You are on page 1of 59

CHAPTER ONE

INTRODUCTION

1.1 STRATIFIED RESERVOIR SYSTEMS

Stratification in many reservoirs is attributable to varying energy levels of the depositional

environment in which the reservoirs originated. Such depositional environments experience some

sorting of the source material by the transporting medium as one layer is overlain by another layer

in the course of deposition. A change in depositional energy level would result in marked

dissimilarities in the composition and texture of overlying strata that are subsequently deposited.

Formations comprising such dissimilar units or strata are referred to as multilayered or stratified

systems or reservoirs. Frequently, some of the layers are separated by other layers with such low

permeabilities that they act as impervious continuous layers inhibiting vertical flow. Reservoir

crossflow between permeable layers is absent in such systems, with crossflow between layers

possible or occurring only through the wellbore that they are called multilayered without crossflow

or commingled in the theory of well testing. If vertical permeability is finite, on the other hand,

reservoir communication occurs between the layers and crossflow occurs in the formation. This

type of system is described as multilayered or stratified with crossflow. It is perhaps stating the

obvious that the fluid type (gas, water or oil) hosted in the reservoir is inconsequential to the

presence or lack of crossflow in the reservoir system.

Wells drilled into natural gas reservoirs are, amongst other uses, a means of accessing such

reservoirs for the purpose of formation evaluation and reservoir characterization.

1
Well tests conducted on gas wells fall into two main categories: Deliverability Tests and Pressure

Transient Tests. Deliverability tests measure the production capability of a gas well under specific

conditions of reservoir and bottomhole flowing pressures. A common productivity indicator is the

absolute open flow (AOF) potential, which is the theoretical maximum rate at which the gas well

can produce against a certain backpressure (atmospheric pressure) at the sandface. Deliverability

tests are also used to generate an Inflow Performance Relationship (IPR) curve or gas backpressure

curve for the well, which describes a relationship between surface production rate and bottomhole

flowing pressure for a specific value of reservoir pressure. The curve can be used to evaluate gas

well current deliverability potential under a variety of surface conditions such as production

against a fixed backpressure.

In pressure transient testing, pressure transients are generated and measured in the well as a

function of time. From the measured pressure response, important formation properties are

determined so as to improve reservoir management for the purpose of optimizing recovery. The

average reservoir pressure for the drainage volume, skin factor (non-Darcy skin inclusive) and

permeability of the reservoir in its dynamic state are some parameters determined from typical

pressure transient tests. This study is concerned with the analysis of well tests (pressure transient

tests) in stratified gas reservoirs and thus would not discuss gas deliverability tests any further.

1.2 PECULIARITIES OF GAS WELL TEST ANALYSIS

Pressure transient analysis of gas reservoirs is complex as compared to that for oil reservoirs. In

comparison with the governing equation for liquid flow, the diffusivity equation which describes

gas flow in porous media is a nonlinear partial differential equation containing pressure dependent

2
terms such as gas viscosity, gas density and compressibility factor (and thus compressibility) as

coefficients. Further complications arise due to different flow regimes. For the flow of real gases

through porous media, the inertial and turbulence effects are important when flow rates are high.

Currently, the full analytical solution for the nonlinear gas flow equation can not be obtained. The

usual practice in gas well testing is to transform the nonlinear gas flow equation into an

approximately linear form which is similar to the liquid flow equation. Foremost among several

analytical methods introduced to linearize the gas flow equation is the use of pseudovariables –

pseuodopressure (Al-Hussainy et al, 1966) and pseudotime (Agarwal, 1979)). With such effective

linearization, the same solutions for the liquid flow equation can be utilized for the gas flow in

porous media. One drawback of the use of pseudovariables is the loss of intuitive feel associated

with the use of familiar units. Meunir et al (1987) introduced the use of normalized pseudovariables

which restores the intuitive use of practical units of pressure and time, makes the flow equations

for gas identical in form to the liquid cases and enables the direct use of the constants of the

working equations of the liquid cases for the gas cases.

It is worthy of note that with the help of modern computers, the nonlinear gas flow equation can

be completely solved with a numerical approach.

Evidently, the analysis of well test in a homogenous gas reservoir is a hard enough task to perform

manually. The theory of pressure transient testing of wells which presupposes a single-layer

homogenous formation has the implication that parameters determined from the analysis of well

tests reflect the “average” or “effective” values for the whole formation. Such single-value

estimates are not indicative of the individual layer properties should layering be present in such

reservoirs. So, reservoir heterogeneity with vertically stacked layers possessing variations in layer

3
properties presents an even more arduous task when analysis of well tests in such stratified gas

reservoirs is performed.

1.3 PRODUCTION LOGGING AND MULTILAYER TESTING

In the treatment of multilayer well testing, production logging tools provide a means of acquiring

useful information. Conventional pressure transient analysis only requires measurement of

downhole pressure and total surface rate. With this restricted amount of information, only average

properties of the reservoir are determined and the total system behavior gives no indication of the

presence of layering. Hence, production logging devices are the key to the proper understanding

of layered reservoirs, especially of the commingled type. The flow profile obtained by up and

down passes of a spinner flow meter is the extra information which must be added to the pressure

response. It must be emphasized that up and down passes of a production logging tool are

necessary to allow determination of the layer pressures and skin factors. For a system where there

is interlayer crossflow, production logging tools are still useful but provide less benefits than for

commingled systems.

If the myriad models resulting from research into multilayered reservoirs are to have any chance

at practical application, advancement in the current tools for production logging is also needed.

The difficulty of measuring individual layer flow rates directly places undesirable limitations on

the full application of methods developed for transient multilayer reservoir testing.

An effective synergy of layer well testing and production logging technology would continue to

provide a useful means of extracting dynamic information from stratified reservoir systems.

4
1.4 STATEMENT OF PROBLEM

In Nigeria and the rest of the world, a number of multilayered gas reservoirs provide the gas energy

needed for both commercial and domestic use. Economically and technically profitable

management of such reservoirs rely heavily on acquisition of both dynamic and static reservoir

data. Pressure Transient Analysis is a useful means of getting important information about such

reservoirs. Conventional analysis methods only yield bulk or average properties of such systems,

whereas individual layer information is desired. Granted, an impressive volume of research has

been provoked in the field of well test analysis of layered oil reservoirs, there is limited literature

focused on the analysis of well tests in stratified gas reservoirs with and without crossflow. This

dearth of research furnishes appropriate justification for more studies such as this work.

1.5 AIMS AND OBJECTIVES OF STUDY

The primary objectives of this research work are:

 To develop an analytical solution for a multilayered gas reservoir to generate pressure

transient responses both for crossflow and commingled situations

 Identify the parameters affecting the pressure transient responses

 Investigate the finite formation damage responses

 To analyze the multilayer responses by layering heterogeneity

The overall intention of this research work is to attempt to provide a new model that will

facilitate the evaluation of individual layer properties of stratified formations. The means of

acquisition of data required for well tests in stratified gas formations is not the focus of this

study.

5
CHAPTER TWO
REVIEW OF RELEVANT LITERATURE

Interest in multilayered reservoirs has spawned numerous studies since the early 1960s as

seen in the first published work on the subject by Jacquard (1960). He modeled a two-layer

crossflow system with transient interlayer flow (in keeping with dual-porosity systems

terminology) with a well of finite wellbore producing at a constant rate and a no-flow

external boundary. Laplace transforms were utilized in his solution method resulting in

purely mathematical solutions as no numerical evaluations of the results were presented.

Lefkovits (1961) is credited with the earliest rigorous work on commingled stratified

reservoir systems. He presented detailed mathematical derivations for a two-layered

commingled reservoir system. Theoretical results on buildup curves were applied to field

examples to determine the average formation properties such as the permeability-thickness

product, the wellbore damage and the static pressure. The paper noted that it was not

possible to determine properties of individual layers from the buildup curves. He also found

that early-time layer flow rate was governed basically by the permeability-thickness

product and skin, and the late-time flow rate increasingly depend on oil-filled volume and

compressibility. It is worthy of note that Templaar-Lietz (1961) had studied the

performance of a two-layer reservoir without crossflow and a no-flow outer boundary prior

to Lefkovits (1961), but the treatment was a simplified approach that did not cover the

entire wellbore pressure history.

6
Studies in multilayered systems with crossflow were pioneered by Russel and Prats (1962a)

and Katz and Tek (1962). A constant pressure inner boundary condition at the wellbore

was assumed in both papers to solve the problem in a two-layer system. An interesting

observation from the work is that the behavior of a multilayered crossflow system is

bounded by that of a commingled system without formation crossflow and a homogenous

system. The former employed successive application of Hankel and Laplace transforms

while the latter utilized Fourier transforms in finding solutions to their models. In a second

paper, Russel and Prats (1962b) dealt with the practical aspects of crossflow, opining that

crossflow between communicating layers is economically advantageous in that operating

life of the reservoir is shortened and primary ultimate recovery is raised.

Concentrating on the effect of permeability distribution between layers, Pendergrass and

Berry (1962) solved the same problem for an n-layered system with the conclusion that it

was not possible to infer stratification from wellbore pressure data except at early time.

Negligence of the vertical permeabilities of the layers may have informed that conclusion.

Kazemi and Seth (1969) used an inner boundary condition where flow entry into the

wellbore was restricted and observed two distinct semilog straight lines in the pressure

response with the early one yielding the flow capacity of the perforated interval and the

latter one giving that of the entire formation. Kazemi (1970) also determined the reservoir

limit using a pressure buildup test. While conventional methods could only be applied

cautiously for the commingled systems, they worked well for crossflow systems.

7
Cobb, et al (1972) unsuccessfully tried to determine individual layer parameters by

examining the wellbore pressure response of a two-layer reservoir for various production

and shut-in conditions. Raghavan, et al (1974) studied a similar problem with unequal

formation thickness and were able to estimate the individual layer properties in some

favorable cases by displaying the data in various ways.

Pressure behavior in buildup tests for bounded multilayered reservoirs without crossflow

(commingled systems) was studied by Earlougher et al (1974). A number of unique

observations were presented, especially for cases of multiple wells and noncircular

reservoir boundaries.

For the commingled systems, the model development was extended by Tariq and Ramey

(1978) whose contributions include the introduction of wellbore storage and skin, and the

novel use of the Stehfest (1970) algorithm for the numerical inversion of Laplace

transforms. The use of this algorithm has opened the way for the determination of

numerous solutions which had proved difficult or even impossible to invert using direct

complex analysis. Practical application of such inverted solutions in well test analysis has

thankfully ensued.

The period beginning from the 1980s experienced substantial efforts to interpret

multilayered systems quantitatively with the introduction of production logging tools

(PLTs) that enable the simultaneous measurement of bottomhole flow rate and pressure.

Studies from then on incorporated this new approach. Gao and Deans (1983) introduced

an idea to reduce the dimension of the equation that govern flow in multilayered systems

8
by one and presented an analytical solution for multilayer reservoirs with formation

crossflow without wellbore storage and skin. They used the “semipermeable wall model”

originally developed by Polubarinova-Kochina (1961) where the terms for vertical

movement were represented by a pseudo-steady state approximation.

Bourdet (1985) solved the pressure response for a two-layer crossflow system including

wellbore storage and skin. He showed that interpretation models such as homogenous

reservoir, two layers without crossflow and double porosity reservoir are limiting forms of

his solution.

Kucuck et al (1984) proposed a new testing method for a two-layer commingled reservoir.

They coupled sandface production rate of each layer with wellbore pressure in the

technique of nonlinear parameter estimation. The coupling of layer production rate is very

significant for multilayered reservoirs because rate transient of each layer reveals

information about that layer while wellbore pressures are determined more by average

reservoir parameters.

Raghavan et al (1985) investigated the pressure response of a two-layer reservoir with

formation crossflow. The effects of various heterogeneities were considered. Even though

they applied the concept of a thick skin, unlike most authors before them, the overall effects

on the wellbore pressures were no different from those of authors who considered a thin

skin. The work demonstrated that the skin factors of the individual layers could be

determined uniquely with production rate data when other parameters, including average

skin factor of the total system, were known.

9
Ehlig-Economides and Joseph (1985) presented a study of well test analysis for

multilayered system with crossflow that represented a major advance. The work includes

an excellent literature survey up to 1985. They developed a complete analytical solution

for the n-layered crossflow system with wellbore storage and skin, and investigated early

tome and late time behavior of the production rate for each layer.

Larsen (1988) reviewed the similarities and differences of the mathematical methods for

various multilayered models. He concluded that the pressure response could be varied by

a change in reservoir modeling, such as the numbers of sublayers into which a single layer

was discretized to simulate a real reservoir under crossflow conditions.

Some researchers have focused on the response of a multilayered reservoir for multi-well

scenarios. Woods (1970) concluded that the apparent transmissivity in a commingled

system obtained by pulse testing is always greater than or equal to the total transmissivity

of the entire system, an observation that was later verified by Larson (1982). He also

proposed the use of flowmeters to obtain flow rate data to determine individual layer

properties from pulse tests between wells. Interference testing for multilayered reservoirs

has been studied by Chu and Raghavan (1981) where they observed that skin factors of the

active well play an important role in the pressure response of the observation well, unlike

the behavior of a single-layer system.

A combination of multilayered systems and fractured reservoirs have been studied by some

authors such as Camacho-V et al (1984) and Bennet et al (1986). Some authors like Lee,

10
et al (1984), Economides and Ayoub (1984) and Bremer et al (1985) have studied the three

layer problem in which a tight layer is present between two permeable layers.

Park and Horne (1989) observed that in the analysis of well test in multilayered formation

with crossflow, the vertical permeability is the only parameter governing the initiation and

termination of the transition from the early time commingled system response to the late

time homogenous system response. They also utilized the method of nonlinear parameter

estimation as a means to determine the layer properties.

Olarewaju and Lee (1989) presented correlations for estimating skin due to partial

penetration for homogenous and layered systems with contrasting layer permeabilities.

Using automatic history matching approach, they presented a field example application of

their analytical model and correlations.

Raghavan (1989) discussed methods to determine layer properties and examined the

consequences of selectively stimulating the layers of the reservoir. The work documented

methods to predict future performance in stratified reservoirs. The advantages of

combining rate and pressure data to identify layer properties such as flow capacity, skin

factor, and average reservoir pressure are documented.

Olarewaju and Lee (1990) presented analytical solutions for the rate performance of a well

intercepting a bounded two-layer reservoir. The solution was for a modelling two-layer

reservoirs with or without crossflow, with each layer having different layer skins and

drainage radii. The effects of partial completion and of different layer skins and drainage

radii on the rate behavior in layered reservoirs were also studied.


11
Kiome (1991) used a multilayered composite reservoir model to study the effects a titled

front on well test analysis. An analytical solution for a multilayered composite reservoir

was employed to generate transient pressure responses.

Bidaux (1992) used a technique based on an analytical conversion of a single-layer

transient pressure response into multi-layer response, taking into account formation

crossflow and a wide variety of inner and outer boundary conditions. Regression was used

to simultaneously match observed pressures and layer rates.

Gomes and Ambastha (1993) developed an analytical solution for multilayered composite

reservoirs with pseudosteady-state interlayer crossflow. The developed analytical solution

for an n-layered composite reservoir, they claimed, is applicable for any irregularly-shaped

discontinuity boundary, and for closed constant pressure and infinite outer boundary

conditions. They validated their model by comparing with some simple well-known

models in the well testing literature.

Poe (1993) presented a semi-analytic reservoir model for gas reservoirs with various

reservoir complexities including multiple reservoir layers in which the reservoir layers may

be infinite or finite in extent, dual or single porosity systems and fractured wells.

Shah and Spath (1993) studied the problems of design and interpretation of layered

reservoir tests in commingled wells when the layer potentials are different where the

difference may be in the conditions either at the initial time or at the outer layer boundaries.

They constructed the multilayer models for commingled wells from existing single layer

analytic solutions to account for different properties and boundary conditions.


12
Sahni et al (1995) presented an analysis of the transient pressure behavior in a multilayered

faulted reservoir with sealing and non-sealing faults. They used a 3D numerical model to

generate transient pressure responses at the producing well. A methodology to estimate the

fault and formation transmissibility as well as the distance of the fault to the producing well

was produced.

Savioli (1999) studied the pressure response of layered reservoirs with crossflow in the

presence of gravity effects. Continuing from the work of Prijambodo et al (1985), they

proposed a model that takes into account the spatial variations of permeability and porosity

together with gravity effects. They concluded that the pressure response of the crossflow

reservoirs lies between the response of the equivalent single layer system and that of the

isolated layer system. The response was divided into three flow periods where at early

times, it behaves as if the layers are isolated. At a transitional period, the pressure response

depends on the interlayer flow; as vertical permeability increases, the transitional period

starts earlier. At later times, the reservoir behaves as an equivalent single layer

homogeneous system.

Hedong et al (2003) presented a model called the semi-permeable wall model for multilayer

gas reservoir where the problem was approximated to have a mathematically simpler

problem. The model accounted for the effect of skin and wellbore storage and considered

all layers open to a single well which flows at a constant total rate. It was found that the

direction of crossflow is governed by the skin factors. The crossflow starts from the less

permeable layer to the more permeable layer in the beginning and from the layer with

13
greater skin factor later. They also observed that when all the layers have skin factors of

the same magnitude, the ratio is exactly the same as the productivity ratio.

Odili (2015) mathematically investigated the performance of a gas well in a bounded

layered reservoir with interlayer crossflow. A diffusivity equation describing the stratified

reservoir with interlayer crossflow was developed and solved analytically using Hankel

and Laplace transforms. A key observation is that, at constant pressure the production rate

was seen to decline exponentially after the early transient behavior has elapsed. The study

also recommends that should a good degree of communication exist between the tight and

permeable layer, a satisfactory completion could result from perforating and treating the

permeable layers since the tight layer can be depleted by formation crossflow from the tight

layer to the more permeable layer thereby reducing completion costs, storage costs and

increasing the life span of the well. However, the study was not applied to the analysis of

well tests in stratified gas reservoirs for the determination of layer parameters.

Despite the volume of literature available on the subject of pressure behavior of stratified

systems, the portion devoted to analysis of well test in stratified gas reservoirs is at best

scanty. This necessitates further studies on the subject. This research work attempts a

methodology for the analysis of well test in stratified gas reservoirs with and without

crossflow.

14
CHAPTER THREE

MODEL DEVELOPMENT

A brief discussion of modelling directions in pressure transient analysis of commingled


and crossflow stratified systems should precede the detailed model development to follow
shortly. The stratified reservoirs without interlayer crossflow (also termed commingled
systems) lend themselves to easier mathematical manipulation as seen in Lefkovits et al
(1961) and Tariq and Ramey (1978). For layered systems with interlayer crossflow, the
situation compels more mathematical rigor. The motley approximations normally used for
such systems both in the mathematical model and analytical methods employed in the
derivation of solutions bear out the validity of that observation. Though exact solutions
based on double integral transformations can be found in the literature, the complexity of
the solutions have dissuaded, and thus limited, their use in interactive analysis which is a
cornerstone of modern well test analysis. Hence, instead of exact mathematical models of
stratified reservoirs that account for a two-dimensional pressure distribution in crossflow
layers, the approach that considers pressure derivatives in one direction and pressure
differences or ignoring flow in the other is employed in this work primarily due to the
advantage of interactive computer-assisted analysis.

From the interactive computer-assisted analysis perspective, the pressure transient or rate
transient analysis of well test data from stratified crossflow systems are best suited for the
foregoing.

3.1 SYNOPSIS OF MODEL ASSUMPTIONS

The model developed in this study considers an n-layer, radial gas reservoir with a
symmetrically located well penetrating the reservoir fully. The reservoir is made up of
horizontal layers each of which is homogenous laterally but different from each other. The

15
reservoir is bounded at the top and bottom by impermeable layers while the typical
conditions of no-flow, constant pressure and infinitely large reservoir are considered at the
outer boundary. A constant rate production of single phase compressible fluid is assumed
at the surface. Formation crossflow is modeled using the pseudosteady-state approximation
introduced in groundwater modelling by Polubarinova-Kochina (1961) and later used in
petroleum literature by Gao (1984). The pseudosteady-state formation crossflow assumes
that crossflow resistance is confined to the interlayer boundary and flow in each layer is
horizontal. Any two layers are assumed to be separated by a “semi-permeable wall”.
Hence, the pressure difference between any two adjacent layers is dependent on lateral
(radial) position and time. Figure 3.1 shows the model schematic.

𝑞𝑇

𝑠1 𝜑1 𝑞1 𝑘𝑟1

𝑠2 𝜑2 𝑞2 𝑘𝑟2

𝑠3 𝜑3 𝑞3 𝑘𝑟3

𝑠4 𝜑4 𝑞4 𝑘𝑟4

𝑠5 𝜑5 𝑞5 𝑘𝑟5

𝑠6 𝜑6 𝑞 6 𝑘𝑟6

𝑠7 𝜑7 𝑞7 𝑘𝑟8

𝑠8 𝜑8 𝑞8 𝑘𝑟8

𝑠9 𝜑9 𝑞9 𝑘𝑟9

𝑠𝑗 𝜑𝑗 𝑞𝑗 𝑘𝑟𝑗

Fig. 3.1 - SCHEMATIC OF A STRATIFIED GAS RESERVOIR WITH INTERLAYER CROSSFLOW 16


3.2 DIFFUSIVITY EQUATION FOR THE STRATIFIED RADIAL GAS
RESERVOIR WITH AND WITHOUT CROSSFLOW

The diffusivity equation for the flow of gas in a stratified gas reservoir with and without
cross flow is developed as follows:

The continuity equation for the system can be written as

𝜕𝜌𝜙
∇(𝜌𝑣̅ ) = − (3.1)
𝜕𝑡

Since the description of flow in stratified systems in this works requires both vertical interlayer
flow and radial flow for the cylindrical geometry assumed, the continuity equation (3.1) can be
written for a 2-D cylindrical system as:

1 𝜕 𝜕 𝜕𝜌𝜙
(𝜌𝑣̅𝑟 ) + (𝜌𝑣̅𝑧 ) = − (3.2)
𝑟 𝜕𝑟 𝜕𝑟 𝜕𝑡

The transport equation for the system is taken to be the Forchheimer equation as formulated by
Swift and Kiel (1962) as,

𝛿𝑟 𝑘𝜕𝑝
𝑣̅ = − (3.3)
𝜇 𝜕𝑟

Where,

1
𝛿= 𝛽𝜌𝑘 (3.4)
1+ |𝑣̅ |
𝜇

And the equation of state for the fluid (gas) is given by the real gas law as,

17
𝑝𝑀
𝜌= (3.5)
𝑍𝑅𝑇

Combining these equations leads to the pressure differential form of the diffusivity equation,

1 𝜕 𝑝𝑀 𝛿𝑟 𝑘𝑟 𝜕𝑝 𝜕 𝑝𝑀 𝛿𝑧 𝑘𝑧 𝜕𝑝 𝜕 𝑝𝑀
(𝑟 𝑧𝑅𝑇 ) + ( ) = ( 𝜙) (3.6)
𝑟 𝜕𝑟 𝜇 𝜕𝑟 𝜕𝑟 𝑧𝑅𝑇 𝜇 𝜕𝑧 𝜕𝑡 𝑧𝑅𝑇

1 𝜕 𝑝 𝛿 𝑘 𝜕𝑝 𝜕 𝑝 𝛿 𝑘 𝜕𝑝 𝜕 𝑝
(𝑟 𝑧 𝑟𝜇 𝜕𝑟
𝑟
) + 𝜕𝑧 ( 𝑧 𝑧𝜇 𝜕𝑧
𝑧
) = 𝜕𝑡 (𝑧 𝜙) (3.7)
𝑟 𝜕𝑟

Use of Kirchhoff transformation has been made by Al-Hussainy, Ramey and Crawford (1966) to
define the real gas pseudopressure as:

𝑝 𝑝
𝑚(𝑝) = 2 ∫𝑝 𝑑𝑝 (3.8)
0 𝜇(𝑝)𝑧(𝑝)

By choosing the variables at the initial reservoir conditions, Eq. (3.8) can be normalized. Thus,

𝑖 𝑖𝜇 𝑧
𝜓(𝑝) = 2𝑝 𝑚(𝑝) (3.9)
𝑖

Where 𝜓(𝑝) is the normalized Pseudopressure.

From Meunier (1987), the above equation can be written in the form:

𝜇 𝑝 𝜌(𝑝)
𝜓(𝑝) = 𝜌𝑖 ∫𝑝 𝜇(𝑝)𝑧(𝑝) 𝑑𝑝 (3.10)
𝑖 0

18
Or
𝜇 𝑝 𝜌(𝑝)
𝜓(𝑝) = 𝑝𝑖 + 𝜌𝑖 ∫𝑝 𝜇(𝑝)𝑧(𝑝) 𝑑𝑝 (3.11)
𝑖 0

Thus, both real and normalized pseudopressure are numerically equal at initial pressure.

Agarwal (1979) introduced the pseudotime as:

𝑡 1
𝑡𝑝 = ∫0 𝑑𝑡 (3.12)
𝜇(𝑝)𝑐𝑡 (𝑝)

Normalizing the pseudotime yields;

𝑡 1
𝑡𝑝𝑛 = 𝜇𝑖 𝑐𝑖 ∫0 𝑑𝑡 (3.13)
𝜇(𝑝)𝑐𝑡 (𝑝)

For constant reservoir and gas properties, a laminar flow solution can be obtained where 𝛿𝑟 = 1;

2
𝜕 𝜓(𝑝)𝑗 1 𝜕𝜓(𝑝)𝑗 𝜕𝜓(𝑝)𝑗
(𝑘ℎ)𝑗 ( 2 +𝑟 𝜕𝑟
) = (𝜙ℎ)𝑗 𝜕𝑡𝑝𝑛
(3.14)
𝜕𝑟

The use of normalized pseudovariables enables the direct use of the liquid equations for the gas
cases. Assuming that pseudo-steady crossflow exists between the layers, discretizing the vertical
flow term with the pseudo-steady state approximation and expressing in terms of normalized
pseudopressure, Eq. (3.7) becomes:

2
𝜕 𝜓(𝑝)𝑗 1 𝜕𝜓(𝑝)𝑗 𝜕𝜓(𝑝)𝑗
(𝑘ℎ)𝑗 ( 2 +𝑟 𝜕𝑟
) = (𝜙ℎ)𝑗 𝜕𝑡 + 𝑋𝐴𝑗 [𝜓(𝑝)𝑗 − 𝜓(𝑝)𝑗+1 ] + 𝑋𝐵𝑗 [𝜓(𝑝)𝑗 −
𝜕𝑟 𝑝𝑛

𝜓(𝑝)𝑗−1 ] (3.15)

19
Where 𝑋𝐴𝑗 and 𝑋𝐵𝑗 are defined as follows:

2
𝑋𝐴𝑗 = ℎ ℎ (3.16)
( ) +( )
𝑘𝑣 𝑗 𝑘𝑣 𝑗+1

2
𝑋𝐵𝑗 = ℎ ℎ (3.17)
( ) +( )
𝑘𝑣 𝑗 𝑘𝑣 𝑗−1

Note that pressure is initially identical for all layers of the stratified gas reservoir. However, this
is not necessarily always the case for all layered systems.

The initial condition (IC) required for the solution is;

𝜓(𝑝)𝑗 [𝑟, 0] = 𝜓(𝑝)𝑖 (3.18)

Inner boundary conditions for a fully penetrating well with wellbore storage and skin can be
written in the form;

𝜕 𝜓(𝑝)𝑗
𝜓(𝑝)𝑤𝑓 = 𝜓(𝑝)𝑖 [𝑟𝑤 , 𝑡] − 𝑠𝑗 (𝑟 ) for j=1,…n (3.19)
𝜕𝑟

𝜕𝜓(𝑝)𝑤𝑓 𝑘ℎ 𝜕 𝜓(𝑝)𝑗
𝑞=𝐶 − 2𝜋𝑟𝑤 ∑𝑛𝑗=1 ( ) ( ) (3.20)
𝜕𝑡 𝜇 𝑗 𝜕𝑟 𝑟=𝑟𝑤

The sandface flow rate from each layer can be explicitly written as:

𝑘ℎ 𝜕𝜓(𝑝)𝑗
𝑞𝑗 = 2𝜋𝑟𝑤 ( ) ( ) (3.21)
𝜇 𝑗 𝜕𝑟 𝑟=𝑟𝑤

Where C is the wellbore storage coefficient, q is the total in situ well flow rate (specified as
constant), and 𝑠𝑗 are the layer skin factors.

20
Three outer boundary conditions are considered as given by the equations below;

Infinite-acting Outer Boundary Condition:

Lim 𝜓(𝑝)𝑗 = 𝜓(𝑝)𝑖 j = 1,…n (3.22)


𝑟→∞

Constant pressure Outer Boundary Condition;

𝜓(𝑝)𝑗 = 𝜓(𝑝)𝑖 𝑓𝑜𝑟 𝑟 = 𝑟𝑒𝑗 j = 1,…n (3.23)

Closed Outer Boundary Condition;

𝜕𝜓(𝑝)𝑗
=0 𝑓𝑜𝑟 𝑟 = 𝑟𝑒𝑗 (3.24)
𝜕𝑟

To reduce clumsiness and improve compactness of the equations for further mathematical
manipulation, the following dimensionless variables are defined for the stratified gas reservoir.

𝑟
𝑟𝐷 = (3.25)
𝑟𝑤

̅̅̅̅)
2𝜋(𝑘ℎ 𝑇𝑠𝑐
𝜓(𝑝)𝐷𝑗 =
𝑞𝑠𝑐 𝜓(𝑝)𝑠𝑐
(𝜓(𝑝)𝑖 − 𝜓(𝑝)𝑗 ) (3.26)

̅̅̅̅)
2𝜋(𝑘ℎ 𝑇𝑠𝑐
𝜓(𝑝)𝐷𝑤 =
𝑞𝑠𝑐 𝜓(𝑝)𝑠𝑐
(𝜓(𝑝)𝑖 − 𝜓(𝑝)𝑤𝑓 ) (3.27)

𝑡𝑝𝑛 ̅̅̅̅
𝑘ℎ
𝑡𝑝𝑛𝐷 = 2 (̅̅̅̅) (3.28)
𝑟𝑤 𝜙ℎ

The diffusivity equation (3.15) can be written in dimensionless variables as,

𝜕2 𝜓(𝑝)𝐷𝑗 1 𝜕𝜓(𝑝)𝐷𝑗 𝜕𝜓(𝑝)𝐷𝑗


Ƙ𝑗 [ 2 + ] = 𝜔𝑗 + 𝐴𝑗 [𝜓(𝑝)𝐷𝑗 − 𝜓(𝑝)𝐷,𝑗+1 ] +
𝜕𝑟𝐷 𝑟𝐷 𝜕𝑟𝐷 𝜕𝑡𝑝𝑛𝐷

𝐵𝑗 [𝜓(𝑝)𝐷𝑗 − 𝜓(𝑝)𝐷,𝑗−1 ] (3.29)

21
(𝑘ℎ)𝑗
Where, Ƙ𝑗 = ̅̅̅̅) (3.30)
(𝜙ℎ

(𝜙ℎ)𝑗
𝜔𝑗 = (3.31)
(̅̅̅̅̅
𝜙ℎ)

𝑋𝐴𝑗 𝑟𝑤 2
𝐴𝑗 = ̅̅̅̅)
(𝑘ℎ
(3.32)

𝑋𝐵𝑗 𝑟𝑤 2
𝑩𝑗 = = (3.33)
(̅̅̅̅̅
𝑘ℎ)

The initial and boundary conditions in dimensionless from are as follows;

Initial condition

𝜓(𝑝)𝐷𝑗 [𝑟𝐷 , 0] = 0 𝑗 = 1, … 𝑛 (3.34)

Inner boundary condition,

𝜕 𝜓(𝑝)𝐷𝑗
𝜓(𝑝)𝐷𝑤 = 𝜓(𝑝)𝐷𝑗 − 𝑠𝑗 𝑗 = 1, … 𝑛 (3.35)
𝜕 𝑟𝐷

𝜕𝜓(𝑝)𝐷𝑤 𝜕𝜓(𝑝)𝐷𝑤
1 = 𝐶𝐷 − ∑𝑛𝑗=1 Ƙ𝑗 𝑗 = 1, … 𝑛
𝜕𝑡𝐷 𝜕𝑟𝐷

𝐶
Where 𝐶𝐷 = = 2 (𝜙ℎ
̅̅̅̅ )𝑐 (3.36)
2𝜋 𝑟𝑤

Sandface flow rate from each layer:

𝑞𝑗 𝜕𝜓(𝑝)𝐷𝑗
𝑞𝐷𝑗 = = −Ƙ𝑗 𝑎𝑡 𝑟𝐷 = 1, 𝑓𝑜𝑟 𝑗 = 1, … 𝑛 (3.37)
𝑞𝑡 𝜕𝑟𝐷

Outer boundary conditions:

Infinite-acting Outer Boundary Condition;

𝜓(𝑝)𝐷𝑗 = 0 𝑟𝐷 → ∞ 𝑗 = 1, . . 𝑛 (3.38)
22
Constant pressure outer boundary condition;

𝜓(𝑝)𝐷𝑗 = 0 𝑟𝐷 = 𝑟𝑒𝐷 𝑗 = 1, . . 𝑛 (3.39)

Closed boundary condition;

𝜕𝜓(𝑝)𝐷𝑗
=0 𝑟𝐷 = 𝑟𝑒𝐷 𝑗 = 1, . . 𝑛 (3.40)
𝜕𝑟𝐷

3.3 SOLUTION TO THE DIFFUSIVITY EQUATION FOR STRATIFIED GAS


RESERVOIR WITH INTERLAYER CROSSFLOW

Utilizing the convenience of solution in Laplace domain, the diffusivity equation and its initial and
boundary conditions are transformed into equivalent equations in Laplace space. The Laplace
transforms of Equations (3.28) through (3.38) are given below:

Laplace Transform of diffusivity equation;

𝑑2 ̅̅̅̅̅̅̅
𝜓(𝑝)𝐷𝑗 ̅̅̅̅̅̅̅
1 𝑑𝜓(𝑝)
𝜓(𝑝)𝐷𝑗 𝑧 + 𝐴𝑗 [𝜓(𝑝)
] = 𝜔𝑗 ̅̅̅̅̅̅̅ 𝜓(𝑝)𝐷,𝑗+1 ] + 𝐵𝑗 [𝜓(𝑝)
̅̅̅̅̅̅̅𝐷𝑗 − ̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅𝐷𝑗 −
𝐷𝑗
Ƙ𝑗 [ +𝑟
𝑑𝑟𝐷 2 𝐷 𝑑𝑟𝐷

̅̅̅̅̅̅̅
𝜓(𝑝)𝐷,𝑗−1 ] (3.41)

Where 𝑧 is the Laplace space variable.

Laplace transform of initial conditions;

𝑑 𝜓(𝑝)𝐷𝑗 ̅̅̅̅̅̅
̅̅̅̅̅̅
𝜓 ̅̅̅̅̅̅
(𝑝)𝐷𝑤 = 𝜓 (𝑝)𝐷𝑗 − 𝑠𝑗 (3.42)
𝑑 𝑟𝐷

1 ̅̅̅̅̅̅̅𝐷𝑤
𝑑𝜓(𝑝)
̅̅̅̅̅̅̅𝐷𝑤 𝑧 − ∑𝑛 Ƙ𝑗
= 𝐶𝐷 𝜓(𝑝) (3.43)
𝑧 𝑗=1 𝑑𝑟𝐷

23
Laplace transform of outer boundary conditions;

Infinite-acting Outer Boundary Condition;

̅̅̅̅̅̅
𝜓 (𝑝)𝐷𝑗 = 0 𝑟𝐷 → ∞ 𝑗 = 1, . . 𝑛 (3.44)

Constant pressure outer boundary condition;

̅̅̅̅̅̅
𝜓 (𝑝)𝐷𝑗 = 0 𝑟𝐷 = 𝑟𝑒𝐷 𝑗 = 1, . . 𝑛 (3.45)

Closed boundary condition;

̅̅̅̅̅̅̅𝐷𝑗
𝑑𝜓(𝑝)
=0 𝑟𝐷 = 𝑟𝑒𝐷 (3.46)
𝑑𝑟𝐷

Equation (3.41) which is the diffusivity equation in Laplace space has the form of the modified
Bessel’s equation with a solution of the form:

̅̅̅̅̅̅̅
𝜓 (𝑝)𝐷𝑗 = 𝐴𝑗 𝐾0 (𝜎𝑟𝐷 ) + 𝐵𝑗 𝐼0 (𝜎𝑟𝐷 ) (3.47)

Differentiating Eq. (3.47) with respect to 𝑟𝐷 gives:

̅̅̅̅̅̅̅𝐷𝑗
𝑑𝜓(𝑝)
= −𝐴𝑗 𝜎𝐾1 (𝜎𝑟𝐷 ) + 𝐵𝑗 𝜎𝐼1 (𝜎𝑟𝐷 ) (3.48)
𝑑𝑟𝐷

Hence,

𝑞 𝐷𝑗 = Ƙ𝑗 𝐴𝑗 𝜎𝐾1 (𝜎𝑟𝐷 ) − 𝐵𝑗 Ƙ𝑗 𝜎𝐼1 (𝜎𝑟𝐷 ))


̅̅̅̅ (3.49)

Introducing Eq. (3.47) into the left-hand side of Eq. (3.41) gives;

24
𝑑2 ̅̅̅̅̅̅̅
𝜓(𝑝)𝐷𝑗 ̅̅̅̅̅̅̅
1 𝑑𝜓(𝑝)
) = Ƙ𝑗 𝜎2 ̅̅̅̅̅̅̅
𝐷𝑗
Ƙ𝑗 ( 2 +𝑟 𝜓(𝑝)𝐷𝑗 (3.50)
𝑑𝑟𝐷 𝐷 𝑑𝑟𝐷

Substitution of Eq. (3.46) into Eq. (3.39) gives the general formulation below:

Ƙ𝑗 𝜎2 ̅̅̅̅̅̅̅ 𝜓(𝑝)𝐷𝑗 𝑧 + 𝐴𝑗 [𝜓(𝑝)


𝜓(𝑝)𝐷𝑗 = 𝜔𝑗 ̅̅̅̅̅̅̅ 𝜓(𝑝)𝐷,𝑗+1 ] + 𝐵𝑗 [𝜓(𝑝)
̅̅̅̅̅̅̅𝐷𝑗 − ̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅𝐷𝑗 − ̅̅̅̅̅̅̅
𝜓(𝑝)𝐷,𝑗−1 ] (3.51)

𝐵𝑗 ̅̅̅̅̅̅̅
𝜓(𝑝)𝐷,𝑗−1 + [𝜎2 Ƙ𝑗 − 𝜔𝑗 𝑧 − 𝐴𝑗 − 𝐵𝑗 ]𝜓(𝑝)
̅̅̅̅̅̅̅𝐷𝑗 + 𝐴𝑗 ̅̅̅̅̅̅̅
𝜓(𝑝)𝐷,𝑗+1 = 0 (3.52)

Eq. (3.52) can also be written in the matrix form:

𝑨. ⃗𝑿
⃗ =𝟎 (3.53)

Where A is a symmetric tridiagonal matrix of the form:

𝑎11 ⋯ 𝑎1𝑛
𝑨=[ ⋮ ⋱ ⋮ ] (3.54)
𝑎𝑛1 ⋯ 𝑎𝑛𝑛

And each element in A is:

𝐵𝑗 , 𝑓𝑜𝑟 𝑘 = 𝑗 − 1 (𝑗 > 1)


𝜎2 Ƙ𝑗 − 𝜔𝑗 𝑧 − 𝐴𝑗 − 𝐵𝑗 𝑓𝑜𝑟 𝑘 = 𝑗
𝑎𝑗𝑘 = (3.55)
𝐴𝑗 , 𝑓𝑜𝑟 𝑘 = 𝑗 + 1 (𝑗 > 𝑛)
[0 𝑓𝑜𝑟 𝑜𝑡ℎ𝑒𝑟 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 ]

While ⃗𝑿
⃗ is a vector of the form:

25
̅̅̅̅̅̅̅
𝜓 (𝑝)𝐷1
̅̅̅̅̅̅̅
𝜓 (𝑝)𝐷2
̅̅̅̅̅̅̅
𝜓(𝑝)𝐷𝑗−1
⃗𝑿
⃗ = . (3.56)
.
.
̅̅̅̅̅̅̅
𝜓(𝑝)𝐷𝑗
̅̅̅̅̅̅̅
[𝜓(𝑝)𝐷𝑗+1 ]

Or with substitution of Eq. (3.47), it can be in the following form:

𝐴1 𝐾0 (𝜎𝑟𝐷 ) + 𝐵1 𝐼0 (𝜎𝑟𝐷 )
𝐴2 𝐾0 (𝜎𝑟𝐷 ) + 𝐵2 𝐼0 (𝜎𝑟𝐷 )
𝐴3 𝐾0 (𝜎𝑟𝐷 ) + 𝐵3 𝐼0 (𝜎𝑟𝐷 )
⃗⃗ = .
𝑿 (3.57)
.
.
𝐴𝑗−1 𝐾0 (𝜎𝑟𝐷 ) + 𝐵𝑗−1𝐼0 (𝜎𝑟𝐷 )
[ 𝐴𝑗 𝐾0 (𝜎𝑟𝐷 ) + 𝐵𝑗 𝐼0 (𝜎𝑟𝐷 ) ]

Eq. (3.52) constitutes a homogenous system of equations that has a nontrivial solution if and only
if the coefficient matrix is singular or noninvertible. The coefficient matrix is a tridiagonal matrix
that is closely banded.

The coefficient matrix must have a determinant of zero. That is:

det 𝑨( 𝜎) = 0 (3.58)

Recall from Eq. (3.55) that the elements of A can be represented as:

26
𝐵𝑗 , 𝑓𝑜𝑟 𝑘 = 𝑗 − 1 (𝑗 > 1)
𝜎 Ƙ𝑗 − 𝜔𝑗 𝑧 − 𝐴𝑗 − 𝐵𝑗
2
𝑓𝑜𝑟 𝑘 = 𝑗
𝑎𝑗𝑘 =
𝐴𝑗 , 𝑓𝑜𝑟 𝑘 = 𝑗 + 1 (𝑗 > 𝑛)
[0 𝑓𝑜𝑟 𝑜𝑡ℎ𝑒𝑟 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 ]

The coefficient matrix whose elements are expressed as in Eq. (3.55) is real symmetric and positive
definite.

The determinant of this sort of matrix can be found by using the general recursive formula:

𝛾𝑗 = 𝑏𝑗𝑗 𝛾𝑗−1 − 𝑏𝑗,𝑗−1 𝑏𝑗−1,𝑗 𝛾𝑗−2 (3.59)

For j=2…n

Where 𝛾𝑗 is the determinant of a tridiagonal matrix of order n, with 𝛾0 = 1 and 𝛾1 = 𝑏1,1

For a given value of n, solving 𝛾𝑛 = 0 will yield the needed n values of 𝜎2 thereby making Eq.
(3.55) singular.

The system above resembles the classic characteristic value problem (also known as the classic
eigenvalue problem). The following real symmetric and positive definite tridiagonal matrix [𝑏𝑗𝑘 ′]
would have 𝜎2 Ƙ𝑗 as the eigenvalues:

−𝐵𝑗 , 𝑓𝑜𝑟 𝑘 = 𝑗 − 1 (𝑗 > 1)


𝜔 𝑧 − 𝐴𝑗 − 𝐵𝑗 𝑓𝑜𝑟 𝑘 = 𝑗
𝑎𝑗𝑘 ′ = 𝑗 (3.60)
−𝐴𝑗 , 𝑓𝑜𝑟 𝑘 = 𝑗 + 1 (𝑗 < 𝑛)
[0 𝑓𝑜𝑟 𝑘 ≠ 𝑗 + 1, 𝑗 𝑜𝑟 𝑗 − 1 ]

If the polynomial in Eq. (3.59) is interpreted as the characteristic equation for eq. (3.60) and
assuming the quantity 𝜎2 Ƙ𝑗 to be real, the n roots 𝜎2 are real and once the n values for have been

27
determined, Eq. (3.48) is used to find the coefficients for the general solution for each layer, which
can be obtained as:

̅̅̅̅̅̅̅𝐷𝑗 = ∑𝑛 [𝐴𝑘 𝐾0 (𝜎𝑘 𝑟𝐷 ) + 𝐵𝑘 𝐼0 (𝜎𝑘 𝑟𝐷 ) ]


𝜓(𝑝) (3.61)
𝑘=1 𝑗 𝑗

Where the subscript on A indicates the layer and the superscript indicates 𝜎. Eq. (3.61) requires
𝟐𝒏𝟐 coefficients to be determined for the solution to be completely specified. But the inner and
outer boundary conditions provide only 𝟐𝒏 equations. For the system of equations of the form
given by Eq. (3.52), Ehlig-Economides and Joseph (1987) showed that the functional relationship
between coefficients can be take advantage of.

𝐴𝑗𝑘 can be expressed in terms of 𝐴1 as follows:

𝑎11 𝑘
𝐴𝑘2 = − 𝐴 =∝𝑘2 𝐴1𝑘
𝑎12 1

1
𝐴𝑘3 = − (𝑎 𝐴𝑘 + 𝑎22 𝐴𝑘2 ) =∝𝑘3 𝐴1𝑘
𝑎21 21 1 (3.62)

1
𝐴𝑘𝑛 = − (𝑎𝑛,𝑛−1 𝐴𝑘𝑛−1 ) =∝𝑘𝑛 𝐴1𝑘
𝑎𝑛,𝑛

Similarly,

𝐵𝑗𝑘 =∝𝑘𝑗 𝐵1𝑘 , for each j. (3.63)

The constants 𝐴𝑗𝑘 and 𝐵𝑗𝑘 have been split into two parts as follows:

𝐴𝑗𝑘 =∝𝑘𝑗 𝐴1𝑘 𝐵𝑗𝑘 =∝𝑘𝑗 𝐵1𝑘 (3.64)

For any pair of layers such as j and j+1, the inner boundary equations are:

𝑑̅̅̅̅̅̅
𝜓(𝑝)𝐷𝑗 𝑑 ̅̅̅̅̅̅
𝜓(𝑝)𝐷𝑗+1
̅̅̅̅̅̅
𝜓 (𝑝)𝐷𝑗 − 𝑠𝑗 ̅̅̅̅̅̅
|𝑟𝐷 =1 = 𝜓 (𝑝)𝐷𝑗+1 − 𝑠𝑗+1 |𝑟𝐷=1 (3.65)
𝑑 𝑟𝐷 𝑑 𝑟𝐷

28
From the outer boundary condition, a relationship can be written between 𝐴1𝑘 and 𝐵1𝑘 .

𝐵1𝑘 .= 𝑏 𝑘 𝐴1𝑘 (3.66)

Where,

For the infinite outer boundary condition.

𝑏𝑘 = 0 (3.67)

For the no-flow outer boundary

𝐾1 (𝜎𝑘 𝑟𝐷 )
𝑏𝑘 = (3.68)
𝐼1 (𝜎𝑘 𝑟𝐷 )

For the constant pressure outer boundary

𝐾0 (𝜎𝑘 𝑟𝐷 )
𝑏𝑘 = − (3.69)
𝐼0 (𝜎𝑘 𝑟𝐷 )

Note that for the outer boundary conditions, 𝑟𝐷 = 𝑟𝑒𝐷

After solving for each eigenvalue, only N unknown coefficients, 𝐴1𝑘 for k = 1,…,n, are to be
determined using N inner boundary conditions. The equation for the dimensionless pressure
becomes:

̅̅̅̅̅̅̅𝐷𝑗 = ∑𝑛 ∝𝑘𝑗 𝐴1𝑘 [𝐾0 (𝜎𝑘 𝑟𝐷 ) + 𝑏𝑘 𝐼0 (𝜎𝑘 𝑟𝐷 ) ]


𝜓(𝑝) (3.70)
𝑘=1

Once, the N coefficients of 𝐴1𝑘 have been solved for, 𝐴𝑗𝑘 can be determined using the functional
dependence relationship of Eq. (3.52) which in combination with the eigenvalues already
determined would yield expressions for pressure and the individual layer sandface production rate:

̅̅̅̅̅̅̅𝐷𝑗 = 𝐴𝑗 𝐾0 (𝜎𝑟𝐷 ) + 𝐵𝑗 𝐼0 (𝜎𝑟𝐷 )


𝜓(𝑝) (3.71)

29
̅̅̅̅̅̅̅𝐷𝑗
𝑑𝜓(𝑝)
𝑞̅𝐷𝑗 = -Ƙ𝑗 |𝑟𝐷 =1
𝑑𝑟𝐷

= 𝐴𝑗 Ƙ𝑗 𝜎𝐾1 (𝜎) - 𝐵𝑗 Ƙ𝑗 𝜎𝐼1 (𝜎)

And wellbore pressure is:

𝜕 𝜓(𝑝)𝐷𝑗
𝜓(𝑝)𝐷𝑤 = 𝜓(𝑝)𝐷𝑗 − 𝑠𝑗 𝑗 = 1, … 𝑛
𝜕 𝑟𝐷

3.4 SOLUTION TO THE DIFFUSIVITY EQUATION FOR STRATIFIED GAS


RESERVOIR WITHOUT INTERLAYER CROSSFLOW

The analytical solution to the diffusivity equation for the commingled system is easier to solve
than that with interlayer crossflow. The solution method employed here utilizes the familiar
Laplace transformation. Recall Eq. (3.29) and the inner and outer boundary conditions ( Eqns. 3.34
to 3.40),

2
𝜕 𝜓(𝑝)𝐷𝑗 1 𝜕𝜓(𝑝)𝐷𝑗 𝜕𝜓(𝑝)𝐷𝑗
Ƙ𝑗 [ 2 +𝑟 𝜕𝑟𝐷
] = 𝜔𝑗 𝜕𝑡 + 𝐴𝑗 [𝜓(𝑝)𝐷𝑗 − 𝜓(𝑝)𝐷,𝑗+1 ] + 𝐵𝑗 [𝜓(𝑝)𝐷𝑗 −
𝜕𝑟𝐷 𝐷 𝑝𝑛𝐷

𝜓(𝑝)𝐷,𝑗−1 ] (3.29)

The initial and boundary conditions in dimensionless from are as follows;

Initial condition

𝜓(𝑝)𝐷𝑗 [𝑟𝐷 , 0] = 0 𝑗 = 1, … 𝑛 (3.34)

Inner boundary condition,

𝜕 𝜓(𝑝)𝐷𝑗
𝜓(𝑝)𝐷𝑤 = 𝜓(𝑝)𝐷𝑗 − 𝑠𝑗 𝑗 = 1, … 𝑛 (3.35)
𝜕 𝑟𝐷

𝜕𝜓(𝑝)𝐷𝑤 𝜕𝜓(𝑝)𝐷𝑤
1 = 𝐶𝐷 − ∑𝑛𝑗=1 Ƙ𝑗 𝑗 = 1, … 𝑛
𝜕𝑡𝐷 𝜕𝑟𝐷

𝐶
Where 𝐶𝐷 = 2 (𝜙ℎ
̅̅̅̅)𝑐 (3.36)
2𝜋 𝑟𝑤

30
Sandface layer rate,

𝑞𝑗 𝜕𝜓(𝑝)𝐷𝑗
𝑞𝐷𝑗 = = −Ƙ𝑗 𝑎𝑡 𝑟𝐷 = 1, 𝑓𝑜𝑟 𝑗 = 1, … 𝑛 (3.37)
𝑞𝑡 𝜕𝑟𝐷

Outer boundary conditions:

Infinite-acting Outer Boundary Condition;

𝜓(𝑝)𝐷𝑗 = 0 𝑟𝐷 → ∞ 𝑗 = 1, . . 𝑛 (3.38)

Constant pressure outer boundary condition;

𝜓(𝑝)𝐷𝑗 = 0 𝑟𝐷 = 𝑟𝑒𝐷 𝑗 = 1, . . 𝑛 (3.39)

Closed boundary condition;

𝜕𝜓(𝑝)𝐷𝑗
=0 𝑟𝐷 = 𝑟𝑒𝐷 𝑗 = 1, . . 𝑛 (3.40)
𝜕𝑟𝐷

Setting 𝐴𝑗 and 𝐵𝑗 to zero since interlayer crossflow is absent, the equation for the commingled
system can be obtained as follows:

2
𝜕 𝜓(𝑝)𝐷𝑗 1 𝜕𝜓(𝑝)𝐷𝑗 𝜕𝜓(𝑝)𝐷𝑗
Ƙ𝑗 [ 2 +𝑟 𝜕𝑟𝐷
] = 𝜔𝑗 𝜕𝑡 (3.72)
𝜕𝑟𝐷 𝐷 𝑝𝑛𝐷

Transforming the dimensionless equation and inner and outer boundary conditions into Laplace
space, these become:

𝑑2 ̅̅̅̅̅̅̅
𝜓(𝑝)𝐷𝑗 ̅̅̅̅̅̅̅
1 𝑑𝜓(𝑝)
] = 𝜔𝑗 ̅̅̅̅̅̅̅
𝐷𝑗
Ƙ𝑗 [
𝑑𝑟𝐷 2 +
𝑟𝐷 𝑑𝑟𝐷
𝜓(𝑝)𝐷𝑗 𝑧 (3.73)

Where 𝑧 is the Laplace space variable.

Laplace transform of initial conditions;

𝑑 𝜓(𝑝)𝐷𝑗 ̅̅̅̅̅̅
̅̅̅̅̅̅
𝜓 ̅̅̅̅̅̅
(𝑝)𝐷𝑤 = 𝜓 (𝑝)𝐷𝑗 − 𝑠𝑗 (3.42)
𝑑𝑟 𝐷

31
1 ̅̅̅̅̅̅̅𝐷𝑤
𝑑𝜓(𝑝)
̅̅̅̅̅̅̅𝐷𝑤 𝑧 − ∑𝑛 Ƙ𝑗
= 𝐶𝐷 𝜓(𝑝) (3.43)
𝑧 𝑗=1 𝑑𝑟𝐷

Laplace transform of outer boundary conditions;

Infinite-acting Outer Boundary Condition;

̅̅̅̅̅̅
𝜓 (𝑝)𝐷𝑗 = 0 𝑟𝐷 → ∞ 𝑗 = 1, . . 𝑛 (3.44)

Constant pressure outer boundary condition;

̅̅̅̅̅̅
𝜓 (𝑝)𝐷𝑗 = 0 𝑟𝐷 = 𝑟𝑒𝐷 𝑗 = 1, . . 𝑛 (3.45)

Closed boundary condition;

̅̅̅̅̅̅̅𝐷𝑗
𝑑𝜓(𝑝)
=0 𝑟𝐷 = 𝑟𝑒𝐷 (3.46)
𝑑𝑟𝐷

Equation (3.73), the diffusivity equation in Laplace space has the form of the modified Bessel’s
equation with a solution of the form:

̅̅̅̅̅̅̅
𝜓 (𝑝)𝐷𝑗 = 𝐴𝑗 𝐾0 (𝜎𝑟𝐷 ) + 𝐵𝑗 𝐼0 (𝜎𝑟𝐷 ) (3.45)

The coefficient 𝐵𝑗 must be set to zero in order to satisfy the outer boundary condition. Hence, the
solution for the pressure in layer j becomes:

̅̅̅̅̅̅̅𝐷𝑗 = 𝐴𝑗 𝐾0 (𝜎𝑟𝐷 )
𝜓(𝑝) (3.74)

Substitution into Eq. (3.73) yields:

Ƙ𝑗 𝜎2 𝐾0 (𝜎𝑟𝐷 ) = 𝜔𝑗 𝐾0 (𝜎𝑟𝐷 )𝑧 (3.75)

32
Solving for 𝜎:

𝜔𝑗 𝑧
𝜎=√ (3.76)
Ƙ𝑗

The production rate from each layer j becomes:

𝑑̅̅̅̅̅̅̅
𝜓(𝑝)𝐷𝑗
𝑞̅𝐷𝑗 = -Ƙ𝑗 𝑑𝑟𝐷
|𝑟 =1
𝐷

= 𝐴𝑗 Ƙ𝑗 𝜎𝐾1 (𝜎) (3.77)

And wellbore pressure is:

̅̅̅̅̅̅
𝑑𝜓 (𝑝)𝐷𝑗
̅̅̅̅̅̅
𝜓 ̅̅̅̅̅̅
(𝑝)𝐷𝑤 = 𝜓 (𝑝)𝐷𝑗 − 𝑠𝑗 |
𝑑𝑟𝐷 𝑟𝐷=1

̅̅̅̅̅̅
𝜓 (𝑝)𝐷𝑤 = 𝐴𝑗 [𝐾0 (𝜎) + 𝑠𝑗 𝜎𝐾1 (𝜎)] (3.78)

𝐴𝑗 can be determined from Eq. (3.78) above as:

̅̅̅̅̅̅
𝜓(𝑝)𝐷𝑤
𝐴𝑗 = (3.79)
[𝐾0 (𝜎)+ 𝑠𝑗 𝜎𝐾1 (𝜎)]

Substituting Eq. (3.79) above into the other inner boundary condition for the constant total
wellhead production, the following can be obtained:

1 ̅̅̅̅̅̅̅𝐷𝑤 𝑧 − ∑𝑛 𝐴𝑗 Ƙ 𝑠𝑗 𝜎𝐾1 (𝜎)


= 𝐶𝐷 𝜓(𝑝) 𝑗=1 𝑗
(3.80)
𝑧

Ƙ𝑗 𝜎𝐾1 (𝜎)̅̅̅̅̅̅
𝜓(𝑝)𝐷𝑤
= 𝐶𝐷 ̅̅̅̅̅̅̅
𝜓(𝑝)𝐷𝑤 𝑧 − ∑𝑛𝑗=1 (𝜎)+ (𝜎)]
(3.81)
[𝐾0 𝑠𝑗 𝜎𝐾1

33
Hence, the wellbore pressure solution in a layered system without interlayer crossflow is:

̅̅̅̅̅̅ 1
𝜓 (𝑝)𝐷𝑤 = Ƙ𝑗 𝜎𝐾1(𝜎) (3.82)
𝑧[𝐶𝐷 𝑧+ ∑𝑛
𝑗=1 ]
𝐾0 (𝜎)+ 𝑠𝑗 𝜎𝐾1 (𝜎)

While the layer production rate is:

𝑞̅𝐷𝑗 = -Ƙ𝑗 𝐴𝑗 𝜎𝐾1 (𝜎) (3.83)

̅̅̅̅̅̅̅𝐷𝑤
Ƙ𝑗 𝜎𝐾1 (𝜎)𝜓(𝑝)
= (3.84)
[𝐾0 (𝜎)+ 𝑠𝑗 𝜎𝐾1 (𝜎)]

3.5 INCORPORATIN OF NON-DARCY FLOW INTO DIFFUSIVITY EQUATION FOR


STRATIFIED GAS RESERVOIR

At low and modest rates such as in the preceding development and solution was for the laminar
flow scenario, the flow of gases can be expressed with Darcy’s law. However, at high velocity or
rates Darcy’s law fails to model accurately the transport of gases through porous media. Hence,
this section of the work attempts to incorporate high velocity effect in the development and
solution already presented for laminar flow. There is no shortage of literature on the subject as
investigators have done so much work on the high velocity effect. The formulation of Lee et al
(1987) would be employed in this work. Following that approach, the non-normalized pseudo-
dimensionless pressure can be written as:

𝑚(𝑝)𝐷𝑁 = 𝐶1 𝐷(𝜇𝑙 )𝑞𝑠𝑐𝐷 (3.85)

𝐶1 is a correlating parameter that depends on the dimensionless flow rate 𝑞𝑠𝑐𝐷 and turbulence
intensity, 𝑁𝑇 for each flow regime.

Specifically, for laminar flow the product of dimensionless flow rate and turbulence intensity is
greater than 0 but less than 0.1, D is equal to zero and 𝐶1 is equal to 1. That is,

34
0.1 < 𝑁𝑇 𝑞𝑠𝑐𝐷 ≤ 1.0 𝐷 = 0

𝐶1 = 1 , (Laminar) (3.86)

0.1 ≤ 𝑁𝑇 𝑞𝑠𝑐𝐷 ≤ 1.0 (Transitional)

𝑟𝑤
𝐶1 = (1 − ) 𝐹𝜇 (3.87)
𝑟𝑖

0.1 ≤ 𝑁𝑇 𝑞𝑠𝑐𝐷 ≤ 1.0 (Turbulent)

𝑟𝑤
𝐶1 = (1 − ) 𝐹𝜇 (𝑁𝑇 𝑞𝑠𝑐𝐷 )−0.028 (3.88)
𝑟𝑖

The following definitions apply to the above formulation:

𝑇𝑞𝑠𝑐 𝑝𝑠𝑐
𝑞𝑠𝑐𝐷 = 𝛼𝑄 2𝑘 (3.89)
𝑟 𝑇𝑠𝑐 𝜋ℎ𝑚(𝑝𝑖 )

𝑘𝑟 2 𝛽𝛾𝑔 𝑚(𝑝𝑖 )
𝑁𝑇 = 𝛼𝑁 (3.90)
𝑇𝜇𝑖 𝑟𝑤

𝜇𝑖
𝐹𝜇 = (3.91)
𝜇

𝑁𝐹𝑜 = 𝑁𝑇 𝑞𝑠𝑐𝐷 (3.92)

Equation (3.91) and (3.92) define the viscosity ratio and Forchheimer number respectively, while
the turbulence factor is defined as:

𝛾𝑔 𝑟𝑑 𝛽𝑘𝑟
𝐷(𝜇) = 𝛼𝐷 ∫𝑟 𝑑𝑟 (3.93)
ℎ 𝑤 𝜇𝑟 2

35
𝛽, the velocity coefficient can be supplied from correlations such as those by Geertsema (1974)
and Firoozabadi and Katz (1979). While former is suitable for unconsolidated sands with possible
porosity variation, the latter is only for compacted sandstones.

𝛼
𝛽 = 𝜑5.5𝛽1 (Geertsema) (3.94)
√𝑘

𝛼𝛽2
𝛽 = 𝑘 1.2 (Firoozabadi and Katz) (3.95)

3.6 LAYER RATE CONVOLUTION AND INCORPORATION OF WELLBORE


STORAGE

The preceding sections contain solutions for laminar and nondarcy flow determined for a stratified
gas reservoir with the assumption of a constant surface flow rate. This assumption requires that
individual layer rates change with time and pressure due to layer variation and wellbore storage
phenomenon. A general convolution approach based on the Duhamel Superposition integral
(1833) is often utilized which satisfies the inner boundary condition with wellbore storage. But a
real-time convolution algorithm originally due to Earlougher (1974) but developed and extended
for multilayered systems by Stewart (2011) is employed in this work. A real time convolution
algorithm for a stratified gas reservoir system where production is commingled into a common
wellbore where no reservoir communication between layers occurs can be formulated in terms of
real gas pseudopressure. The algorithm computes the rate response of a layer when the measured
pressure history and a model with determined parameter are available. The real time convolution
algorithm can be written as:

1422𝑇
𝑚(𝑝)𝑗 − 𝑚(𝑝)𝑤𝑓 =
(𝑘ℎ)𝑗
( ∑𝑀
𝑖=1 𝑄𝑖𝑗 (𝑝𝐷𝑗 (𝑡 − 𝑇𝑖−1 )𝐷 − 𝑝𝐷𝑗 (𝑡 − 𝑇𝑖 )𝐷 ) +

𝑄𝑗 (𝑡)𝑝𝐷𝑗 (𝑡 − 𝑇𝑀 )𝐷 + 𝐷𝑄𝑗 (𝑡)) (3.96)

36
In the above superposition equation, the first term on the left-hand side is supposed to be
superscripted for its reference to the initial reservoir pressure in each layer which may be the same
or different in all layers. Since the production from all the layers is commingled, the wellbore
pressure (which is the second term on the left-hand side) is common to all layers. Each layer is
described by a 𝑝𝐷 function with no restriction placed on the form of the model describing any
layer apart from the requirement that no interlayer crossflow exists.

𝑄𝑖𝑗 is the rate of layer j during time i . Equation (3.96) assumes the availability of rate history
for layer j up to time, 𝑇𝑀 .

The solution for the layer rates may be written as

𝑎 𝑄𝑗 2 (𝑡) + 𝑏𝑄𝑗 (𝑡) + 𝑐 = 0 (3.97)

Where 𝑎 = 𝐷𝑗

𝑏 = 𝑝𝐷𝑗 (𝑡 − 𝑇𝑀 )𝐷 + 𝑆𝑗

𝑀
(𝑚(𝑝)𝑗 − 𝑚(𝑝)𝑤𝑓 )(𝑘ℎ)𝑗
𝑐 = (∑ 𝑄𝑖𝑗 (𝑝𝐷𝑗 (𝑡 − 𝑇𝑖−1 )𝐷 − 𝑝𝐷𝑗 (𝑡 − 𝑇𝑖 )𝐷 ) − )
1422𝑇
𝑖=1

The rate at a new time level can be found in an analytic fashion by using the quadratic formula.

If the layer total flow rate is specified, and the layer rate and the pressure responses have to be
determined, simultaneous equations can be set up and the resulting formulation solved using any
Gaussian technique.

The above convolution algorithm for stratified systems can be presented for the normalized
pseudovariables (pressure and time) formulation using the definitions for normalized
pseudopressure and pseudotime under the model development section.

Since it is easy to measure the wellbore pressures accurately, the layer flow rate response for a
forced pressure transient and an identified model can form an integral part of the testing method.

37
3.7 SOLUTION METHODOLOGY FOR LAMINAR AND NON-DARCY FLOW IN
STRATIFED SYSTEMS WITH AND WITHOUT CROSSFLOW

The following steps can be followed in computing the solutions for the laminar and nondarcy flow
equations:

 Eigenvalues are calculated using an appropriate subroutine from the IMSL Math
Library(Version 7.0)
 The boundary conditions are used to set up a system of equations to be solved used an
appropriate Gaussian elimination routine from the IMSL Math Library(Version 7.0) for the

constants 𝐴𝑗𝑘 and 𝐵𝑗𝑘

 Dimensionless pressure in Laplace space is calculated and numerically inverted into real
space by means of Stehfest Algorithm (1970)
 Real variables (pressure and time) are obtained from the definitions of normalized
pseudovariables. This involves the use of correlations for the compressibility factor and
viscosity. The Hall and Yarborough (1973) correlation is employed for the compressibility
factor while the Carr-Kobayashi-Burrows (1954) correlation.
 For any normalized dimensionless pseudotime, the corresponding laminar solution is used
as the initial guess for the iteration process to evaluate the nondarcy flow responses
 The real variables are computed for this initial guess and the radius of investigation is
determined.
 The value of 𝛽 is computed which requires the equivalent permeability and porosity along
the radial direction. The turbulence intensity, 𝑁𝑇 is also determined.
 The constant dimensionless flow rate is determined, the product of which and the
turbulence intensity gives the Forchheimer number. From the value of the Forchheimer
number, the flow regime can be determined from which the viscosity ratio is evaluated
using the present wellbore and corresponding fluid properties.
 The normalized pseudopressure drop due to inertial and turbulence effect can be calculated
and the total value for the quantity computed.

38
 After the wellbore pressure response is obtained, the normalized pseudopressure
derivatives can be determined by means of a simple two point computation algorithm that
uses one point before and one point after the point of interest, takes the difference of these
two values weighted by the interval.

39
CHAPTER FOUR

RESULTS AND DISCUSSIONS

4.1 MODEL VALIDATION

A semi-analytical model for stratified gas reservoirs with and without interlayer crossflow has

been developed. The model is validated by generating homogeneous reservoir responses using the

model and comparing them with pressure transient responses for corresponding familiar

homogeneous cases. The homogeneous reservoir condition for this model is simulated by

assigning the same porosity, thickness and permeability values for each of the layers. The

following comparisons have been made in this work:

4.1.1 This model and Al-Hussainy et al (1966)

Al-Hussainy et al presented correlated solutions for flow of ideal gases in homogeneous

reservoirs with liquid flow solutions by computing the viscosity term in the dimensionless

time at the initial pressure without considering high velocity effects. Laminar responses of

the present model were compared with the Al-Hussainy et al (1966) solutions as shown in

Figure 4.1. The comparison indicates that the laminar solutions are not functions of the

flow rate since the change in the flow rate does not effect a change in the dimensionless

laminar pressure responses. The laminar responses are shown to be equal to the liquid

solutions.

40
4.1.2 This model and Wattenbarger & Ramey (1968)

Assuming homogeneous reservoir conditions, Wattenbarger & Ramey (1968) used a

numerical model to obtain solutions for real gas flow problems including formation damage,

wellbore storage and high velocity effects. Figure 4.2 shows the influence of inertial effects

on the responses of infinite-acting homogeneous reservoirs. A match of this study with that

of Wattenbarger & Ramey (1968) is seen for different flow rates. The pressure drop is seen

to increase with increasing flow rate but normalized dimensionless pseudopressure drop is

unaffected by flow rate.

4.1.3 This model and Oren et al (1988)

Oren et al (1988) utilized a model originally developed by Lee et al (1987) where the

turbulence factor was modified using a correlating parameter, 𝐶1 . Oren et al (1988)

incorporated wellbore storage effect in their model and used an exponential decay factor

multiplied by the solution obtained by Lee et al (1987). Figure 4.3 shows the responses of

the present model compared with those of Oren et al (1988).

4.2 EFFECTS OF DIFFERENT PARAMETERS ON THE PRESSURE TRANSIENT

RESPONSES

The effect of various parameters on the pressure transient responses for homogeneous reservoir

situations are discussed as follows:

41
4.2.1 Wellbore Storage and Skin

For early times, the plot of normalized pseudopressure and normalized pseudotime

gives a unit slope, which suggests that the gas produced at first comes from the

well. During wellbore storage dominated flow, the period depends on gas

compressibility and the storage volume of the wellbore.

Skin effect is due to the altered permeability in the vicinity of the wellbore.

A high pressure drop is expected for high skin values, with high skin values

implying permeability damage.

Figures similar to Agarwal type-curves can be generated for the model in this work.

An initial pressure of 3000 psia, dimensionless flow rate of 0.0025, reservoir

permeability of 250md and Geertsma (1974) correlation as the velocity coefficient

have been utilized in the plot.

It was observed that as the value of skin increases, higher pressure drops are seen.

Also, higher early time pressure drop decreases with increase in the wellbore

storage coefficient.

4.2.2 Velocity or Turbulence Coefficient

Correlations of the velocity coefficient with the permeability and porosity of a reservoir

exist. Other works have shown the dependence of the velocity coefficient on saturation.

The Geertsma (1974) and Firoozabadi and Katz (1979) correlation are used to investigate

the effect of the velocity coefficient on the dimensionless normalized pseudopressure and

derivative responses for a homogeneous infinite-acting system with various skin values (0,

2, 5), an initial reservoir pressure of 3000 psia, wellbore storage coefficient of 1000,

42
reservoir porosity of 0.25 and permeability of 250md, and a constant dimensionless flow

rate of 0.01. The Geertsma (1974) correlation gives a lower pressure drop than that of

Firoozabadi and Katz (1979). Hence, the constant velocity coefficient approach does not

seem to represent the actual condition.

4.2.3 Flow Rate or Velocity

Unlike oil reservoirs, the velocity or flow rate has a significant effect on the pressure transient

responses because of the inertial or turbulence effect. A number of responses have been

generated with different dimensionless flow rates for a homogenous reservoir, where the

dimensionless flow rate is proportional to the actual flow rate for the same reservoir condition.

4.2.4 Initial Pressure

The initial pressure does not seem to affect the pressure responses even though the

definition of turbulence intensity might lead one to assume that it does. High velocity

responses depend on the Forchheimer number (product of turbulence intensity and

dimensionless flow rate) which is not dependent on the initial pressure.

4.2.5 Permeability

Laminar responses do not depend on the permeability of the reservoir but high velocity

effects have a strong dependence on the permeability.

4.2.6 Formation Damage

Near wellbore formation damage affect the productivity of the well. Responses of the

well with a certain skin were investigated.

43
4.2.7 Layering

The effect of ordering of the layers is investigated to see whether there is any variation

when the bottom layer capacity, kh, is changed by the same multiples of the top of that

layer. The plot shows that there is no significant change in the responses when the layer

ordering is changed.

𝜓𝐷

𝑡𝑝𝑛𝐷

44
𝜓𝐷

Figure 4.2 - This model and Wattenbarger & Ramey (1968) for 𝑞𝐷 =
0.05(𝑏𝑙𝑢𝑒), 0.01(𝑔𝑟𝑒𝑒𝑛), 0.02(𝑏𝑙𝑎𝑐𝑘), 0.03(𝑟𝑒𝑑)

45
𝜓𝐷

Figure 4.3 -This model and Oren et al (1988) for 𝐶𝐷 =0 (green) and 𝐶𝐷 = 1000 (blue)

46
CHAPTER FIVE

CONCLUSION AND RECOMMENDATIONS

5.1 CONCLUSION

A stratified gas reservoir model that includes wellbore storage, skin, and high velocity effects for
a system with and without interlayer crossflow was developed and sensitivity studies on key
factors affecting the pressure transient behavior were carried out. A number of observations lead
to the following conclusions:

a) Flow rate has the most significant effect on the pressure transient responses of the stratified
gas reservoirs. High normalized dimensionless pseudopressure drops are seen for higher
flow rates, with the logarithmic derivative of normalized dimensionless pseudopressure
attaining values in excess of 0.5 in the transient state for very high flow rates.

b) The initial pressures have no noticeable effect on the pressure transient responses of
stratified gas reservoirs.

c) The normalized dimensionless pseudopressure drop is observed to increase with reservoir


permeability increase. The reason for this is not far-fetched: As permeability increases, an
increase in flow rate occurs for a constant value of dimensionless flow rate resulting in a
more prominent high velocity effect. Whereas, the permeability does not have a great effect
on the responses for a constant flow rate.

d) When the top layer of the system has a higher flow capacity than the bottom layer for a
two-layer commingled system, the normalized dimensionless pseudopressure drops are
higher. The amount of increase in the normalized dimensionless pseudopressure depends
on the flow rate. The layer position does not seem to affect the pressure responses. The
early time responses for a stratified gas reservoir with interlayer crossflow are almost

47
similar to those for the commingled system. At late time when interlayer crossflow has
developed, the pressure transient responses are like that of a homogeneous system.

e) Unlike in pressure transient analysis for oil wells, the concept of a thin skin does not
adequately accommodate formation damage in gas reservoirs due to the presence of high
velocity effect.

f) The effect of the velocity coefficient β, cannot be ignored in the pressure transient
responses of gas reservoirs. Firoozabadi and Katz (1979) correlation for β gives a higher
value of pressure drop than the Geertsma (1974) correlation. Analyses that resort to
arbitrarily fixing β are not representative of true reservoir conditions. Needless to say, no
high velocity phenomena is expected for oil reservoirs.

5.2 RECOMMENDATIONS

The model developed is for a stratified gas reservoir with and without interlayer crossflow.
Horizontal heterogeneity as seen in composite systems can be incorporated without modest
modifications to the model. Only single phase flow of gas is studied in this work. Hence,
the model can be extended to multiphase flow for any reservoir for the purpose of pressure
transient analysis.

The assumption of a single vertical well is utilized in this thesis. The pressure transient
analysis of slanted wells in stratified gas reservoirs with and without interlayer crossflow
is bound to be a fascinating subject of further studies.
Various reservoir conditions such as water drive, water coning, etc can be incorporated
when an extension of this work is desired.

Current well testing software have not yet fully integrated stratified reservoirs into the
models used in their well testing suites. Good programming skills, mathematical

48
competency and sound reservoir engineering could guide in selecting a student group that
might undertake the development of a multilayer well testing software.

49
REFERENCES

Abramowitz, M., and Stegun, I.A. (1970): “Handbook of Mathematical Functions”,

Dover Publications, Inc., New York

Agarwal, R.G., Al-Hussainy, R. and Ramey, H.J., (1970): “ An Investigation of Wellbore

Storage and Skin Effect in Unsteady Liquid Flow: I. Analytical Treatment,” SPEJ

(September) 279-290

Al-Hussainy, R., Ramey, H.J., and Crawford, P.B. (1966): “The Flow Real Gases

Through Porous Media,” JPT AIME 237 (May) 624-636

Atkinson, K.E. (1978): “An Introduction to Numerical Analysis,” John Wiley and Sons,

New York

Bard, Y. (1974): “Nonlinear Parameter Estimation,” Academic Press, Inc. Ltd., New

York.

Bennett, C. O., Reynolds, A. C., and Raghavan, R. (1986): “Analysis of Finite

Conductivity Fractures Intercepting Multilayer Reservoir,” SPE FE, 259-274.

Bourdet, D., (1985): “Pressure Behavior of Layered Reservoirs with Crossflow,” SPE

paper NO. 13628, presented at the 55th Annual California Regional Meeting in

Bakersfield, CA.,

Bremer, R. E., Winston, H., and Vela, S. (1985): “Analytical Model for Vertical

Interference Tests Across Low-Permeability Zones,” Soc. Pet. Eng. J. 407-418.

Camacho-V., R. G., Raghavan, R., Reynolds, A. C. , (1984): “Response of Wells

Producing Layered Reservoirs: Unequal Fracture Length,” SPE paper No. 12844

presented at the SPE/DOE/GRI Unconventional Gas Recovery Symposium, Pittsburgh.

50
Carr, N.L, Kobayashi, R. and Burrows, D,B. (1954): “Viscosity of Hydrocarbon Gases

Undcr Pressure,” Trans AIME 201 (1954) 264-272.

Chen, T. (1987): “Pressure Drawdown in a Layered Reservoir with Linear Boundaries,”

SPE paper No. 16767, presented at the 62nd Annual Technical Conference and

Exhibition in Dallas, TX.,

Chu, W. C., and Raghavan, R. (1981): “The Effect of Non-communicating Layers on

Interference Test Data,” J. Pet. Tech., 370-382.

Churchill, R.V. (1972): Operational Mathematics, 3rd Edition, McGraw-Bill, New York.

Civan, F. and Evans, R.D, (1991): “Non-Darcy Flow CocJ1icients and Relative

Permeabilities for Gas/Brine Systems,” paper SPE 21516 presented at SPE Gas

Technology Symposium, Houston, IX.

Cobb, W. M., Ramey, H. J. Jr., and Miller, F. G. (1972): “Well Test Analysis for Wells

Producing Commingled Zones,” J. Pet. Tech. 27-37.

Deans, H. A., and Gao, C-T., (1983): “Pressure Transients and Crossflow in a Multilayer

Reservoir: Single-Phase Flow,” SPE paper No. 11966, presented at the 58th Annual

Conference and Exhibition in San Francisco, CA.

Earlougher, R. C., Kersch, K. M., and Kunzman, W. J. (1974): “Some Characteristics of

Pressure Buildup Behavior in Bounded Multiple-Layered Reservoirs without Crossflow,”

J. Pet. Tech Oct., 1178-1186.

Ehlig-Economides, C. A., and Joseph, J. A. (1985).: “A New Test for Determination of

Individual Layer Properties in a Multilayered Reservoir,” SPE paper No. 14167,

presented at 60th SPE Annual Technical Conference and Exhibition, Las Vegas, NV.

51
Forchheimer, P,H. (1901): “Wasscrbewegung durch Boden,” Z. Ver Deutsch Ing;.45

1781-1788.

Gao, C-T. (1984): “Single-Phase Fluid Flow in a Stratified Porous Medium with

Crossflow,” Soc. Pet. Eng. J., 97-106.

Gao, C-T. (1983b): “The Crossflow Behavior and the Determination of Reservoir

Parameters by Drawdown Tests in Multilayer Reservoirs,” SPE paper No. 12580,

Geertsma, J. (1974): “Estimating the Coefficient of Inertial Resistance in Fluid Flow

through Porous Media,” SPFJ (October) 445-450.

Householder, A. S. (1953).: “Principles of Numerical Analysis,” McGraw-Hill Book Co.

Inc., New York .

Katz, M. L., and Tek, M. R. (1962): “A Theoretical Study of Pressure Distribution and

Fluid Flux in Bounded Stratified Porous Systems with Crossflow,” Soc. Pet. Eng. J.,

Mar., 68-82.

Kazemi, H. (1970): “Pressure Buildup in Reservoir Limit Testing of Stratified Systems.”

J. Pet. Tech., Apr., 503-511.

Kucuk, F., Karakas, M., and Ayestaran, L. (1984).: “Well Test Analysis of Commingled

Zones Without Crossflow,” SPE paper No. 13081 presented at the 59th Annual Technical

Conference and Exhibition in Houston, TX.

Kucuk, F., and Ayestaran, L. (1985): “Analysis of Simultaneously Measured Pressure

and Sandface Flow Rate in Transient Well Testing,” J. Pet. Tech. 323-334.

Kucuk, F., Karakas, M., and Ayestaran, L. (1986): “Well Testing and Analysis

Techniques for Layered Reservoirs,” SPE FE, 342-354.

52
Larsen, L. (1981): “Wells Producing Commingled Zones with Unequal Initial Pressures

and Reservoir Properties,” SPE paper No. 10325, presented at the 56th Annual Technical

Conference and Exhibition in San Antonio, TX.

Larsen, L. (1982): “Determination of Skin Factors and Flow Capacities of Individual

Layers in Two-Layered Reservoirs,” SPE paper No. 11138, presented at the 57th Annual

Technical Conference and Exhibition in New Orleans, LA.

Lee, R.L., Logan, R. W. and Tek, M.R, (1987): “Effect of Turbulence on Transient Flow

of Real Gas through Porous Media,” SPEFE (March) 108-120.

Lefkovits, H. C., Hazebroek, P., Allen, E. E., and Matthew, C. S. (1961): “A Study of the

Behavior of Bounded Reservoirs Composed of Stratified Layers,” Soc. Pet. Eng. J., 43-

58.

Meunier, D., Wittmann, M. J., and Stewart, G. (1961): “Interpretation of Pressure

Buildup Test Using In-Situ Measurement of Afterow," J. Pet. Tech., 143-152.

Meunier, D.F., Kabir, C.S., and Wittmann, M.J (1985): “Gas Well Test Analysis: Use of

Normalized Pressure and Time Functions,” paper SPE 13082 presented at the 1984 SPE

Annual Technical Conference and Exhibition, Houston, Sept. 16-19.

Oren, P.E., Lee, R.L. and Tek, M.R. (1988): “The Effects of Well bore Storage, Skin and

Turbulence Intensity on Early-Time Transient Flow of Real Gas Through Porous Media,”

SPEFE. 547-554.

Odili, O.N. (2015): “Well Test Analysis in Stratified Gas Reservoirs with and without

Crosssflow”, MSc Project, Department of Petroleum Engineering, University of Ibadan.

Raghavan, R. (1993): Well Test Analysis, P T R Prentice Hall, Englewood Cliffs. NJ,

218-267.

53
Smith, R,Y. (1961): “Unsteady-State Gas Flow into Gas Wells,” Trans, AIME Vol.

222,1151.1159

Stehfest, H. (1970): “Algorithm 368, Numerical Inversion of Laplace Transforms, D-5,”

Comm, Of ACM. 13, No.1, (Jan) 49.

Wattenbarger, R.A. and Ramey, H.J, (1968): “Gas Well Testing Turbulence, Damage,

and Wellbore Storage," JPT 877-887.

Whitson, C.H. and Sognesand, S, (1990): “Application of the van Everdingen-Meyer

Method for Analyzing Variate-Rate Wel1 Tests,” SPEFE (March) 67-75.

54
APPENDIX

COMPUTER CODES

INCLUDE ‘MATHD.FI’
IMPLICIT REAL*8(A-H, M, O-Z)

PARAMETER (L1=2, L2=L1+1, L3= L1+2, ITRING=0 SCTEMP =2.88.71D00)


PARAMETER (SCPRESS=1.01325D05, F1M = 0.3048D00, DAYS = 8.64D04)
PARAMETER (MDMTS = 9.86923D-16, MMCFTMTC = 2.831685D04)
PARAMETER (PSIPA = 6.894757D03, CPPAS = 1.0D-03, PSCPR = (PSIPA**2)/CPPAS
PARAMETER (NC = 9, NCPOL = NC*22, NCPOL1 =NCPOL +1)
PARAMETER (ICYC=22)
DIMENSION RK (L2, L1), RKV (L2, L1), PHI (L2, L1), H (L1), HW (L1), TD (22)
DIMENSION TIMR MR (NCPOL), DMPSPR (NCPOL), DMDPSPR (NCPOL), DMPST1M (NCPOL)
DIMENSION PDMAT (NCPOL), TRD1 (NCPOL}, PSPRSMAT (NCPOL,)
DIMENSION EXTEMP (1000), PRMAT (NCPOL}, DPSPRF (NCPOL,)
DIMENSION PSPRESMAT (NCPOL), TRD1 (NCPOL}, PSPRSMAT (NCPOL,)
DIMENSION DDMDT (NCPOL, NCPOL), DMTM (L1, NCPOL)
DIMENSION FINDD (L1), FINDC (L1), FRAXQ (L1, NCPOL,)
DIMENSION WORA (L2, L1), WORB (L2), WORX (L2)
DIMENSION DMPRJMT (L1, NCPOL)

COMMON/GM/CD, INOBC, NL, NZ, NRD, INFP, JWELL, SK (L1), NPL(L1), RD(L3), + ST(L2,L1),
TS(L2,L1), RM(L1, L1), XA(L2,L1), XD(L2,L1)
COMMON/CINY/TEMPR, PRESSINI
COMMON/CINT1/Z1,Z2,N,ERT,TPC1,TPC2,WM,SG,YCO2,YH2H,YN2,BASE

55
COMMON/CINT4/MINI, QST, TMPR1, PI, TTS, QD, HT
COMMOM/FRAC/X (12)
COMMON/CINT5/JBETA, RW
COMMON/PRPS/PSEUDOPR (IPRLEN), REALPR (IPRLEN)
COMMON/CIN6/PERM (L2, L1), RDR (L3), PHIR (L2, L1)

REAL*8 PSTIMMATA [ALLOCATABLE] (.), TPRMAT [ALLOCATABLE] (:)

LOGICAL CHECK
EXTERNAL DQDVAL
CHECK = .TRUE.
OPEN (UNIT=13, FILE=’SKN25.IN’, STATUS = ‘OLD’)
OPEN (UNIT=23, FILE=’PERM2.IN’, STATUS = ‘OLD’)
OPEN (UNIT=33, FILE=’PERMK2.IN’, STATUS = ‘OLD’)
OPEN (UNIT=43, FILE=’POR2.IN’, STATUS = ‘OLD’)
OPEN (UNIT=53, FILE=’H2.IN’, STATUS = ‘OLD’)
OPEN (UNIT=63, FILE=’RD2.IN’, STATUS = ‘OLD’)
OPEN (UNIT=333, FILE=’BAKJ.DAT’, STATUS = ‘OLD’)

CALL UMACH (-2, 3)


OPEN (UNIT=1, FILE = ‘FCDA11.OUT’)
OPEN (UNIT=2, FILE = ‘FCDB11.OUT’)
OPEN (UNIT=3, FILE = ‘FCDC11.OUT’)
OPEN (UNIT=4, FILE = ‘FCDD11.OUT’)
Z1=0.3 D00
Z2=1.0D00
BASE = 14.7D00

56
N =10000
ERT = 1.0D-07
READ (333,*) (X(I) , I =1,12)

WRITE (*,*) ‘ENTER THE CODE FOR THE GAS PROPERTIES EVALUATION’
WRITE (*,*) ‘1 –COMPOSITIONAL ANALYSIS’
WRITE (*,*) ‘GIVEN SPECIFIC GRAVITY’
READ (*,*) IFLAG

IF (IFLAG.EQ.O) THEN
WRITE (*,*) ‘ENTER THE SPECIFIC GRAVITY OF THE GAS’
READ (*,*) SG
WRITE (*,*) ‘ENTER H2S & CO2 MOLE FRACTIONS’
READ (*,*) YH2S, YCO2
WM = 28.964D00*SG
EPS1 = EPS (YH2S, YCO2)
TPC1 = TPC (SG, EPS1)
PPC1 = PPC (SG, TPC1, YH2S, EPS1)
ELSE
CALL GASGRA (SG, WM, PPC1, TPC1, EPS1)
PPC2 =PPC1
TPC2 = TPC1
YH2S = X (1)
YCO2 = X (2)
YN2 = X (3)
ENDIF

57
NL = L1
NZ = NL +1
NRD = NL +2
PI = DACOS (O.O DOO)*2.ODOO
PRINT*, ‘ENTER THE LAYER NUMBER, JWELL FOR WHICH WELLBORE PRESSURE
WILL BE CALCULATED’
READ (*,*) JWELL
PRINT*, ‘ENTER RESPONSE FUNCTION CODE’
PRINT*, ‘1 –DRAWDOWN’
PRINT*, ‘2 –BUILDUP’
READ (*,*) INRES

IF (INRES.EQ.2) THEN
PRINT*, ‘ENTER THE DIMENSIONLESS PRODUCING TIME, TPD, (BASED ON MIN
FRONT RADIUS)’
READ (*,*) TPD
ENDIF
PRINT*, ‘SELECT CODES FOR OUTER BOUNDRY CONDITIONS, INOBC’
PRINT*, ‘1 ---INFINITE’
PRINT*, ‘2 ---CLOSED’
PRINT*, ‘3 --- CONSTANT PRESSURE’
READ (*,*) INOBC
INOBC = 2
PRINT*, ‘PLEASE ENTER WELLBORE STORAGE COEFFICIENT, CD’
READ (*,*) CD
PRINT *, ‘ENTER RESERVOIR TEMPERATURE (KELVIN)’
READ (*,*) TEMPR
TEMPR = TEMPR *1.8D00

58
PRINT*, ‘ENTER INITIAL RESERVOIR PSEUDOREDUCED PRESSURE’
READ (*,*) PPRI
PRESSINI = PPCI * PPRI
WRITE (3,*) TRESSINI (PSI) ‘ , PRESSINI
READ (13,*) (SK (J) , J =1, N1)
CLOSE (13)
PRINT*, ‘ENTER THE NUMBER OF TERMS TO BE USED IN STEHFEST ALGORITHM’
READ (*,*) NT
PRINT*, ‘ENTER THE DIMENSIONLESS OUTER RADIUS’
READ (*,*) ROUT
READ (23,*) RK
CALL DWRRRN (‘RK [M2]’, NZ, NL, RK, L2, ITRING)
CLOSE (23)
READ (43,*) PHI

59

You might also like