showed a case where the casing-heading period is increased by
40% when the reservoir dynamics are incorporated in the transient simulation of wells. More important
than that, we also showed a case where neglecting the reservoir dynamics can lead one to conclude that
SPE-181691-MS 17

an oscillatory well is actually stable. This can affect both the production performance and the safety of
production operations.
Another possible application of the Dynamic IPR is to anti-slug control system design. Some simplified
well model is usually employed in the calculation the "gains" (constants of the controller). The Dynamic
IPR is already generated in a format suitable for control system applications, the transfer function. On the
other hand, there are more complex situations that were not addressed in this work, such as multi-phase
flow within the reservoir, and the dynamic formation of gas and water cones. At least in principle, a generic
Dynamic IPR given by Eq. 8 could be applied in those cases.

Acronyms
FT Fourier transform
IPR Inflow Performance Relationship
PI Productivity Index
VFP Vertical Flow Performance

Nomenclature
ak Coefficient dependent on reservoir and fluid properties
Ar Cross sectional area of the tubing below the injection point, m2
Aw Cross sectional area of the tubing above the injection point, m2
AΔP Amplitude of periodic drawdown oscillation, Pa
bk Coefficient dependent on reservoir and fluid properties
Bo Oil formation volume factor
ct Total compressibility, Pa-1
Civ Constant of the gas lift valve model, m2
Cpc Constant of the production choke model, m2
Ei Integral exponential function
f Pwf oscillation frequency, Hz
g Standard acceleration of gravity, m/s2
h Reservoir height/Net pay, m
i Imaginary number
k Permeability, Darcy
La Annulus length, m
Lr Distance from wellbore to gas lift valve, m
Lw Length of the tubing, m
M Molecular weight of the gas, kg/mol
pr Reservoir pressure used in Case 4, Pa
ps Separator pressure, Pa
P Pressure, Pa
Average reservoir pressure, Pa
Pe Reservoir pressure, Pa
Pe Pressure at the effective radius, Pa
Pe Reservoir static pressure and initial pressure, Pa
Pi Initial reservoir pressure, Pa
Pwf Bottomhole flowing pressure, Pa
PI Productivity index, m3/d/Pa
qo Oil flow rate, m3/d
18 SPE-181691-MS

re Effective radius of the reservoir


rgo Gas oil ratio, kg/s/kg/s
rw Wellbore radius, m
R Ideal gas constant, J/Kmol
t Time, s
Ta Temperature of the gas in the annulus, K
Tw Temperature of the fluid in the tubung, K
Va Annulus volume, m3
wgc Mass flow rate of gas into the annulus, kg/s
wiv Mass flow rate of gas through the gas-lift valve, kg/s
wrg Mass flow rate of gas from the reservoir into the tubing, kg/s
wpg Mass flow rate of gas through the production choke, kg/s
wro Flow rate of oil from the reservoir into the tubing, kg/s
wpo Mass flow rate oil through the production choke, kg/s
μ Dynamic viscosity, Pa.s
ρo Oil density, kg/m3
ω Angular frequency, rad/s
ϕ Reservoir porosity

References
Bahonar, M., Azaiez, J., and Chen, Z.J. 2011. Transient Nonisothermal Fully Coupled Wellbore/Reservoir Model for Gas-
Well Testing, Part 1: Modelling. Journal of Canadian Petroleum Technology.
Barron, R.F. and Barron, B.R. 2012. Design for thermal stresses. In. Hoboken, N.J.:: Wiley.
Beggs, H.D. 1991. Production Optimization: Using NODAL Analysis: OGCI Publications. Original edition. p. 2.
Dake, L.P. 1983. Fundamentals of Reservoir Engineering: Elsevier Science. Original edition. p. 159, 166, 177.
Economides, M.J., Hill, A.D., Ehlig-Economides, C., and Zhu, D. 2011. Petroleum Production Systems: Pearson
Education. Original edition. p. 19 –120, 261, 262, 275.
Eikrem, G.O., Aamo, O.M., and Foss, B.A. 2008. On Instability in Gas Lift Wells and Schemes for Stabilization by
Automatic Control.
Garcia, A.P. 2013. Stability analysis and stabilization of gas lift systems. In 22ndInternational Congress of Mechanical
Engineering, November: 3-7.
Garcia, A.P., Neto, G.M., Machado, M.B., and Carvalho, R.d.S. 2014. Dynamic IPR - Modeling Reservoir Well
Interactions to Improve Transient Simulations of Wells. Society of Petroleum Engineers.
Haberman, R. 2004. Applied Partial Differential Equations: With Fourier Series and Boundary Value Problems. Upper
Saddle River, N.J.: Pearson Prentice Hall. Original edition. p. 48, 108.
Hu, B. 2004. Gas Lift Instabilities, Norwegian University of Science and Technology, Trondheim, Norway
Hu, B., Sagen, J., Chupin, G., Haugset, T., Ek, A., and Sommersel, T. 2007. Integrated Wellbore- Reservoir Dynamic
Simulation. Society of Petroleum Engineers.
Izgec, B., Kabir, S., Ding, z., and Hasan, A.R. 2006. Transient Fluid and Heat Flow Modeling in Coupled Wellbore/
Reservoir Systems. Society of Petroleum Engineers.
Nennie, E.D., Alberts, G., Belfroid, S., Peters, L., and Joosten, G.J.P. 2007. An Investigation Into the Need of a Dynamic
Coupled Well-Reservoir Simulator. Society of Petroleum Engineers.
Pagan, E.V., Williams, W., and Waltrich, P.J. 2016. A Simplified Transient Model to Predict Liquid Loading in Gas Wells.
Society of Petroleum Engineers.
Sagen, J., Sira, T., Ek, A., Selberg, S., Chaib, M., and Eidsmoen, H. 2007. A Coupled Dynamic Reservoir and Pipeline
Model - Development and Initial Experience. BHR Group.
Vepa, R. 1976. On the use of Pade approximants to represent unsteady aerodynamic loads for arbitrarily small motions
of wings. In 14th Aerospace Sciences Meeting, Aerospace Sciences Meetings: American Institute of Aeronautics and
Astronautics.

Vous aimerez peut-être aussi