Nodal

Analysis of Oil and Gas

Production Systems
 Nodal Analysis of Oil and Gas
Production Systems

Jan Dirk Jansen

The MATLAB software accompanying the book can

be downloaded from www.tudelft.nl/nodal

Society of Petroleum Engineers

© Copyright 2017 Society of Petroleum Engineers

All rights reserved. No portion of this book may be reproduced in any form or by any
means, including electronic storage and retrieval systems, except by explicit, prior
written permission of the publisher except for brief passages excerpted for review
and critical purposes.

Printed in the United States of America.

Disclaimer
This book was prepared by members of the Society of Petroleum Engineers and their
well-qualified colleagues from material published in the recognized technical literature
and from their own individual experience and expertise. While the material presented
is believed to be based on sound technical knowledge, neither the Society of
Petroleum Engineers nor any of the authors or editors herein provide a warranty
either expressed or implied in its application. Correspondingly, the discussion of
materials, methods, or techniques that may be covered by letters patents implies no
freedom to use such materials, methods, or techniques without permission through
appropriate licensing. Nothing described within this book should be construed to
lessen the need to apply sound engineering judgment nor to carefully apply accepted
engineering practices in the design, implementation, or application of the techniques
described herein.

ISBN 978-1-61399-564-8

First Printing 2017

Society of Petroleum Engineers

222 Palisades Creek Drive
Richardson, TX 75080-2040 USA
http://www.spe.org/store
[email protected]
1.972-952.9393
 Preface
This book is based on material used in a three-month course for first-year graduate
students in the Petroleum Engineering MSc program at Delft University of Technology
(TU Delft). This course aims at providing skills in the development and use of
mathematical and computer models for flow through the various parts of an oil and
gas production system. The underlying idea is that a basic understanding of the
numerical implementation of theoretical concepts should be an important element in
the education of all engineering students, even though only a few of them will become
tool developers and most of them will just use software tools whose inner workings
are hidden behind slick user interfaces.
Although the book is primarily intended for classroom use, it may also be of value
for practicing engineers or researchers who want to get acquainted with the
essentials of nodal analysis or learn more about the underlying theory.* Prerequisites
are a basic understanding of physical transport phenomena; calculus, including first-
order differential equations; and preferably an acquaintance with hydrocarbon
properties and oil and gas well completions.
Questions and exercises are provided at the end of each chapter. Some of those
involve computations that can be performed by hand with a calculator, while others
require the use of a computer to run the MATLAB routines that accompany the book.
MATLAB questions and assignments form an important ingredient of the course, and a
basic level in computer programming will therefore be of help to benefit maximally
from the book. Hand calculations are primarily seen as a means to obtain quick order-
of-magnitude estimates or to verify the results of computer simulations. However, the
theory is presented in the book to be understandable without the reader going
through the MATLAB exercises. Worked-out answers to all questions are provided in
Appendix G.
An early version of this book was based on material written by prof. Peter K.
Currie. Since I took over his course in 2001, the text has gradually been rewritten and
expanded. During Spring 2011 I revised major parts while teaching a similar course at
Stanford University. Very little of the theory described in the text is new, and I
acknowledge the contributions of all those who have pioneered the field of nodal
analysis and the underlying models for the various production system elements. I
have tried to refer to the original sources as far as I am aware of them. Moreover, I
acknowledge the contribution of many students and teaching assistants at TU Delft
and Stanford who helped to spot errors, inconsistencies, or material that was unclear,
both in the text and in the accompanying MATLAB code. No doubt there is still room for
further improvement. If you find errors, note missing references, or have other
comments, please let me know at [email protected].
JDJ, Delft, July 2017.
* Those readers who are just looking for a quick introduction to nodal analysis should read Sections 1.5.1 and 1.6,
perform the assignment in Section 1.8 (no MATLAB required), and then jump to Chapter 8, which gives an overview of
the most common nodal analysis applications.
 Contents
Preface
1 Introduction
1.1 What Is Covered in This Chapter?
1.2 About This Book
1.3 Production Engineering
1.4 Oil and Gas Production Systems
1.5 System Models
1.6 Nodal Analysis
1.7 Questions
1.8 MATLAB Assignment: Nodal Analysis of a Simple System
2 Properties of Reservoir Fluids
2.1 What Is Covered in This Chapter?
2.2 Fluid Properties
2.3 Production Variables
2.4 Pressure/Temperature Phase Diagram
2.5 Equations of State
2.6 Oil Models
2.7 Fluid Property Calculations
2.8 Questions
2.9 MATLAB Assignment: Hydrocarbon Properties
3 Single-Phase Flow in Wells and Pipelines
3.1 What Is Covered in This Chapter?
3.2 Governing Equations
3.3 Single-Phase Oil Flow
3.4 Single-Phase Gas Flow
3.5 Tubing Performance and Intake Curves
3.6 Flow in an Annular Geometry
3.7 Questions
3.8 MATLAB Assignment: Single-Phase Gas Flow
4 Multiphase Flow in Wells and Pipelines
4.1 What Is Covered in This Chapter?
4.2 Multiphase Flow Concepts
 4.3 Pressure Drop Analysis
4.4 Tubing Intake Curves
4.5 Lift Tables
4.6 Questions
4.7 MATLAB Assignment: Drift Flux
5 Flow Through Restrictions
5.1 What Is Covered in This Chapter?
5.2 Restrictions
5.3 Single-Phase Oil Flow
5.4 Single-Phase Gas Flow
5.5 Multiphase Flow: Empirical Models
5.6 Multiphase Flow: Theoretical Models
5.7 Questions
5.8 MATLAB Assignment: Choke Flow
6 Inflow Performance: The Basics
6.1 What Is Covered in This Chapter?
6.2 The Importance of Inflow Performance
6.3 Well Operation and Reservoir Flow Stages
6.4 Governing Equations
6.5 Single-Phase Oil Flow
6.6 The Skin Factor
6.7 Single-Phase Gas Flow
6.8 Multiphase Flow: Empirical Models
6.9 Multilayer Inflow Performance
6.10 Related Topics Not Considered in This Chapter
6.11 Questions
6.12 MATLAB Assignment: Commingled Production With a Smart Well
7 Inflow Performance: Further Topics
7.1 What Is Covered in This Chapter?
7.2 Two-Dimensional Reservoir Flow
7.3 Horizontal Wells, Part 1
7.4 Horizontal Wells, Part 2
7.5 Semi-Analytical Approach
7.6 Multiphase Flow: Theoretical Models
7.7 Related Topics Not Considered in This Chapter
 7.8 Questions
7.9 MATLAB Assignment: Horizontal Well Inflow and Pressure Drop …
8 Well Performance
8.1 What Is Covered in This Chapter?
8.2 Analyzing Well Performance
8.3 Field Development Planning and Field Management
8.4 Artificial Lift
8.5 Related Topics Not Considered in This Chapter
8.6 Questions
8.7 MATLAB Assignment: Well Performance
Appendix A: SI Units and Field Units
A-1 Conversion Factors
A-2 SI Prefixes
A-3 Standard Conditions
A-4 Pitfalls in Unit Conversion
Appendix B: Fluid Properties and Correlations
B-1 Fluid Properties
B-2 Oil Correlations
B-3 Gas Correlations
B-4 Water Properties
Appendix C: Wellbore Surveying
C-1 Coordinates
C-2 Survey Evaluation
C-3 Numerical Implementation
Appendix D: Numerical Methods
D-1 Root Finding
D-2 Differential Equations
Appendix E: Multiphase Flow Correlations
E-1 Hagedorn and Brown
E-2 Mukherjee and Brill
E-3 Drift Flux Models
Appendix F: Relative Permeabilities
F-1 Physics
F-2 Corey Expressions
Appendix G: Answers to Questions
G-1 Answers for Chapter 1, Introduction
G-2 Answers for Chapter 2, Properties of Reservoir Fluids
G-3 Answers for Chapter 3, Single-Phase Flow in Wells and Pipelines
G-4 Answers for Chapter 4, Multiphase Flow in Wells and Pipelines
G-5 Answers for Chapter 5, Flow Through Restrictions
G-6 Answers for Chapter 6, Inflow Performance: The Basics
G-7 Answers for Chapter 7, Inflow Performance: Further Topics
G-8 Answers for Chapter 8, Well Performance
Appendix H: MATLAB Files
H-1 Guidance for Use
H-2 Conversion Factors
H-3 Survey
H-4 Fluid Properties
H-5 Pipe Flow
H-6 Lift Tables
H-7 Choke Flow
H-8 Reservoir Flow
H-9 Systems
H-10 Questions
H-11 Graphical User Interface
H-12 Demo
Nomenclature
Glossary
References
Index
Chapter 1

Introduction

1.1 What Is Covered in This Chapter?

This introductory chapter is meant to clarify the scope and objectives of the book,
give a brief overview of the field of production engineering and the relevant hardware,
describe the basic concepts used in oil and gas production system modeling, and
introduce the method of nodal analysis.

1.2 About This Book

1.2.1 Objectives. Oil and gas wells are essential elements of oil and gas production
systems. A production system can be roughly defined as the equipment required to
produce hydrocarbons from a subsurface reservoir to the point of sale. The activities
related to the conceptual design and the operation of a production system are
collectively referred to as production engineering. An important element of production
engineering is the quantification of pressures, temperatures, and flow rates inside the
production system for various operating conditions, an activity that usually requires
some mathematical model of the system, either in the form of written equations or
implemented in a computer program. A particular form of modeling, known as nodal
analysis, describes the system as a collection of separate elements with distinct flow
properties (Brown 1984; Brown and Lea 1985; Beggs 1991) and builds on theory
developed in the classic paper of Gilbert (1954). (Note that literature references are
listed alphabetically in the References section of this book.) Nodal analysis has
proven to be an effective tool for analyzing production system performance, and
forms the basis of several commercial and proprietary software packages.
The objectives of this book are:

• To provide an understanding of the concepts used in computer models for the

nodal analysis of flow through a production system. This includes understanding
the physics, capturing the essential physics in the form of mathematical
expressions, and coding these in a computer model.
• To provide basic skills in using a production system computer model to perform
production engineering calculations. This concerns, in particular, generating
predictions of pressures and flow rates and using them to optimize the design
of the production system.

1.2.2 Scope. This book is not explicitly meant to give a complete overview of the field
of production engineering, but, rather, to treat some essential concepts in enough
detail to achieve the objectives listed above. In particular, it concentrates on the
modeling of the flow in a well system, including the near-well reservoir, but does not
treat surface facilities in any detail, nor complex networks of multiple wells. Also, the
treatment of the physical principles is aimed at explaining the essential concepts but
does not completely cover all aspects involved. For example, the book does treat the
equations for pressure drops in various elements of a well system and demonstrates
how they can be implemented in a simple computer code. However, it does not cover
the temperature drops in those elements, because the equations and the computer
implementation are analogous to those for pressure drops, and the details can be
found in specialized books. Another example concerns the models for multiphase flow
in pipes covered in Chapter 4. There, rather than cover the vast range of available
models and correlations, the book describes just a few typical models, not
necessarily the most accurate ones, to explain the underlying concepts.
Understanding the essential concepts in nodal analysis of a well system should
provide the readers with enough knowledge to use commercial nodal analysis
software for more complex systems, explore the literature in this field, and expand
their knowledge as required. Moreover, the scope of this book does not include a
detailed description of hardware components. Several other books are available for
this purpose, such as Bellarby (2009); various chapters in the SPE Petroleum
Engineering Handbook (Lake 2007) also contain good descriptions of hardware
components.

1.2.3 Empiricism and First-Principles Modeling. Many textbooks on production

engineering present a large number of formulas without their derivation. In the present
book, I tried, wherever feasible, to avoid this approach and instead present the
derivation of formulas starting from first principles. However, many production
engineering calculations are performed with empirical correlations, or at least
semiempirical ones with coefficients “tuned” through experiments; in those instances
the presentation of equations without derivation is, of course, unavoidable. Moreover,
the description of full physical models would sometimes be too involved and outside
the scope of this book—e.g., as in the case of fully “mechanistic” multiphase pipe
flow models. Therefore, in such cases the book will revert to (semi)empirical
approaches.

1.2.4 MATLAB Routines. Many of the equations developed here have been
programmed in MATLAB routines that can be downloaded from the website
accompanying the book. MATLAB is a programming language for numerical computing
with extensive plotting capabilities and a large number of pre-programmed routines
for mathematical operations. For an overview and tutorials, see MATLAB (2017). The
understanding and use of the routines accompanying the book requires basic
programming skills, but no advanced MATLAB knowledge. In particular, the use of
compact vector notation as offered by the MATLAB programming language has mostly
been avoided, thus somewhat sacrificing computational and notational efficiency for
the benefit of readability and educational value. Questions are provided at the end of
the chapters. Some of these can be answered through reasoning, others by
performing calculations by hand or with a simple calculator, others require the use of
MATLAB routines. In line with modern production engineering practice, hand calculations
are seen primarily as a means to obtain quick order-of-magnitude estimates or to
verify the results of computer simulations.

1.2.5 Unit Systems and Notation Conventions. Formulas, data, and example
calculations will be presented primarily in consistent SI units. Occasionally the
corresponding field units are added to allow easy comparison with results from
literature or to give the reader a feel for units still used in oil field practice. In the oil
industry, the expression “SI units” is often loosely used to indicate both “strict”
(consistent) SI units and “allowable” (possibly inconsistent) SI units. The strict units
can be subdivided into the seven base SI units (m, kg, s, A, K, mol, and cd) and
derived SI units such as N, Pa, °C, or J. Allowable SI units typically include d (day)
and a (year), and are defined in the SPE Metric Standard (SPE 1982).
A brief list of conversion factors is given in Appendix A (for a more extensive list,
see SPE 1982). In addition, a number of MATLAB “m-files” for unit conversion can be
downloaded from the website accompanying the book (web address listed on title
page at time of printing). These have a self-explanatory syntax. Thus, e.g., to convert
a value of 1,000 psi into Pa, type:
» from_psi_to_Pa(1000)

which produces the answer

ans = 6894757

SI units will be used directly in the text. Non-SI units will be enclosed in round
brackets when necessary to avoid confusion. To distinguish between temperatures
expressed in °C (or °F) and absolute temperatures expressed in K (or °R), absolute
temperatures will be labeled with a subscript: Tabs . Following the SPE Style Guide
(SPE, 2015), different notations, expressed in either SI or field units, can be used for
the same unit. For example, “day” is indicated with “d” in SI units, while it is indicated
with “D” in field units. Oil volumes in field units are indicated with “bbl”, but in
combined units with “B”, such that oil flow rates in field units are expressed in B/D.
Also, products of units in the denominators of fractional expressions are indicated
with a “ ” in SI units and a “-” in field units. Thus, kg/m3 s should be read as kg/(m3 ×
s) and lbs/ft3-s as lbs/(ft3 × s). Dimensions will be enclosed in square brackets, as,
for example, in “Js is expressed in m2/s Pa (B/D-psi-ft) and has dimensions of [L3 m–1
t].”
Dimensions appear as follows, using the SPE Symbols Standard (Lake 2007, Vol.
7, pp. 103-140):
L length
 m mass
M money
n amount of substance
q electrical charge
t time
T temperature
Following the SPE Style Guide (SPE 2015), variables are always written in italics,
while subscripts are italicized when they represent SPE standard symbols or
acronyms; they are written in Roman font when they represent abbreviations. For
example, pmf and psep are used to indicate ‘manifold pressure’ and ‘separator
pressure’ respectively.

1.3 Production Engineering

Fig. 1.1 displays a high-level overview of activities during oil and gas exploration and
production, known as E&P. This process diagram, often referred to as the petroleum
life cycle model, can of course be refined to display subactivities at deeper levels.
The material in this book is of relevance to the production engineering activities during
the development and production phases of the petroleum life cycle, in particular to the
sub-activities involving field development planning, detailed design of wells and
facilities, and operation of wells and facilities.

1.3.1 Development. Unlike what is suggested in Fig. 1.1, the petroleum life cycle is
not just a sequential process. In particular, during the design phase, a lot of activities
are performed in an iterative fashion. Fig. 1.2, e.g., displays some of the activities
involved in designing a well during a field development project, clearly indicating the
iterative nature of the process. At a higher level, several cycles of reappraisal (e.g.,
based on production performance or new seismic data); redevelopment (e.g., through
either the recompletion of existing wells or in-fill drilling of new ones); and production
may take place during the life of a field. Each of these activities involves aspects of
production engineering.

Fig. 1.1—Petroleum life cycle model.

Fig. 1.2—An example of iterative processes during well design in a field development project. Not shown
are the links to other iterative activities during the development process such as geological modeling or
design of surface facilities.

The key objective during field development is maximizing the economic benefits
within the constraints of the project. This optimization process involves comparing
multiple development concepts, usually in combination with multiple subsurface
models to reflect geological uncertainties. Early cooperation between geophysicists,
geologists, reservoir engineers, production engineers, and well engineers, supported
by the appropriate organizational structure and systems (software), is essential to
achieve the objective.
Traditionally, the concept of production optimization is used in a somewhat
narrower context. For example, the textbooks of Brown (1984) and Beggs (1991)
focus on optimizing the various components in the flow path from the reservoir to the
separator, and elaborate on the detailed analysis of flow in flowlines, chokes, wells,
and the near-well section of the reservoir. This book takes a similar approach. All
optimization activities require the use of some kind of model of the production system.
Traditionally, these models consisted of relatively simple mathematical equations,
accessible to hand analysis, sometimes with the aid of charts or tables. Now, the
models are usually much more complicated and require the use of a computer.
1.3.2 Production. Apart from its seemingly linear character, Fig. 1.1 has another
flaw. It suggests that all phases of the petroleum life cycle are equally important, or
that they take similar amounts of time. In reality, the majority of the life of an oil field
is spent in the production phase (see Fig. 1.3). Just as during the design phase,
various repetitive processes occur during the production phase. This is illustrated in
Fig. 1.4, which represents oil and gas production as a feedback control process,
involving measurement, modeling, and control. Two major feedback cycles occur,
each on its own timescale (see, e.g., Rossi et al. 2000).

• Daily Production Control. On a scale of days to weeks, typical input variables

are wellhead chokesettings, water-injection pressures, or lift-gas rates.
Measured output from the process includes production variables such as
tubing-head pressures and oil, gas, and water rates. Control will often be driven
by short-term optimization objectives—e.g., production targets or utilization
rates of surface facilities. Models of flow through wells and surface facilities
can play an important role in the process of optimizing daily production (Bieker
et al. 2007). A typical short-term optimization problem is maximizing oil
production by distributing a limited amount of lift gas over a number of
producing wells.
• Reservoir Management. On a time scale of months to years, the production
process consists essentially of draining the reservoir. In addition to the
variables that control daily production, input includes production engineering
activities such as water or gas shutoff, recompletion, stimulation, or even
sidetracking or infill drilling.

Fig. 1.3—Petroleum life cycle model emphasizing the importance of the production phase.
Fig. 1.4—Oil and gas production represented as a feedback control process, involving measurement,
modeling, and control.

Measured output includes production histories (i.e., records of wellhead

pressures and well flow rates over time), and the results of well tests (i.e.,
measurements of downhole pressure buildup after a well is shut in) and
production logs (i.e., downhole measurements of pressure, temperature, and
flow rates made with a tool run on a wireline), and reservoir images obtained
from time-lapse seismic data (i.e., the results of seismic measurements
repeated over time to monitor the movement of fluid fronts). Control is usually
focused on maximizing the asset revenues, by maximizing ultimate recovery and
minimizing operating expenditures.

System modeling will often involve extensive reservoir simulation, in addition to

wellbore- and surface-flow modeling. In particular, when redevelopment activities are
initiated at a later stage in the producing life of a field, the reservoir management
process coincides in many aspects with the field development process described
above (Jansen et al. 2008). Sometimes short-term production optimization is
considered to be an activity for production engineers only, whereas reservoir
management would be the domain of the reservoir engineering and production
geology disciplines. Such a distinction is somewhat artificial because these activities
are linked. An important production engineering activity is surveillance, the systematic
collection and analysis of well and facilities performance data. Such production data
are essential not only for optimizing the production system but also managing the
reservoir over the long term. In turn, understanding the long-term objectives of field
development is essential to optimizing the depletion of a reservoir. Because of the
need to integrate the short- and long-term optimization activities, many oil companies
have reorganized their production organizations around assets, rather than around the
traditional disciplines.

1.4 Oil and Gas Production Systems

The main functions of an oil and gas production system are to:

• Provide a conduit for the flow of fluids from the reservoir to the off-take point at
the surface, or from the surface back to the subsurface
• Separate the produced reservoir fluids from each other
• Minimize the production of and the negative effects of byproducts
• Store the produced fluids if they cannot be exported immediately
• Measure the amounts of fluids produced and control the production process
• Provide (part of) the energy required to transport fluids through the system

The basic elements of a production system are shown in Figs. 1.5 through 1.7:

• Near-wellbore region of the reservoir, i.e., a zone of several meters in a radial

direction around the wells throughout at the depth of the reservoir
• Wells, from the reservoir to the wellheads at the surface
• Flowlines, from the wellheads to the surface facilities
• Surface facilities, consisting of gas/liquid separators, pumps, compressors,
and other equipment for treatment and measurement
• Storage tanks (also called stock tanks) and pipelines (also called trunk lines)
up to the off-take point (also called the sales point), which can, e.g., be a valve
at the entrance of a gas transport pipeline or the off-loading point of an oil
terminal that supplies tankers
Fig. 1.5—Schematics of a vertical and a horizontal well. The vertical well (top right) is completed with a
tubing, a packer, and a perforated casing. The horizontal well (bottom left) is completed with a tubing, a
packer, and an uncemented slotted pipe.

Fig. 1.6—Surface facilities. The five wellheads are connected to four production wells and one water
injection well. Oil is exported to a terminal, gas into an export pipeline.
Each element of the system can be subdivided into subelements. In particular, the
flow path through the wellbore may consist of:

• Perforations (“perfs”) in the formation, in the cement around the casing, and in
the casing itself
• Sand control equipment, consisting of densely-packed gravel or metal screens
at the bottom of the well
• Tubing, a pipe running from the bottom of the well to surface
• Surface-controlled subsurface safety valve (SCSSV), to close in the well if
surface control is accidentally lost
• Wellhead (often called Christmas tree or tree), consisting of manually or
remotely controlled valves to shut in the well and allow access with wireline
tools. One of the valves is usually a choke valve or bean, which is a variable-
sized restriction used to control the flow from the well

The downhole equipment in a well is usually referred to as the completion. Some

wells are not completed with a cemented production casing over their entire depth but
instead have an openhole completion, which is an open, uncased hole in the
producing unit, also called a barefoot completion; or they may have a perforated or
slotted pipe. There is always a cemented casing running from the top of the reservoir
(i.e., the seal or the caprock) to the surface to avoid uncontrolled flow or co-mingling
of reservoir fluids. The tubing is usually anchored to the casing just above the
reservoir with the aid of an inflatable rubber packer. As opposed to the casing, which
is cemented in place, the tubing can be replaced if it is worn or corroded, or if the
flow rate of the well can be increased by changing the tubing diameter. Some wells
have a dual completion, which means two tubings are installed, each producing from
a different reservoir at a different depth.

Fig. 1.7—Schematic representation of a gas/liquid separator. The throughput is controlled with the aid of a
pressure controller (PC) and a level controller (LC).
 The surface facilities are usually more complicated than those depicted in Fig. 1.6.
Often, two or more separators are mounted in series to allow a stepwise reduction of
the pressure, rather than a single pressure drop. These steps are performed in order
to maximize the amount of oil produced. During separation of light and heavy
hydrocarbon components, a certain amount of the intermediate components
disappears with the lighter ones. The lower the pressure drop, the lower the amounts
of intermediate components that disappear. A multiple separator configuration also
allows coping with a drop in tubinghead pressure (i.e., the pressure in the tubing at
the wellhead), an effect that often occurs during the life of a well when water
production increases and oil production drops. In that case it is possible to connect
the well to the low-pressure separator directly while those wells that still produce at
high tubinghead pressures remain connected to the high-pressure separator. The
pressure in the stock tank is always atmospheric because crude oil (degassed and
dewatered oil) is transported under atmospheric conditions.
A special role is played by the test separator, which is another, smaller separator
equipped to measure oil, gas, and water flow rates. Individual wells can be rerouted
to the test separator for these measurements. Such a production test, which takes
several hours to obtain accurate data, is typically performed once a month for each
well. This has traditionally been the only way to assess a well’s production.
Increasingly, however, more continuous measurements are being taken with the aid of
multiphase flowmeters directly connected to the flowlines. Some form of continuous
measurement of pressures and temperatures at various parts of the surface
production system is quite common. Downhole measurements with the aid of
permanent downhole gauges (PDGs) are less common, although their application is
steadily increasing. Automatic measurements are usually stored in an electronic
control system, which can allow full control of the surface facilities from a local or
even remote control room. This level of instrumentation and automated process
control is quite common in expensive, high-production operations, typically in an
offshore environment. However, many production facilities, especially those on land,
are relatively simple and are still operated manually.
Gas production often requires specialized treatment facilities to dry the gas and
remove corrosive components such as hydrogen sulfide (H2S) or carbon dioxide
(CO2). Furthermore, various types of pumps and compressors, both centrifugal and
reciprocating, are applied to export oil and gas or to reinject produced water or gas
into the subsurface. Gas compression is also often used to enable gas lift, which is
the injection of gas into the wellbore to reduce the hydrostatic head of the liquid and
thus increase production. This is an example of artificial lift, the process of supplying
external energy to force the wellbore liquid from the reservoir to surface. Artificial lift
is required when the reservoir pressure is too low to make the well flow naturally, a
situation that often occurs at a later stage in the life of the reservoir. The most well-
known methods, apart from gas lift, are pumping with beam pumps (“nodding
donkeys”) or electrical submersible pumps (ESPs). For further information on
surface facilities, see, e.g., Chilingarian et al. (1987) or Arnold and Stewart (1998).
For further information on artificial lift, see, e.g., Bellarby (2009) or Economides et al.
(2013).

1.5 System Models

1.5.1 Topology. Flow through a complicated system, such as a production system,

must be broken down into its component parts for analysis. In this book we will
examine the flow behavior in several of the component parts: inflow into the wells,
flow within the wells, and flow through chokes and flowlines. In performing these
detailed analyses it is essential to realize that we are looking only at separate
components of a larger system. Optimizing the performance of each component will
not normally result in an optimized system. For example, if we improve the well inflow
so much that the tubing is unable to handle the production, we have wasted money.
We can describe the components of a production system as a network of
elements connected at nodes. For example, the flow from the reservoir through the
well, the surface facilities, and the pipeline can be represented as a series of
elements and nodes as shown in Fig. 1.8. This figure represents the simplest form of
a network: a cascade, i.e., a chain of elements in which each node is connected to
not more than two elements. In reality, a production system is not a cascade, and the
associated network has a more complex topology. It may, for example, contain
branches, i.e., three or more elements connected at a node. A next step in
complexity involves loops, which are a chain of elements with a beginning and ending
connected to the same node. Fig. 1.9 displays a production system with several
nested loops formed by multiple wells, one of which is a multilateral well, connecting
two reservoirs to a single production facility. If we look in detail at some of the other
components of the system, we can further refine the system model. Thus, for
example, the manifold may be more complex, the facilities may consist of many
components, or the pipeline may have branches with flow coming in from other fields.

1.5.2 Flow and Effort Variables. If we consider the flow of single-phase fluid through
a network, the interaction between the various elements can usually be described in
terms of two pairs of variables: pressure and flow rate, and temperature and heat
flow, respectively. These are examples of pairs of effort and flow variables, concepts
that play a key role in the branch of engineering known as systems dynamics. Other
familiar pairs of effort and flow variables are the electric potential and current used in
electrical network analysis, force and velocity used in mechanical systems analysis,
and torque and angular velocity used also in mechanics. A common feature of most
pairs of effort and flow variables is that their product represents power, also known
as energy rate or energy per unit time:
Fig. 1.8—Network representation of a simple production system.

Fig. 1.9—More complex production system forming a network with branches and loops.

where P is power, E is energy, t is time and where the other symbols have been
defined in Table 1.1, together with their SI units and physical dimensions. Eq. 1.1 is
valid only for consistent sets of units, such as SI units. For use with field units it will
be necessary to introduce numerical factors, for example to account for differences
between quantities expressed in feet and inches. Furthermore, it should be noted that
the product of temperature and heat flow is not power. (Note that a more consistent
formulation of thermal systems is possible in terms of temperature and entropy flow,
an effort and flow variable pair whose product does represent power.) For an in-
depth treatment of system dynamics, see Karnopp et al. (2000).

1.5.3 Element Equations. In the following, we will consider single-phase flow of a

fluid with density and viscosity that are functions of pressure and temperature. For
the sake of simplicity, we will neglect the occurrence of heat flow and assume that
the temperature distribution within the system is known.
Within an element of a production system, pressure and temperature will generally
be functions of time t and spatial coordinates x, y, and z. Most elements, however,
can be represented as 1D systems with a single spatial coordinate s. Furthermore,
we generally restrict ourselves to the analysis of steady-state flow—i.e., flow
independent of t. If we consider a wellbore element, for instance, we can then
describe the flow behavior with the following variables:

Hydraulics Electricity Translation Rotation Heat Flow

Effort Pressure Potential Force Torque Temperature
Symbol p V F M Tabs
SI units Pa V N N m K
Dimension [L–1 m t–2] [L2 m q–1 t–2] [L m t–2] [L2 m t–2] [T]
Flow Flow rate Current Velocity Angular velocity Heat flow rate
Symbol q I v ω Q
SI units m 3/s A m/s rad/s J/s
Dimension [L3 t–1] [q t–1] [L t–1] [t–1] [L2 m t–3]
Product p × q = dE/dt V × I = dE/dt F × v = dE/dt M × ω = dE/dt Tab s × Q ≠ dE/dt
Table 1.1—Analogies between system variables in different domains.

where round brackets indicate functional relationships, and where along-hole distance
s is the independent variable, while vertical depth from surface z(s), wellbore
inclination α(s), and temperature T(s) are given functions of s.
We need four equations to solve for the four unknowns p, q, ρ, and μ. In Chapter 3
we will discuss the nature of these equations in detail. Here we state only that it is
generally possible to solve the equations over the length of an element and express
the pressure and flow rate at one end of the element in terms of the pressure and
flow rate at the other end with input/output relationships:

where f1 and f2 are functions. They are usually strongly nonlinear such that they
cannot be obtained in closed form but have to be determined numerically, as will be
treated in more detail in Chapter 3. Density and viscosity can be computed anywhere
in the element because they are a function of p and T only. We could have expressed
Eq. 1.3 in terms of mass flow rates ṁin = qinρin and ṁout = qoutρout instead of volume
flow rates qin and qout. In that case we would have found that ṁout = ṁin because we
are considering a steady-state situation, where no mass can accumulate in an
element. The same result could have been reached by expressing qin and qout in terms
of a reference flow rate at a given pressure and temperature. In the oil industry such
a reference flow rate is usually defined at standard conditions, representing “typical”
atmospheric conditions of 15°C and 100 kPa. In that case Eq. 1.3 reduces to

where f3 is another nonlinear function and where the subscript sc indicates standard
conditions. Eq. 1.4 illustrates that single-phase flow through an element can be
completely determined with a single relation between pressure and flow rate. In
theory it is also possible to derive the flow rate from the pressure drop with the aid of
the inverse relation:

although in practice this can hardly ever be done directly and requires an iterative
procedure. Moreover, for flow in oil and gas production systems, the situation is
usually more complex because we encounter multiphase flow—i.e., flow involving a
gas phase, one or two liquid phases (oil and water), and sometimes even solid
phases (e.g., wax, asphaltenes, hydrates, ice). As a result, we cannot use a single
rate q to characterize the flow.
Both oil and the gas phases may contain a large number of hydrocarbon
components in compositions that vary with pressure and temperature. However, in
this book we restrict ourselves by assuming that the oil and gas phases are
composed of just two pseudocomponents that are present in a variable ratio
depending on the local pressure and temperature. Each of the phases will have its
own density and viscosity, while in addition the interfacial tension σ comes into play.
Often the pseudocomponents chosen are the gas and oil that result from surface
separation at standard conditions. We can then express the input/output relationships
either in terms of local oil- and gas-phase flow rates qo and qg, or in terms of the
reference pseudocomponent flow rates qo,sc and qg,sc:

or

As was the case in single-phase flow, at steady-state conditions the difference

between flow-in and flow-out vanishes in the equations expressed in
(pseudocomponent) flow rates at standard conditions, i.e., qo,sc,in = qo,sc,out = qo,sc, and
qg,sc,in = qg,sc,out = qg,sc. Eq. 1.7 shows that, just like for single-phase flow, for two-
phase flow the pressure drop over an element can be reduced to a single expression
in terms of the flow rates. In this case, however, we cannot reconstruct the flow rates
from the pressure drop using the inverse of Eq. 1.7 alone, and we need a second
equation that provides information about the ratio of the flow rates qo,sc and qg,sc.

1.5.4 System Equations. The input/output representations (Eqs. 1.3, 1.4, 1.6, and
1.7) are of course perfectly suited for the analysis of cascade systems with the aid of
a marching algorithm. We can then start from the known values for pressure and flow
rate(s) at one end of the system and work our way through to the other end by using
the input/output relations fi for the elements one after another. A system with
branches but without loops can also be analyzed using this approach. For a system
with loops, however, the situation is more complicated and requires solving a system
of nonlinear equations in an iterative fashion. The same approach can be used for the
analysis of fluid-flow networks with multiple components, and obviously the number of
equations increases with the number of components taken into account. Furthermore,
the analysis could be extended to include the temperature T and heat flow rate Q in
the system. Mass flow and heat flow are strongly coupled through convective heat
transport and viscous dissipation. Therefore, in the most general situation of thermal
compositional network analysis, we end up with a large system of coupled nonlinear
equations that may require considerable computing power. Such an analysis is
outside the scope of this book.

1.6 Nodal Analysis

1.6.1 Principle. The analysis of cascade systems with the aid of a marching
algorithm is known in the oil industry as nodal analysis. Written in capitals, NODAL™
analysis has even been registered as a trademark by a major service company. For
any given cascade network, we can march between the system boundaries—i.e.,
between the ends of the cascade, from beginning to end or vice versa. If we know
the pressure and the flow rate at one of the ends, such a one-pass analysis is
sufficient to obtain the pressures and flow rates at all nodes. However, we often only
know the value of one variable at each end of the cascade. For example, we may
know the reservoir pressure and the separator pressure at the ends of a cascade
representing a single well. In that case we need to guess the flow rate at one of the
ends and repeat the marching algorithm several times, either upward or downward,
to establish the correct flow rate in an iterative fashion. (Note that “correct flow rate”
means the flow rate that gives the correct pressure at the other end.) Instead of
marching all the way from one end to the other, we could just as well perform two
shorter marches, each one starting at an end and finishing in a joint node, also
referred to as the analysis node (see Fig. 1.10).
Furthermore, instead of performing the iteration automatically, we could plot both
pressures (top-down and bottom-up) at that particular node for a large number of
flow rates, and determine the correct flow rate graphically. This is indeed the
approach followed in traditional nodal analysis of production systems, which was
developed in the 1950s and relied on tabulated pressure drop values and graphical
analysis rather than computer methods.
Fig. 1.10—Procedure for nodal analysis.

1.6.2 Classic Procedure. In traditional nodal analysis, popular locations for the
analysis node are those corresponding to any of the following types of pressures:

• Flowing bottomhole pressure, pwf, at the bottom of the tubing (see Fig. 1.11)
• Flowing tubinghead pressure, ptf, just upstream of the wellhead choke
• Flowline pressure, pfl, at the entrance of the flowline just downstream of the
wellhead choke
• Manifold pressure, pmf, at the end of the flowline

Frequently-used abbreviations for (flowing) tubinghead pressure and (flowing)

bottomhole pressure are (F)THP and (F)BHP, in which the adjective “flowing” is used
to distinguish the pressures from the closed-in values (also referred to as the static
values) for THP and BHP. These occur when the well is closed in at the surface, and
are sometimes abbreviated as CTHP and CBHP.
Suppose we want to determine the production rate qo,sc of the oil well depicted in
Fig. 1.11 at given values of the reservoir pressure pR and separator pressure psep.
We choose the analysis node at the bottom of the hole and take a first guess for qo,sc.
Starting from psep, we compute the pressure increase over the manifold, the flowline,
the choke, the top part of the tubing, the safety valve, and finally the bottom part of
the tubing, resulting in the FBHP pwf—the pressure at the analysis node. Repeating
this procedure for different flow rates, we can plot a performance curve representing
the relationship between qo,sc and pwf; as determined by the system elements
downstream of the analysis node. Similarly, starting from the reservoir, we can
perform pressure drop calculations for the near-well reservoir and the perforations,
and determine a second performance curve for the upstream part of our system.
(Note that upstream and downstream performance curves are sometimes referred to
as inflow and outflow curves, respectively. Moreover, the downstream performance
curve, which is the outflow curve, is sometimes referred to as the intake curve.) The
upstream and downstream performance curves can be plotted on the same p-q
graph. In general, there will be two possibilities:

1. The two curves do not intersect. The system cannot be operated under the
assumed conditions (i.e., the given reservoir and manifold pressures).
2. The curves intersect at one or more points. Usually we simply find a single
intersection, which is referred to as the operating point or working point (see
Fig. 1.12). The desired value of the flow rate can be read from the horizontal
axis and the corresponding pressure from the vertical axis.

Fig. 1.11—Example of nodal analysis configuration with analysis node at the well bottom.

Note that we could have chosen any other node as the analysis node, but we would
have found an identical value for qo,sc. However, it is sometimes convenient to choose
a specific location for the analysis node to investigate the performance of a specific
element or combination of elements. Moreover, as will be discussed in later chapters,
the particular flow behavior or modeling assumptions in certain elements may prohibit
marching in either an upstream or downstream direction over those elements which
restricts the freedom to select the analysis node.

1.6.3 Production Optimization. Nodal analysis can be used during the FDP phase to
optimize the design of the production system, and in particular to optimize the
configuration and the dimensions of the various system components. Later, during the
production phase, nodal analysis can be used to analyze production measurements
and to optimize system performance through the design of modifications to the
system components. Nodal analysis is particularly useful for debottlenecking, aimed
at removing or modifying those elements in the production system that form the
largest restrictions to flow. Other forms of production optimization involve more than
just nodal analysis and can be framed as formal mathematical optimization problems.
These require maximizing or minimizing one or more objectives by systematically
changing control variables (also known as manipulated or input variables) under the
presence of constraints. A classic example is gas-lift optimization, which involves
optimizing the distribution of a limited amount of available lift gas over a set of oil
producers. Note, however, that in this book we will not cover production optimization
beyond nodal analysis.

Fig. 1.12—Nodal analysis using performance curves. The intersection between the downstream and
upstream performance curves defines the operating point.

1.7 Questions
Note that answers to all questions can be found in Appendix G.

1.1 What is displayed on the axes of performance curves used in nodal analysis?
1.2 Why can the two performance curves for an analysis node intersect in more
than one point?
1.3 Under steady-state flow conditions in a well, the mass flow rates of oil and
gas (pseudo)components do not change over the depth of the well. However,
the volume flow rates of the oil and gas phase rates will change. Why?
1.4 For each of the following wells, would it be theoretically possible to determine
the gas, oil, and/or water flow rates from two pressure measurements, one at
the wellhead and one at the bottom of the well, through iterative use of a
wellbore flow simulator? Assume that you have full knowledge of the fluid
 properties (e.g., densities, viscosities), the wellbore geometry, and the
temperature profile along the wellbore. Briefly explain your answers.
(a) Water injection well.
(b) Single-phase gas production well.
(c) “Dead oil” production well (i.e., without gas production) with known
water/ oil ratio.
(d) “Live oil” production well (i.e., with gas production) and with known
water/ oil ratio but unknown gas/oil ratio.
1.5 An electric motor operates with 90% efficiency at 300 V and draws a current
of 16 A. The shaft of the motor rotates with 240 revolutions per minute (rpm)
and drives an oil pump by means of a reduction gear with an efficiency of
98%. The pump creates a pressure differential of 160 kPa at a flow rate of 22
× 10–3 m3/s. What is the torque generated by the motor? What is the efficiency
of the pump?

Questions 1.6 through 1.10 involve unit conversions. Consult Appendix A for
conversion factors and further information. In addition, you may want to make use of
the MATLAB “m-files” for unit conversion (see Appendix H for an overview).

1.6 A well produces 12,000 STB/D of oil at a gas/oil ratio (GOR) of 15,00
scf/STB. The oil gravity is 38°API and the gas gravity is 0.82. What are the oil
and gas production rates and densities in SI units?
1.7 A mixture of 1 lbm mol of C1H4 and 0.3 lbm mol of CO2 is kept at a
temperature of 83°R and a pressure of 30 psig. What are the mass, the
temperature, and the pressure of the gas mixture in SI units?
1.8 Calculate the pressure in Pa and in psi in a well open to the atmosphere and
filled with salt water (specific gravity 1.03) at a depth of 2000 m.
1.9 The pressure drop over a choke for an incompressible liquid is given by

∆p = ρv2 / (288gc C2),

where ∆p is the pressure drop expressed in psi, ρ is the liquid density in

lbm/ft3, v is the liquid velocity in ft/s, and C is a dimensionless choke
coefficient. The nature of the dimensional constant gc is discussed in Appendix
A. Convert the expression to SI units.
1.10 Fig. 1.13 depicts a vertical well penetrating a reservoir containing gas, oil,
and water. The acronyms GOC and OWC indicate the gas/oil contact and the
oil/water contact respectively. The aquifer is hydrostatic. The atmospheric
pressure is 100 kPa. The oil has a 30°API gravity, and the water gradient is
0.45 psi/ft. The top perforations are 50 ft below the GOC. The total
perforation height is 100 ft. What is the pressure at the GOC in SI units?
1.8 MATLAB Assignment: Nodal Analysis of a Simple System
1.8.1 Objectives.

• Get a first impression of the use of nodal analysis to determine the pressure
drop and flow rate in a simple production system.
• Become aware of some typical, and sometimes surprising, features of nodal
analysis.
• Get a first impression of the use of MATLAB to perform nodal analysis.

1.8.2 Assignment. Note that if you do not have access to MATLAB, you can run the
executable file nodal.exe. This requires first installing the free-of-charge MATLAB
runtime compiler by running MyAppInstaller_web.exe. Both files are located in
directory “Demo.” (Keep in mind that depending on the speed of your computer, the
installation of the runtime compiler and the startup of the application may require
some patience.)

Task 1.
• Download the MATLAB files listed in Appendix H and store them in a working
directory (possibly with subdirectories). If necessary adjust the path to the
directory (and the subdirectories) using the command “Set Path” under the tab
“File” in the MATLAB workspace.
• Type the command nodal in the MATLAB workspace. This should result in the
appearance of the graphical user interface (GUI) shown in Fig. 1.14.
At the left side of Fig. 1.14 a schematic is displayed of a simple production system
consisting of a near-well reservoir, a vertical tubing, a wellhead choke (usually
represented by two touching triangles in process engineering drawings), and a
horizontal flowline. The sliders at the center can be used to set the pressures at the
boundaries of the system—i.e., the reservoir pressure, pR, and the separator
pressure, psep. The analysis node has been chosen at the top of the tubing; the
corresponding FTHP, ptf, is one of the two desired outcomes of the nodal analysis, the
other one being the oil flow rate at standard conditions, qo,sc. These output quantities
are displayed on the axes of the graph at the right. Note that in this book we use the
convention that the flow in production wells is negative (because we define the flow in
injection wells as positive). The graph also depicts two curves, an upstream (dashed)
and a downstream (solid) pressure drop curve. As explained in Section 1.6, the
intersection between the downstream and upstream curves should define the
operating point, but in this case there is clearly no intersection and therefore no flow
through the system.
Fig. 1.13—Oil reservoir with gas cap and aquifer.

Fig. 1.14—GUI for a simple production system. The upstream (dashed) and downstream (solid) pressure
drop curves do not intersect, indicating that for the present system configuration and boundary
conditions, flow through the system is not possible.

Question 1. Which system elements contribute to the upstream and downstream

pressure drops respectively?
Task 2.
• Move the sliders to see what happens. Note that depending on the speed of
your computer, the response to the slider movements may be more or less
delayed.
• Set the sliders to (approximately) reproduce the situation shown in Fig. 1.15.

Question 2.
 2(a) What are new values of the system boundary conditions pR and psep?
2(b) What are the values of the corresponding FTHP, ptf, and oil flow rate, qo,sc?
2(c) Verify, with the aid of the downstream curve, that for a zero flow rate the
FTHP, ptf, becomes equal to psep. Why?

Task 3.
• Determine what will happen if you reduce the reservoir pressure. Is the flow
rate going to increase or decrease?
• Check your reasoning with the aid of the bottom slider. Note that at low
reservoir pressures two intersections occur. Only the one corresponding to the
highest flow rate is physically relevant. This is not immediately obvious and
requires a dynamic analysis—i.e., a more complex analysis than the steady-
state pressure drop approach used in our nodal analysis—to prove this.
Learning point:
Nodal analysis requires understanding of the underlying physical mechanisms
and assumptions to correctly interpret its results.

Task 4.
• Restore the situation as in Fig. 1.15.
Determine what will happen if you reduce the separator pressure. Is the flow
rate going to increase or decrease?
• Check your reasoning with the aid of the top slider. Surprised? Learning point:
Nodal analysis requires understanding of the behavior of the individual system
components to explain the behavior of the total system.

Task 5 (only if you have access to MATLAB)

• Close the GUI.
• Type the command system_01 in the MATLAB workspace. This invokes the
running of MATLAB m-file system_01.m (which is located in the subfolder
“systems”) and should result in a figure popping up similar to the one in the
GUI. In fact, the GUI is based on a slightly modified version of system_01.m.
Fig. 1.15—GUI for the same production system as in Fig. 1.14, but with different boundary conditions. Now
the upstream and downstream pressure drop curves do intersect, resulting in an operating point indicated
with a black dot. The values of the corresponding FTHP, ptf, and oil flow rate, −qo,sc, can be read off from
the axes and are also displayed at the bottom of the GUI.

• Open the m-file in the MATLAB editor and save it under a different name in the
same folder (e.g., system_01_test.m). Rather than by using the GUI, you can
now change the reservoir and separator pressures by editing them in the
(renamed) m-file. Give it a try and rerun the renamed file to inspect the results.
Moreover, you now have the freedom to change many more parameters.
• Choose pR = 21.5 MPa and psep = 0.7 MPa.
• Read off the corresponding oil flow rate in the graph (it should be
approximately 1.2 × 10–3 m3/s). Determine what would happen if you increase
the tubing size. Do you expect the well to produce more or less oil?
• Verify your reasoning by changing the tubing diameter dw in the renamed m-file
from 0.07600 m to 0.10053 m (which corresponds to replacing a 3½-in.-
diameter tubing by a 4½-in.-diameter one.) Surprised? See the learning points
stated above.
Chapter 2

Properties of Reservoir Fluids

2.1 What Is Covered in This Chapter?

The two-phase (gas/liquid) behavior of hydrocarbons is a major reason for the
nonlinear relationship between flow rate and pressure drop in the various production
system elements. Single-phase reservoir oil will start to boil on its journey to the
surface and will separate into oil and gas phases as a result of reducing pressure.
Consequently, the oil shrinks while the gas massively expands at reducing pressures.
Likewise, single-phase reservoir gas may experience condensation when brought to
the surface and thus also may separate in two phases.
We will focus on relatively simple models for hydrocarbon phase behavior in which
the many components that make up the oil and gas phases are reduced to a mixture
of two pseudocomponents. These two are conveniently chosen as stock-tank oil and
stock-tank gas, which are the oil and gas phases at ambient pressure and
temperature (also known as standard conditions). The so-called black oil or volatile
oil models that we will consider in detail are capable of describing the local density
and flow rate of the gas and oil phases at any combination of pressure and
temperature in terms of the two pseudocomponents. That is, they allow us to express
all flow equations for oil, gas, and water at downhole conditions in terms of just three
(pseudo)components (stock-tank oil, stock-tank gas, and water) with constant
properties. This will simplify the flow equations in the subsequent chapters because at
steady-state conditions the flow rates of the three (pseudo)components in a well
remain constant throughout their journey from reservoir to surface.
Accompanying this chapter is Appendix B, which provides a number of
(semi)empirical correlations to compute the relevant hydrocarbon properties required
to perform production engineering calculations.

2.2 Fluid Properties

The thermodynamic properties of reservoir fluids, such as pressure, temperature,
density, or viscosity, can have a strong influence on the flow in a well and the
production rate. For engineering-oriented information see, e.g., Whitson and Brulé
(2000), Danesh (1998), or McCain (1990); for a more theoretical treatment see
Firoozabadi (1999). During the exploration and appraisal phase of an oil or gas field,
determination of the fluid properties is an important activity. Fluid samples, often
called bottomhole samples, can be collected from the bottom of the wellbore with the
aid of specialized wireline tools. Specialized laboratories perform
pressure/volume/temperature (PVT) analyses to determine the composition—i.e.,
type and relative quantity of each component in the fluid mixture—and the properties
of the fluids at a wide range of pressures and temperatures. The composition is
usually specified in terms of fractions of the various components per mol (or lbm mol)
of fluid sample. For a good description of the reporting of other properties, see
Whitson and Brulé (2000). Some important properties of the most frequently
encountered reservoir fluid components have been reproduced in Tables B-1 and B-2
in Appendix B.
To analyze multiphase flow of hydrocarbons in production systems, we need to
know the state—i.e., thermodynamic condition—of the fluid mixture in each point of
the production system. Apart from the thermodynamic properties of the components,
this requires knowledge of the phase behavior of the mixture. We usually distinguish
three distinct phases: gas, oil, and water, where we consider oil and water as
different phases because they are immiscible. In this book we will not consider the
effect of the presence of a solid phase as may occur when, e.g., asphaltenes or
waxes are present. To what extent the various components of a reservoir fluid mixture
are in the liquid or the gas phase is fully determined by the composition of the mixture
and by a minimal set of fluid properties, called state variables (i.e., PVT). The state
variables are related to each other through an algebraic relationship known as the
equation of state (EOS), as will be discussed in Section 2.5. As a result it suffices to
know only two of the three variables to completely specify the state of an oil/gas
mixture. It is customary to specify fluid properties at a reference state. In the E&P
industry this is done through the definition of standard conditions: a pressure psc =
100 kPa (14.7 psi) and a temperature Tsc = 15°C (60°F), which can be considered as
typical for atmospheric conditions in temperate climates. Oil at standard conditions is
often referred to as stock-tank oil, gas at surface conditions sometimes as stock-tank
gas. The term separator gas usually refers to gas at a slightly higher than
atmospheric pressure and should not be confused with gas at standard conditions.
The fluid properties that are of most interest for production engineering calculations
are these:

• Oil density at standard conditions ρo,sc. In SI units, this is the density of the oil
in kg/m3. In field units, oil density is usually specified by the oil specific gravity
γo, which is the density of the oil relative to that of pure water, both measured
at standard conditions: ρw,sc = 999 kg/m3 (62.4 lbm/ft3 or 8.34 lbm/gal).
However, it is also common to use the API gravity γAPI, which is related to the
specific gravity as ρo = 141.5 / (131.5 + γAPI) and therefore to the density as ρo
= 141.5 × 103/(131.5 + γAPI). Closely related to the oil density is the oil gradient
go, defined as the pressure gradient in an oil column: go = ρo g Pa/m (psi/ft),
where g is the acceleration of gravity: g = 9.80665 m/s2 (32.174 ft/s2).
• Gas density at standard conditions ρg,sc or gas specific gravity γg. The latter is
the density of the gas relative to air, both measured at standard conditions:
ρair,sc = 1.23 kg/m3 (76.3 × 10–3 lbm/ft3). This is equal to the ratio of the gas
 molar mass M (expressed in kg/mol) to the molar mass of air, where Mair =
28.97 × 10–3 kg/mol. Alternatively, it is the ratio of the gas molecular weight Mw
(expressed in g/mol or lbm/(lbm mol) to the molecular weight of air, where Mw,air
= 28.97 g/mol [lbm/(lbm mol)]. In an analogy to the oil gradient, the gas
gradient is gg = ρg g Pa/m (psi/ft).
• Water density at standard conditions ρw,sc or water specific gravity γw, which is
the density of the formation water relative to that of pure water, both measured
at standard conditions. The water gradient follows as gw = ρw g Pa/m (psi/ft).
Formation water will contain many dissolved salts. An equivalent measurement
for density is therefore the NaCl equivalent water salinity.
• Bubblepoint pressure pb . This is the pressure at which first gas is formed when
oil is subjected to a decreasing pressure at a given temperature. If the
pressure at the top of a reservoir is above the bubblepoint pressure, all gas is
dissolved in the oil. However, if the top part of the reservoir is below
bubblepoint pressure, a gas cap exists and the oil is gas saturated. The
bubblepoint pressure is therefore also known as the saturation pressure.
• Solution gas/oil ratio Rs. This is the volume of stock-tank gas, which will
dissolve in a unit volume of stock-tank oil when both are transferred to the given
pressure and temperature conditions. Rs, also referred to as gas solubility, is a
ratio of volumes and hence is dimensionless, but it is dependent on the choice
of units (m3/m3 or scf/STB). The abbreviation “scf” indicates standard cubic
feet, or “ft3 at standard conditions”; “STB” indicates stock-tank barrel or
standard barrel, or “bbl at standard conditions.” We will indicate the solution
gas/oil ratio (GOR) at the bubblepoint pressure with the symbol Rsb . If the
pressure is increased to above the bubblepoint pressure, Rs remains equal to
Rsb and the oil becomes undersaturated. Oil with a zero GOR—i.e., with a
bubblepoint pressure equal to atmospheric pressure psc —is referred to as
dead oil. All other oil is known as live oil. Both pb and Rsb are strongly
dependent on the composition of the oil.
• Solution oil/gas ratio rs. This is the volume of stock-tank oil that will vaporize in
a unit volume of stock-tank gas when both are transferred to the given pressure
and temperature conditions. The oil/gas ratio (OGR) plays an important role in
the production of gas condensates and is therefore also referred to as
condensate/gas ratio (CGR). Yet another name is volatilized oil/gas ratio,
which leads to the use of the symbol Rv (instead of rs) in some publications.
Also rs is a ratio of volumes, and hence dimensionless, but it is dependent on
the choice of units (m3/m3 or STB/ scf). Gas with a very low CGR is referred to
as dry gas.
• Oil formation volume factor Bo. This is the volume occupied by one stock-tank
unit volume of oil, transferred to another condition with a given pressure p and
temperature T. The oil volume will change because of compression, thermal
 expansion, and, in particular, the uptake of gas. The net effect is an increase of
oil volume with increasing depth. In reservoir engineering, Bo is usually specified
at reservoir conditions pR and TR; but in production engineering, Bo may also be
specified at other conditions that occur between the reservoir and the
separator. Bo is a ratio of volumes and hence is dimensionless; unlike the GOR
and OGR, it is independent of the choice of units (m3/m3 or bbl/STB).
• Gas formation volume factor Bg. This is the volume occupied by a unit volume
of gas at standard conditions, transferred to another condition with a given
pressure p and temperature T. (Note that this definition concerns the dry gas
formation volume factor. As discussed later, there is also a wet gas formation
volume factor.) The gas volume will change because of compression, thermal
expansion, dissolution in the oil phase, or uptake of condensate. The net effect
is a decrease of gas volume with increasing depth. Bg is also a dimensionless
ratio of volumes, but in field units different definitions can be used, leading to a
dependency on the choice of units (m3/m3, ft3/scf, or bbl/scf).
• Water formation volume factor Bw. Not surprisingly, this is the volume occupied
by a unit volume of water at standard conditions, transferred to another
condition with a given pressure p and temperature T. Also Bw is dimensionless,
but like Bo, independent of the choice of units (m3/m3 or bbl/STB). Bw usually
has a value close to one, because of the low compressibility and low gas
solubility capacity of water.
• Oil, gas, and water viscosities μo, μg, μw: Usually the dynamic viscosities are
used, with SI units (Pa s) or field units (cp). The viscosities are strongly varying
functions of temperature.
• Interfacial tensions σog, σow, σgw. These quantities play a role in multiphase flow
behavior in production systems, but to a much lesser extent than in flow through
porous media.
• Specific heat capacities. These quantities play a role in the flow of gas through
restrictions. A unit increase in temperature of a unit mass of an ideal gas at
constant volume (and increasing pressure) results in an increase of energy with
cv J, where cv is the specific heat capacity (or specific heat) at constant
volume. The specific heat capacity is therefore expressed in J/kg K (lbf-ft/lbm-
°R). Similarly, an increase in temperature at constant pressure (and increasing
volume) results in an increase of energy with cp J/kg K (lbf-ft/lbm-°R), where
cp is the specific heat at constant pressure.

A simple interpretation of the oil and gas formation volume factors is depicted in
Fig. 2.1. When an amount of Bo m3 of reservoir oil is brought to surface it yields 1 m3
of stock-tank oil, and Rs m3 of stock-tank gas. Or, when 1 m3 of reservoir oil is
brought to surface it yields 1/Bo m3 of stock-tank oil and Rs/Bo m3 of stock-tank gas.
(This stock-tank gas originating from reservoir oil is referred to as solution gas.)
Although the oil itself slightly expands under reducing pressure, the escaping solution
gas makes the oil effectively shrink when it comes to surface. Therefore the ratio 1/Bo
is known as the shrinkage factor, where 1/Bo < 1.

Fig. 2.1—Oil and gas volume flow rates at different two-phase conditions. (Volume flow rates q are used
instead of volumes V based on the underlying assumption that the flow rates are steady-state, i.e., time
invariant.)

Similarly, when an amount of Bg m3 of reservoir gas is brought to surface, it yields

1 m3 of stock-tank gas and rs m3 of stock-tank oil. (This stock-tank oil originating from
reservoir gas is referred to as condensate or volatilized oil.) Even at large amounts
of condensate dropout, the gas still expands so much under reducing pressure that
the ratio 1/Bg is always much larger than unity. In gas engineering it is common
practice to use the symbol E to indicate this gas expansion factor, E = 1/Bg. Values
for Rs, rs, Bo, and Bg as a function of p and T can be determined from a laboratory
analysis of a bottomhole sample in a so-called flash test, in which oil at reservoir
pressure and temperature is brought to standard conditions through a stepwise
reduction of pressure and temperature. Different types of flash tests are in use, and
we refer to the references mentioned at the beginning of this chapter for a detailed
treatment.

2.3 Production Variables

Related to the fluid properties Rs and rs defined above are two frequently used
production process variables:

• Producing gas/oil ratio Rgo. This is the instantaneous ratio of the gas and oil
flow rates measured at surface during production: Rgo = qg,sc / qo,sc. If water is
present in the production stream, we can extend the concept of the producing
GOR to a producing gas/liquid ratio (GLR) Rgl = qg,sc / (qo,sc + qw,sc). These
 quantities are also referred to as the instantaneous GOR and GLR. The
producing GOR should not be confused with the produced GOR Rp, as used by
reservoir engineers in material-balance calculations, which refers to the
cumulative ratio of gas and oil produced from the reservoir since the start of
production. Many publications use R, without subscript, to indicate the
instantaneous producing GOR; but confusingly, some also use Rp. When
producing oil from a reservoir above bubble-point pressure, Rgo will be identical
to Rsb —i.e., to Rs at bubblepoint pressure and reservoir temperature. If during
production the reservoir pressure drops below bubblepoint pressure, Rgo could
in theory drop below Rsb . In practice, however, usually the opposite occurs
because the gas viscosity is much lower than the liquid viscosity, which makes
the gas flow to the well much more easily than the oil. In addition, gas coning
may occur, which implies that free gas is pulled in from the gas cap. Together
with the associated gas that is released from the oil during its travel up the
wellbore, this may result in an Rgo that is significantly higher than Rsb .
• Producing oil/gas ratio rog. In analogy to the producing GOR, this is the
instantaneous ratio of the oil and gas flow rates measured at surface: rog = qo,sc
/ qg,sc. As mentioned before, the OGR plays an important role in the production
of gas condensates and is therefore also referred to as the producing CGR or
the condensate yield. Note that in reporting field production we could either use
Rgo (the usual choice for black or volatile oil reservoirs) or rog (the usual choice
for gas/condensate reservoirs).

A third production variable that is frequently used concerns the combined production
of water and oil:

• Water/oil ratio Rwo. This is the volume of water produced at surface together
with a unit volume of oil, both measured at standard conditions, or, in terms of
flow rates: Rwo = qw,sc / qo,sc. An alternative measure is the water cut, which is
the fraction (or percentage) of water in the total volume of produced liquids (oil
and water) measured at standard conditions: fw,sc = qw,sc / (qo,sc + qw,sc). Both
measures are dimensionless and are independent on the choice of units (m3/m3
or STB/STB). Oil with a zero or very low water/oil ratio (WOR) is often referred
to as dry oil. Sometimes the concept of ‘base sediment and water’ (BSW) is
used to indicate the amount of solids and water as a fraction of the total
amount of solids and liquids in the wellbore flow. Because the amount of solids
is usually very low, the BSW value is in practice almost identical to the water
cut.

2.4 Pressure/Temperature Phase Diagram

Fig. 2.2 displays the phase diagram for a hydrocarbon mixture. To the left of the
bubblepoint line, the system acts as a single-phase liquid and all the gas is dissolved.
To the right of the dewpoint line, the system acts as a gas. If the bubblepoint line is
crossed when coming from the liquid phase, gas is liberated to form a two-phase
system. Moving farther toward the dewpoint line, an increasing amount of gas comes
out of solution. Conversely, if the dewpoint line is crossed when coming from the gas
phase, liquid condenses. At the critical point, the distinction between liquid and gas
cannot be made because at that particular pressure and temperature the liquid and
gas phases have identical densities. In general, at pressures above the cricondenbar
(the maximum pressure at which two phases can coexist) in combination with
temperatures above the cricondentherm (the maximum temperature at which two
phases can coexist), the fluid is in a supercritical state, also known as dense phase.

Fig. 2.2—Phase diagram for a hydrocarbon mixture.

Fig. 2.2 also indicates a classification of reservoir types based on the phase
diagram:

• Undersaturated oil reservoirs have initial pressures above the bubblepoint line
and temperatures to the left of the critical point. During production of a
reservoir, the reservoir pressure will drop while the reservoir temperature
remains unchanged. This can be represented by a vertical line in p-T space.
When the line crosses the bubblepoint line, gas is coming out of solution.
 • Saturated oil reservoirs, also known as gas cap reservoirs, have initial
pressures already below the bubblepoint line.
• Gas condensate reservoirs have initial pressures above the dewpoint line and
initial temperatures between the critical temperature and the cricondentherm.
During production of a gas/condensate reservoir, condensation occurs when the
pressure drops below the dewpoint line. This effect, which is called retrograde
condensation, may seem somewhat counterintuitive because we usually
experience condensation when the pressure of a gas/liquid mixture increases
rather than decreases. Although it appears from the phase diagram that at
even lower pressures the condensate would return to the gas phase again, this
is usually not the case. Because the condensed liquids are much less mobile
than the gas, they stay behind in the formation while the gas is produced. As a
result, the reservoir fluid composition changes and the entire phase diagram
changes its form and moves to the right such that vaporization of the remaining
condensate may never occur.
• Dry gas reservoirs have temperatures to the right of the cricondentherm and
do not experience this problem.

Using a similar terminology but a slightly different classification, we can distinguish

four categories of hydrocarbon fluids:

• Black oil: oil for which, at reservoir conditions, the solution GOR Rs < 350
m3/m3 (approximately 2000 scf/STB)
• Volatile oil: oil for which Rs > 350 m3/m3 (approximately 2000 scf/STB)
• Wet gas (or condensate gas): gas for which the condensate/gas ratio rs > 30
m3/million m3 (approximately 5 STB/million scf)
• Dry gas: gas for which the condensate/gas ratio rs < 30 m3/million m3
(approximately 5 STB/million scf)

As shown in Fig. 2.3, when oil flows up the production tubing, it follows a path in p-T
space in which gas is liberated and expands as it goes up the tubing. As a result, the
amount of liquid decreases, so there is a shrinkage in the volume of oil. Furthermore,
unlike the pressure drop in the reservoir, which is isothermal, the pressure drop in the
tubing is accompanied by a drop in temperature. When producing from a wet gas
reservoir, this may result in condensate dropout and liquid loading of the wells.

2.5 Equations of State

2.5.1 Vapor/Liquid Equilibrium. As discussed in Section 2.2, an EOS specifies an

algebraic relationship between the state variables pressure, volume, and
temperature. With the aid of such an EOS, all properties of hydrocarbon mixtures as
used in production engineering can be determined, provided an accurate
compositional description is available from laboratory experiments on fluid samples. In
particular, we can use an EOS to determine the so-called equilibrium factors (K
values) that are needed to describe the equilibrium between components in the liquid
and the vapor phase. The K values allow us to compute the composition of the liquid
and the gas phases in a multiphase fluid mixture at any given pressure and
temperature. These so-called flash calculations or vapor/liquid equilibrium (VLE)
calculations, which use an EOS to determine the K values, require a level of numerical
computation outside the scope of this book. For more information on the EOS-based
approach, see the references mentioned at the beginning of this chapter. Although we
do not discuss multiphase EOS definitions, we will use simple expressions for the
EOS of single-phase fluids as discussed in the following two sections.

Fig. 2.3—Path in p-T space as the oil/gas mixture flows from the reservoir to the terminal.

2.5.2 Single-Phase Gas. The relationship between pressure, volume, and

temperature of an ideal gas is given by an EOS known as the ideal-gas law, which
can be expressed in different forms in SI units as
or in field units as

where p is pressure, Pa, (psia); V is volume, m3; (ft3), n is amount of gas, mol, (lbm
mol); R is the universal gas constant equal to 8.314 J/K mol, (10.73 psia-ft3/°R-(lbm
mol)); Tabs is absolute temperature, K, (°R); m is mass, kg, (lbm); ρg is gas density,
kg/m3, (lbm/ft3); M is molar mass, kg/mol; Mw is molecular weight, (lbm/lbm·mol). See
Section A-4.2 of Appendix A for the numerical relationship between the molar mass
M, the molecular weight Mw, the specific gravity γg, and the density ρg,sc.
The ideal-gas law is valid only at pressures much below those normally
encountered in the E&P industry. Approximate relationships, valid at higher pressures,
are given in SI units by

and in field units by

where Z is the gas deviation factor, also known as the gas compressibility factor or
simply the Z factor. Correlations developed by Standing and Katz (1942) are normally
used to extend this relationship to hydrocarbon gas mixtures (see Appendix B).

2.5.3 Single-Phase Oil. The relationship between pressure, volume, and temperature
of single-phase oil is given by an EOS in the form of an expression for compressibility.
Isothermal oil compressibility is defined as the decrease in volume per unit of
pressure, –dV/dp, per unit of volume V (or the increase in density per unit volume,
dρ/dp, per unit of density ρ) at constant temperature T:
where co is the compressibility coefficient, which typically is itself a function of
pressure and temperature. The large heat capacity of an oil reservoir justifies the
assumption that oil expansion in the reservoir during production is an isothermal
process. When flowing through the wellbore to the surface, the fluid mixture gradually
cools down, but it is often still assumed that the expansion occurs pseudoisothermally
with a compressibility coefficient that gradually changes with decreasing pressure and
temperature. (An empirical correlation for co is given in Section B-2.6 in Appendix B.)
Eq. 2.13 can be interpreted as a differential equation in V and p that allows for
separation of variables:

We specify a boundary condition in the form of

and we assume that the dependence of co on p is small enough to linearize the right
side of Eq. 2.14. The solution can then be written as

Remembering that co is a function of p and T, Eq. 2.16 can be interpreted as an EOS

that describes the PVT behavior of single-phase oil. Because the density ρo is
inversely proportional to the volume V, we can also write Eq. 2.16 as

Single-phase oil is at a pressure above the bubblepoint pressure, and a natural choice
for the reference pressure pref is therefore the bubblepoint pressure at a certain
temperature T, leading to

where the values of pb and ρob can be determined from laboratory experiments or
from empirical correlations as discussed in Appendix B. A special case is obtained for
dead oil, which has a bubblepoint pressure equal to the atmospheric pressure, in
which case pref = psc.

2.6 Oil Models

2.6.1 Compositional Models. In a compositional model of a multiphase hydrocarbon
mixture, the compositions of the liquid and the gas phases are functions of pressure
and temperature and need to be determined from experiments, possibly combined
with flash calculations using an EOS. Hydrocarbon mixtures may consist of many tens
of components, and a full compositional analysis taking into account all components
would be very time consuming. Furthermore, it is often quite difficult to accurately
establish the amount and the properties of all components, in particular those with a
high molar mass, the so-called heavy fractions. Therefore it is customary to lump
these into a pseudocomponent. This is typically done for heptane and all heavier
fractions, in which case the pseudocomponent is referred to as C7+. In its most simple
form, a compositional model consists of only two pseudocomponents, one for the
lighter and one for the heavier hydrocarbons, often referred to as heavies and lights.
Such a two-component model, also known as a binary-mixture model, is too crude to
accurately describe the behavior of gas/condensate systems. However, it is usually
sufficient to describe the behavior of black oils or even volatile oils.

2.6.2 Volatile Oil Model. The volatile oil model is a two-component model that
accounts for compositional variations in both the liquid and the gas phase. Other
names are modified black oil model, extended black oil model, or volatilized oil
model. The pseudocomponents in the volatile oil model are stock-tank oil and stock-
tank gas (“heavies” and “lights”), which can each be characterized with just a single
parameter per pseudocomponent: ρo,sc and ρg,sc in SI units, or γo (or γAPI) and γg in
field units. Note that this is an assumption. In a more accurate model, it should be
taken into account that the properties of stock-tank gas originating from reservoir gas
are different from those of stock-tank gas originating from reservoir oil (i.e., solution
gas). The same holds for the properties of stock-tank oil resulting from reservoir oil or
from reservoir gas (i.e., condensate). The difference may be of relevance—e.g., for
the accurate design of separation facilities for condensates or volatile oils. However,
for most reservoir engineering or production engineering calculations, the assumption
is acceptable.
The composition of the oil at or above bubblepoint pressure is fully defined by the
bubblepoint GOR Rsb . The compositions of the oil and the gas at pressures below
bubblepoint are specified by the solution GOR Rs and the solution OGR rs, which are
both functions of pressure and temperature. To describe the change in density (or
volume) of the oil and gas phases with changing pressure and temperature, we need
two additional variables, the oil and gas formation volume factors Bo and Bg. With the
aid of Fig. 2.1 we can derive the mass balance equations for oil and gas that are
brought from downhole conditions (also known as local or in-situ conditions) to
standard conditions:
which give us the required expressions in terms of densities:

Since Bg, Bo, Rs, and rs are functions of pressure and temperature, Eqs. 2.21 and
2.22 can be interpreted as EOSs. They can be conveniently written in matrix form,
and if we also include the mass balance equation for water,

this results in

or

where we have assumed that Bw = 1 and Rsw = 0, which implies that gas solubility,
compressibility, and thermal expansion for water are so small that they can be
neglected. We can use Fig. 2.1 to derive similar matrix expressions for the volume
flow rates, resulting in
or

Note that we could equally well have used volumes instead of flow rates. For
example, the second row in Eq. 2.26 can also be expressed as Vo,sc = rs/Bg × Vg +
1/Bo × Vo.
Eqs. 2.24 and 2.27 have been programmed in MATLAB function local_q_and_rho.m.
Determination of the necessary values of Bg, Bo, Rs, and rs as functions of pressure
and temperature is usually made with the aid of PVT tests and compositional
analysis. This is outside the scope of our book, and therefore we will mainly use an
even further simplified formulation, the so-called black oil model as described in the
following section. An example of the use of tabulated volatile oil parameters Bg, Bo,
Rs, and rs is discussed in the MATLAB assignment (Hydrocarbon Properties) at the end
of this chapter. Several varieties of the volatile oil model have been developed (for an
overview, see Whitson and Brulé 2000). A graphical representation of the volatile oil
parameter definitions as used in our text is given in Fig. 2.4.

2.6.3 Black Oil Model. Just like the volatile oil model, the black oil model is a two-
component model. However, it assumes a constant composition of the gas phase and
accounts only for compositional variations in the liquid phase. The relevant equations
of the model follow directly from those of the volatile oil model by substitution of rs =
0. In particular, Eqs. 2.24 to 2.27 become
Fig. 2.4—Graphical representation of the volatile oil parameter definitions (after Whitson and Brulé 2000).
Superscripts o and g are used to indicate the origins of contributions to the volume flow rates at standard
conditions. For example, represents the part of the gas phase at standard conditions that originates
from the downhole oil phase.
Just as for the volatile oil models, the necessary values of Bg, Bo, and Rs as functions
of pressure and temperature may be obtained with the aid of PVT tests and
compositional analysis. However, during the early development phase of an oil field no
fluid samples may be available. In that case it is necessary to fall back on
correlations, which are relationships for “typical” oil and gas compositions, based on
experimental data. Especially for black oils such correlations can be reasonably
accurate. In addition, production engineering calculations based on correlations
require much less computational effort than do calculations based on compositional
analysis using an EOS. Therefore, black oil correlations are widely used. However, to
describe the behavior of volatile oil or gas/condensate systems, correlations are
usually of limited value, and thus performing PVT analyses on fluid samples is
required to allow proper compositional calculations.

2.7 Fluid Property Calculations

2.7.1 Gas Formation Volume Factor. An expression for the gas formation volume
factor follows from Eqs. 2.24 and 2.27 as

For black oil—i.e., for rs = 0—this reduces to

Numerical values for Bg can be computed as follows. With the aid of the gas law for
nonideal gases, Eq. 2.7, we obtain

where the last equality results from the fact that, by definition, Zsc = 1. Here we
added the subscript w to indicate that this equation concerns the wet gas formation
volume factor, because it is based on the assumption that we consider volumes of
identical amounts of substance n at downhole and standard conditions. This
assumption is fully justified for single-phase gas; but where applying Eq. 2.34 in two-
phase computations, take into account that condensate may be present at surface
conditions. Usually we are more interested in the dry gas formation volume factor Bg,
which is related to its wet gas counterpart according to
Note that for black oil and single-phase gas, rs = 0 and therefore Bg = Bgw.

2.7.2 Oil Formation Volume Factor. An expression for the oil formation volume
factor follows from Eqs. 2.24 and 2.27 as

which for black oil, i.e., for rs = 0, reduces to

The actual computation of Bo is based on either experiments or correlations. For

pressures above the bubblepoint pressure the oil formation volume factor is given by

An empirical correlation for co is given in Appendix B.

2.7.3 Black Oil Correlations. The standard reference for black oil correlations is
Standing (1952), and these “Standing correlations” for pb , Rs, and Bo have been
programmed in the MATLAB function black_oil_Standing.m. For details of the
correlations see Appendix B, and for an overview of the corresponding MATLAB
routines see Appendix H. A flow chart displaying the various computational steps in
MATLAB function black_oil_Standing.m has been displayed in Fig. 2.5. Many other
correlations have been developed over the past half-century, and an extensive
overview is given in Appendix B of Brill and Mukherjee (1999); further information can
be found in the references mentioned in Section 2.2 above. As an example of an
alternative correlation, the “Glasø correlations” (Glasø 1980) have been programmed
in MATLAB function black_oil_Glaso.m. Figs. 2.6 and 2.7 illustrate the typical behavior
of Bo and Rs with increasing pressure, reflecting the increasing amount of gas
dissolving in the oil. After the bubblepoint pressure has been reached—i.e., after the
maximum amount of gas has been dissolved—Rs stays constant (at a value of Rsb , by
definition). The value of Bo slightly decreases because of compression of the
undersaturated oil with increasing pressure.

2.8 Questions

2.1 What is the difference between the solution GOR and the producing GOR in
 an oil well?
2.2 What is the difference between ordinary and retrograde condensation?
2.3 The thermodynamic behavior of a hydrocarbon mixture can be characterized
with an EOS if we know its composition (in terms of component fractions) and
the thermodynamic properties of the components. A strongly simplified
description of hydrocarbon mixtures is the black oil model. How are the EOS,
the composition, and the (pseudo)component properties defined in the black
oil model?
2.4 What is the difference between the oil phase and the oil (pseudo)component
in a volatile or black oil model?

Fig. 2.5—Flow chart displaying the various computational steps in MATLAB function black_oil_Standing.m.
Fig. 2.6—Oil formation volume factor as a function of pressure. The graph is based on the Standing (1947,
1952) correlations applied to a gas/oil mixture with fluid properties: ρg,sc = 0.95 kg/m3 (γg = 0.77), ρo,sc =
850 kg/m3 (γAPI = 35°), and Rsb = 100 m3/m3 (561 scf/STB) at a temperature of 60°C (124°F). The
corresponding bubblepoint pressure is 14.4 MPa (2,089 psi).
Fig. 2.7—Solution GOR as a function of pressure. The graph is based on the Standing (1947, 1952)
correlations applied to a gas/oil mixture with the same fluid properties and at the same temperature as in
Fig. 2.6.

2.5 Why do we use pseudoreduced pressures and temperatures (and not just
reduced values) to compute the Z factor of hydrocarbon mixtures?
2.6 Demonstrate that for the volatile oil model we have ρg · qg + ρo·qo = ρg,sc · qg,sc
+ ρo,sc·qo,sc. What is the meaning of this equality?
• The PVT data for a volatile oil well producing at 0% water cut and
corresponding to a pressure of 50 MPa and a temperature of 50°C are
represented as Bg = 0.0025 m3/m3, Bo = 1.55 m3/m3, Rs = 200 m3/m3, and rs =
0.0008 m3/m3. What is the local gas/liquid fraction if the well produces 1,000
B/D at a producing GOR of 2000 scf/STB? (Give your answer in SI units).

For questions 2.8 through 2.12, use the oil and gas correlations in Appendix B. The
correlations have been programmed in various MATLAB routines (see Appendix H).
However, the computations can also be performed by hand calculation.

2.7 An oil reservoir has a pressure of pR = 17 MPa at the GOC and a

temperature of 76°C. The bubblepoint pressure of the oil is pb = 19.5 MPa.
The gas and oil densities at standard conditions are ρg = 1.11 kg/m3 and ρo =
910 kg/m3. What is the solution GOR of the oil?
 2.8 Refer to Question 2.8. What is the specific gravity of the gas produced at
surface? And what is the density of the gas-cap gas just above the GOC in
the reservoir?
2.9 Refer to Question 2.8. What are the oil and gas viscosities if the reservoir
pressure is 22 MPa and all other parameters remain the same?
2.10 Refer to Question 2.8. What is the compressibility coefficient co of the oil?
2.11 Consider a well that produces dry black oil with a producing GOR of 250
m3/m3. The production history shows no indication of free gas production. The
densities of the gas and oil at standard conditions are given by ρg,sc = 1.02
kg/m3 and ρo,sc = 805 kg/m3. What is the oil formation volume factor at the
following pressure and temperature combinations: p = 15 MPa and T = 85°C,
and p = 30 MPa and T = 105°C?

2.9 MATLAB Assignment: Hydrocarbon Properties

2.9.1 Objectives

• Perform a simple MATLAB programming exercise and create annotated plots.

• Become familiar with the oil and gas parameters Bg, Bo, rs, and Rs and their
dependency on pressure and temperature.
• Become familiar with the MATLAB function files black_oil_Standing.m and
local_q_and_rho.m and with the use of tabulated volatile oil parameters.

2.9.2 Assignment

Task 1
• Recreate Figs. 2.6 and 2.7 (i.e., plots of Bo and Rs vs. pressure p) by running
script file example_plot_black_oil_1.m. Note that in addition to creating plots
of Bo and Rs, the file also creates two plots of Bg vs. p, one on a linear scale
and one on a semilogarithmic scale.
• Inspect the contents of the script file. It uses a simple for-loop, which calls the
function file black_oil_Standing.m to obtain the three black oil parameters Bg,
Bo, and Rs for increasing values of the pressure p at a fixed temperature T. The
values are stored in a matrix named results with n_step rows and three
columns.
• Inspect the contents of black_oil_Standing.m and verify the sequence of
computational steps with the aid of the flow chart depicted in Fig. 2.5.
Notes
• All preprogrammed MATLAB function files listed in Appendix H start with a series
of comment lines to briefly explain the actions performed in the function and to
list the input and output parameters in the header, i.e., the first line of the
 function. Note that the second line of the function is always a repetition of the
header—but with a comment symbol % in front. This is done to echo the header
information with the aid of the help function. For example, type help
black_oil_Standing in the MATLAB workspace and watch the result.
• If you are an experienced MATLAB user, you may recognize that the use of a
forloop in plot_black_oil_1.m could have been avoided by using a technique
called vectorization. However, in the MATLAB examples used in this book we will
not make use of this technique but will instead use classical loops to increase
the readability of the code.
• The size of the matrix results has been initialized just before the for-loop by
defining it as a matrix filled with zeros. This is to avoid the matrix increasing one
row in size every time the for-loop is executed, which would slow down the
computation.
Question 1
1(a) What should be the value of Bo at standard conditions?
1(b) Is this in line with the result in your plot. Why not?
1(c) Repeat the plot with modified parameter values such that Bo obtains the
value corresponding to standard conditions. What did you change?
1(d) Why has the Bg curve a (near-)hyperbolic shape? Which equation explains
this behavior?
Task 2
• Inspect the function file local_q_and_rho.m. The variables q, q_sc, rho and
rho_sc are column vectors of three elements each. What is their meaning?
• Run script file example_plot_black_oil_2.m. It creates plots of local
(downhole) oil and gas rates as a function of pressure p and at a fixed
temperature T for a given unit surface oil rate and the corresponding surface
associated gas rate Rsb . The black oil properties are identical to those used in
example_plot_black_oil_1.m. Inspect the contents of the script file.

Question 2
2(a) Why is the local oil rate curve identical to the oil formation volume factor
curve plotted under Task 1?
2(b) Is the local gas rate curve also identical to the gas formation volume factor
curve plotted under Task 1? If not, what explains the difference?
Task 3. Create a MATLAB script file that performs the following steps:
• Load the volatile oil parameter table with the following commands:
file name = ‘vol oil table 01’;
load(file name,’vol oil’);
• The variable vol_oil is a 3D matrix containing values of T, p, Bg, Bo, rs, and Rs.
The first dimension is used to loop over the temperature values; i.e., it just
contains integers 1, 2, …, m, where m is the number of temperature values.
The second dimension loops over the pressure values (i.e., integers 1, 2, …,
 n), and the third dimension contains the values of T, p, and the four volatile oil
parameters.
• Determine the number of elements in each dimension, as well as the lower and
upper bounds of the temperature and pressure ranges, with the following
statements:
[m,n,k] = size(vol_oil)
T_lo = vol_oil(1,1,1)
p_lo = vol_oil(1,1,2)
T_hi = vol_oil(m,1,1)
p_hi = vol_oil(1,n,2)
• Create plots of Bg, Bo, rs, and Rs, as a function of pressure over the range
determined above, for a temperature T of choice (but within the range
determined above). For this purpose, save the file
example_plot_black_oil_1.m under the name
example_plot_volatile_oil_1.m and, inside this new file, modify the call to
function black_oil_Standing.m into a call to volatile_oil.m. Adjust all
comment lines as appropriate.
• Optionally, display two graphs in each plot, one for the lowest and one for the
highest value of T as determined above. Hint: Repeat the plot statement and
use the command hold on between the two plot statements. Add a legend.
• Inspect the contents of volatile_oil.m and note how 2D interpolation (i.e.,
between p values and between T values) is used to obtain the volatile oil
parameters.
• Note that when using a volatile oil table, the input parameters ρg,sc, ρo,sc, and
Rsb should be consistent with those used to generate the tabulated values. In
particular, for vol_oil_table_01 one should use rg,sc = 0.80 kg/m3 (γg = 0.65),
ρo,sc = 800 kg/m3 (γAPI = 45°), and Rsb = 450 m3/m3 (2,526 scf/STB).
Question 3
3(a) What could be a reason for the minimum in the curves for rs?
3(b) What is different in local_q_and_rho.m when using a volatile oil table
instead of a black oil model?

2.9.3 Deliverables

• MATLAB program listing for Task 3

• Plots
• Answers to the questions
Chapter 3

Single-Phase Flow in Wells and

Pipelines

3.1 What Is Covered in This Chapter?

As described in Chapter 1, nodal analysis of a production system is based on a
representation of the system in the form of individual elements connected at nodes. In
this chapter we will consider elements that describe the flow through circular conduits
such as pipelines, flowlines, or wells. We use a 1D description and consider single-
phase flow of oil, gas, or water. Multiphase flow will be the topic of the next chapter.
Under the assumption of a known temperature profile along the pipe, we start from
differential equations for conservation of mass and momentum and an algebraic EOS
in terms of pressures and densities of the fluid. Assuming steady-state conditions, we
arrive at a first-order nonlinear ordinary-differential equation for pressure along the
pipe, which reveals that there are three contributions to pressure drop: gravity,
friction, and acceleration. We discuss how, given a boundary condition, this equation
can be integrated numerically with the aid of standard numerical integration routines
to produce a traverse—i.e., a graph of pressure as a function of distance for a given
flow rate. Under simplifying assumptions, the integration can also be performed
analytically both for single-phase liquid flow and for single-phase gas flow, although
the latter requires some additional tricks to deal with the nonlinear effects of
compressibility.
Once we can compute the pressure drop over a pipe for a given flow rate, we can
repeat the computation many times for different flow rates, while keeping the
pressure at one end of the pipe fixed, to produce a graph of the varying pressure at
the other end as a function of rate. This procedure results in (upstream or
downstream) performance curves that can be used to characterize the flow
performance of the pipe element. As discussed in Chapter 1, the use of such
performance curves forms a key step in classic nodal analysis.

3.2 Governing Equations

3.2.1 Mass Balance, Momentum Balance, and Equation of State. In this section
we will derive the 1D equations for single-phase fluid flow in circular pipes under the
assumption that the temperature profile along the pipe is known. For a detailed
treatment of the nature of the equations see, e.g., Bird et al. (2002), who also treat
the case in which the temperature is not known in advance. Consider a section of an
inclined pipeline with constant cross-sectional area (see Fig. 3.1). We can write the
mass balance per unit time for the section as

where A is cross-sectional area of the pipe, m2; ρ is fluid density, kg/m3; v is fluid
velocity averaged over the cross section, m/s; s is coordinate along the pipe, m; and t
is time, s. In reality the velocity will change over the cross section of the pipe. It is
zero at the wall and reaches its maximum value at the center. However, the use of an
averaged velocity is accurate enough for our purpose.
Initially we assume that the fluid velocity is always positive—i.e., that the fluid
always flows in the positive coordinate direction. The momentum balance per unit
time (Newton’s law) can then be written as

where p is pressure, Pa; Fg(ρ,s) is gravity force per unit length, N/m; Ff (ρ,μ,v) is
friction force per unit length, N/m; and μ is dynamic viscosity, Pa·s. The nature of the
gravity force Fg(ρ,s) and the friction force Ff (μ,ρ,v) will be discussed in more detail in
Sections 3.22 and 3.23. The viscosity μ is a known function of pressure and
temperature, where the temperature is a known function of s. A third equation for the
remaining unknown variables ρ, v, and p is given by the equation of state (EOS) for
the fluid. For example, if we consider the flow of single-phase gas, we can use Eq.
2.9 derived in Section 2.5; for single-phase oil flow we can use Eq. 2.17. If we
expand Eqs. 3.1 and 3.2, drop all terms higher than first order in the differential ds,
and simplify the results, we can write the three equations as

Fig. 3.1—Segment of an inclined pipeline.

where the compressibility co and the gas deviation factor Z are known functions of p
and T; note that as discussed above, T itself (and therefore also Tabs ) is a known
function of s.

3.2.2 Head Loss. The head loss or gravity loss (i.e., the gravity force per unit length)
for a pipe element is defined as

where g is acceleration of gravity, m/s2; and θ(s) is pipeline inclination, rad. In

pipeline engineering, the inclination θ is defined as the angle of the pipeline axis with
the respect to the horizontal plane. The term “sin θ ”, which can be positive or
negative, is therefore a measure of the change in elevation of the pipeline axis per
unit length of measured distance s. The inclination is usually known as a function of s,
either theoretically as one of the design parameters of the pipeline or actually from
measured data obtained during a pipeline survey. In well engineering it is common
practice to define the wellbore geometry with a slightly different set of parameters
(see Fig. 3.2). The inclination α of the well is defined as the angle between the
wellbore axis and the vertical direction. The term “cos α ” is therefore a measure of
the change in true vertical depth z, which is measured downward, per unit length of s,
which is now known as along-hole depth (AHD), or measured depth, and naturally is
also positive in the downward direction. As a result of the different definition of the
inclination and because s is positive going downward, the head loss follows as

During the design phase, the well path, also known as the well trajectory, is specified
either in rectangular coordinates or in terms of survey coordinates: AHD, inclination,
and azimuth (i.e., the angle with respect to north). During and after drilling a well, the
actual trajectory can be obtained from a wellbore survey which can be performed
with survey tools run on wireline or with so-called measurement-while-drilling (MWD)
tools incorporated in the drillstring. Appendix C gives an algorithm to convert from
survey coordinates to rectangular coordinates, and describes MATLAB functions for
survey evaluation and plotting of well trajectories. Note that the choice for a
downward positive direction of s also implies that wellbore flow to surface has a
negative velocity. We will therefore use the sign convention that flow rates related to
oil, gas, or water production have a negative sign, whereas all flow rates related to
injection are positive. We will use this convention for flow in pipelines, flowlines,
wellbores, and the near-wellbore region in the reservoir.

Fig. 3.2—Segment of a deviated well.

3.2.3 Friction Loss. The friction loss for single-phase flow in pipes with a circular
cross section can be expressed as

where d is the inside diameter of the pipe, m; f is the Moody friction factor (also
known as the Darcy-Weisbach friction factor); and q = v × A is the flow rate, m3/s.
Note the use of the absolute sign in the definition of the friction force: The
dependency of the friction force on −v|v| (or −q|q|) implies that it is always pointing in
a direction opposite to the velocity (or the flow rate). For detailed information on the
physics underlying the friction loss in pipe flow we refer to Bird et al. (2002) or one of
the many other available fluid mechanics textbooks, e.g. White (2016). Here we
merely state the key aspects.
The dimensionless friction factor f is a function of μ, ρ, and v (or q) through its
dependence on two other dimensionless numbers: the Reynolds number NRe and the
scaled pipe roughness ε. The Reynolds number is defined as

where, as discussed before, we assume that μ is a known function of p and T(s). It

provides a measure of the relative importance of inertial forces (governed by ρ) and
friction forces (governed by μ). The scaled pipe roughness is defined as

where e is the pipe roughness expressed in m, which we assume to be constant over

the length of the pipe. For Reynolds numbers lower than 2,000 the flow is laminar,
and f is given explicitly by

whereas for Reynolds numbers larger than 3,000 the flow is turbulent and f is given
by the Colebrook (1939) equation:

Eq. 3.13 is an implicit functional relationship between the three governing

dimensionless numbers (f, NRe, and ε) and was derived from a large number of
experiments by various researchers (see Colebrook 1939, Moody 1944 and
references therein). Because of its implicit nature it has to be solved iteratively which
is conveniently done numerically, e.g., with Newton-Raphson iteration or the
predefined MATLAB function fzero (see Appendix D). Alternatively, one of the several
available explicit approximations can be used—e.g., the one of Zigrang and Sylvester
(1985), which is defined as

Fig. 3.3, known as the Moody chart, displays the change of friction factor f with
increasing Reynolds number NRe for various values of the dimensionless roughness ε,
as computed with the original implicit Eq. 3.13. For flow in the intermediate regime,
characterized by Reynolds numbers between 2,000 and 3,000, one can use a linear
interpolation between Eqs. 3.12 and 3.13. The figure has been generated with the aid
of the MATLAB file Moody_friction_factor.m which combines Eqs. 3.12 and 3.13 with
linear interpolation to cover the entire range of low and high Reynolds numbers. As
similar expression that uses the explicit approximation Eq. 3.14 instead of Eq. 3.13, is
available as Zig_and_Syl_fric_fact.m.
Fig. 3.3—Moody chart: friction factor f as function of Reynolds number NRe for various values of
dimensionless roughness ε.

3.2.4 Pressure Drop Components. If we restrict the analysis to steady-state flow—

i.e., if we assume that ρ, v, and p are functions only of s, not of t—Eqs. 3.3 and 3.4
can be rewritten as

Using the equality

substituting Eqs. 3.7, 3.9, and 3.14, and reorganizing the outcome results in an
expression for the pressure drop per unit length dp/ds,
Here we have taken into account the possibility of negative fluid velocities (which
according to our sign convention correspond to production) through the use of the
absolute value of the velocity |v|. The three terms at the right side of Eq. 3.18 can be
interpreted as follows:

• Head loss or gravity loss is the static change in pressure caused by the
change in the pipe’s elevation. In near-horizontal pipelines this component is
negligible, but it is usually the most important component in a well. The
pressure between surface and well bottom changes greatly, simply because of
the weight of the column of fluid in the well, even if it is not flowing.
• Friction loss is caused by the dissipation of energy by viscous forces in the
fluid. This term depends strongly on the fluid properties, the flow regime
(laminar or turbulent), and the fluid velocity. It is usually the most important
component in pipelines.
• Acceleration loss (sometimes referred to as kinetic energy loss) is caused by
the change in momentum when the fluid is accelerated in the well because of
expansion. Generally this term is less important, but it can become of
significance for very high-rate gas wells.

3.3 Single-Phase Oil Flow

Here we focus on single-phase oil flow; we will not treat the case of single-phase
water flow separately because it can be directly obtained from the single-phase oil
case.

3.3.1 Governing Equations. Maintaining the assumption of steady-state flow, the

right side of the mass conservation Eq. 3.3 vanishes, which reduces it to

This is a first-order ordinary-differential equation, though a trivial one, requiring one

boundary condition that can be chosen at an arbitrary reference point s = sref as

Integrating Eq. 3.19 and using the boundary condition in Eq. 3.20, we find that
Next, the acceleration term ρv dv/ds in Eq. 3.18 can be rewritten as

where we dropped the subscripts o to improve the readability of the expression, and
where we made use of Eqs. 3.6 and 3.21 to compute dρ/dp and dv/dρ. Combining
Eqs. 3.6, 3.18, 3.21, and 3.22, we can now write the governing equation for steady-
state single-phase oil flow as

where v and ρ are given by Eqs. 3.21 and 3.6 respectively (while continuing to drop
subscripts). Eqs. 3.6, 3.21, and 3.23 together form a set of three differential-
algebraic equations for the three unknowns ρ, v, and p. (Note that in some
publications μ is considered a fourth unknown variable, in which case a fourth
equation is introduced in the form of the relationship between μ, p, and T, sometimes
referred to as a second equation of state.)
Comparison with Eqs. 3.3 through 3.6 shows that the mass balance differential Eq.
3.3 has now been replaced by an algebraic equation (Eq. 3.21) as a result of the
restriction to steady state. The set of three equations can be simplified by recognizing
that the compressibility of single-phase liquids is often quite small, which implies that
the acceleration term (i.e., the first term at the right side of Eq. 3.23) is nearly equal
to unity. If we furthermore make use of Eqs. 2.28 and 2.31 (with Rs = Rsb ) and Eq.
2.38, the simplified set of equations for single-phase oil flow becomes

Alternatively, we can combine Eqs. 3.24 through 3.26 in a single differential equation.
Using the fact that for circular conduits it holds that A = ¼πd2, this results in

For dead oil we have Rsb = 0. If we further assume that the oil is incompressible we
also have Bo = 1 such that ρo = ρo,sc and qo = qo,sc, and Eq. 3.27 further simplifies to
3.3.2 Solutions. Each of the differential equations Eqs. 3.23, 3.24, 3.27, and 3.28 is
of first order and therefore requires one boundary condition, specifying the pressure p
at a certain value of s:

where we have used a hat above the variables to indicate that their value is
prescribed. The solution to, e.g., Eq. 3.24 can now be expressed as

In the general case, where θ is a function of s, and f is a function of μ and therefore

of T(s), it will usually not be possible to obtain the integral in closed form, in which
case it could be obtained numerically. In particular, the MATLAB file oil_dpds.m
computes the pressure drop for single-phase oil flow. It contains Eqs. 3.24 through
3.26 and can be called on by one of the standard numerical integration routines in
MATLAB (see Section D-2 in Appendix D).
Under some simplifying assumptions the integral can be solved analytically. For
example, consider the case of a pipeline with constant inclination θ transporting dead
oil. If, furthermore, we assume incompressible flow (i.e., a constant ρ) and a constant
temperature along the line, the viscosity also remains constant and therefore the
friction factor f. Alternatively, we could approximate nonconstant inclinations and
friction factors with their average values θav and fav. The integration then becomes
trivial, and the pressure p is a linear function of the measured distance s:

3.4 Single-Phase Gas Flow

3.4.1 Governing Equations. In the case of steady-state single-phase gas flow, we

can rewrite the acceleration term in Eq. 3.18 as

where we used Eq. 3.5 to derive the expression for dρ/dp. The governing set of
equations for steady-state single-phase gas flow through a pipeline now follows by
combining Eqs. 3.5, 3.18, 3.21, and 3.32, modifying them where necessary to be
applicable to gas:

where we also made use of Eq. 2.28 and 2.31 with qo,sc = 0. Just as for single-phase
liquid flow, Eq. 3.33 requires a single boundary condition. A difference with the single-
phase liquid equations (Eqs. 3.24 through 3.26) is the presence of the acceleration
term between square brackets at the right side of Eq. 3.33. The pressure drop dp/ds
approaches infinity when this term approaches zero. This happens when the absolute
value of the velocity approaches

which can be interpreted as the sonic velocity for gas under isothermal conditions.
The ratio |v|/vs is therefore a measure for the importance of acceleration losses.
Generally, they can be neglected. Only in extremely high-rate gas wells, or in a
situation of uncontrolled gas flow, such as a wellbore blowout, might the gas velocity
in a well approach the sonic velocity. However, in that case the flow can no longer be
approximated as isothermal; i.e., temperature effects start to become important, and
a more elaborate analysis is required, which is beyond the scope of this book. Flow
at sonic velocities also plays an important role in the behavior of choke valves, and
that aspect is discussed in more detail in Chapter 5.

3.4.2 Solutions. Eqs. 3.33 through 3.35, together with a boundary condition, can be
solved numerically with the aid of MATLAB (see m-file gas_dpds.m and further details in
Section D-2 in Appendix D). An approximate analytical solution can be obtained by
assuming that the acceleration losses may be neglected, i.e., that |v| << vs and that f,
Tabs , θ, and Z may be taken as constant “average” values fav, Tav,abs , θav, and Zav over
the length of the pipeline. In that case the set of equations reduces to a single
differential equation in p:

where the coefficients k1 and k2 are given by

and where we used the relationship A = πd2/4. Note that k1 should be positive for all
wells, whereas k2 should be positive for producers and negative for injectors. (Watch
out for the difference between α and θ and the associated signs: θ = α − π/2.) Eq.
3.37 can be rewritten as

and therefore as

which has the solution

If we use boundary condition (Eq. 3.29) to solve for the integration constant C, we
arrive at

Usually it will be necessary to perform one or more iterations to obtain the average
values fav, Tav,abs , θav, and Zav because they depend on the unknown pressure p(s). For
horizontal flowlines, θav = 0 and therefore k1 = 0, and thus the solution (Eq. 3.43)
breaks down; but starting from Eq. 3.41, it is easily shown that in that case the
solution becomes

3.5 Tubing Performance and Intake Curves

Referring back to Section 1.5, we can now determine the element equations for
single-phase well or pipeline flow that were already specified in concise form in Eq.
1.4 as
If we choose the boundary condition at one end of the element (i.e., ), we can
compute the pressure pout at the other end through numerical integration of Eqs. 3.24
through 3.26 or Eqs. 3.33 through 3.35, or directly from the analytical expressions
Eq. 3.31 or Eq. 3.43, depending on the assumptions and the required accuracy. The
velocity v appearing in all these expressions can be related to qsc with the aid of Eq.
3.21 or its equivalent for gas, Eq. 3.34, keeping in mind that negative values of v and
q indicate flow in a production situation.
The MATLAB file pipe.m provides an element equation according to expression Eq.
3.45, which can be used to compute the pressure drop over an element of a well or a
flowline. The file can also be used to compute the pressure drop over an entire well
or flowline as long as the geometrical properties of the pipe (diameter, roughness) do
not change along its length. This allows, e.g., for the computation of the flowing
tubinghead pressure (FTHP) ptf of a single-phase liquid or gas well for a known
flowing bottomhole pressure (FBHP) pwf, or, conversely, pwf for a known ptf. Similarly
the file can be used for flowlines to compute the manifold pressure for a given flowline
pressure and vice versa. Examples of how to use the MATLAB file are provided by the
MATLAB script files example_flowline.m and example_well.m.
With the aid of pipe.m it is also possible to create plots of the wellbore pressure p
as a function of AHD s. Such plots are often referred to as traverses, and an example
of how to create them is given in the script file example_traverse.m. Another file,
example_traverse_deviated.m, gives an example of how the traverse of a deviated
well can be obtained using a survey file to specify the well trajectory. Moreover, it is
possible to repeat the computation of the FTHP, at a fixed FBHP, for a large number
of flow rates. This results in a so-called tubing performance curve, which depicts the
FTHP as a function of flow rate. (Confusingly, the term tubing performance curve is
also used for a graph that represents the combined pressure drop over the tubing
and the near-well reservoir area.) Alternatively, one can repeat the computation of the
FBHP, at a fixed FTHP, for varying flow rates. This results in a so-called tubing intake
pressure curve, which depicts the FBHP as a function of flow rate (see the script file
example_intake_curve.m). Shorter names are tubing intake curve or intake pressure
curve.
Figs. 3.4 and 3.5 depict traverses for a dry gas well with parameters given in
Table 3.1. They were both computed for the same FBHP, but with different flow rates
and therefore different FTHPs. Fig. 3.4 corresponds to a relatively low rate, where
gravity losses dominate the pressure drop over the well. Fig. 3.5 corresponds to
twice the rate, and it can be seen that friction losses play a much larger role.
Acceleration losses are of no significance, and only in Fig. 3.5 can they be noticed for
very low pressures close to the surface. Going from the bottom of the well to the
surface, the gravity losses gradually decrease because the density of the gas
decreases, as can clearly be observed in Fig. 3.5. Because the mass flow rate
remains the same, the reduction in density with elevation also causes an increase in
volume flow rate and therefore in gas velocity. The increased velocity, in turn, results
in an increase in friction losses, which is also clearly visible in Fig. 3.5.
 An example of a tubing performance curve has been displayed in Fig. 3.6, where
the FBHP was kept constant at 29.0 MPa (4,206 psi). The figure illustrates that for
decreasing FTHP a point is reached at which the curve becomes vertical (i.e., a point
where no further increase in gas flow rate is possible because the gas velocity has
reached the sonic velocity). Fig. 3.7 displays the tubing intake curves for fixed FTHPs
of 1.5 MPa (218 psi) and 19.2 MPa (2,785 psi).

Fig. 3.4—Traverse for a low-rate single-phase gas well with parameters given in Table 3.1.
Fig. 3.5—Traverse for the same well as in Fig. 3.4 but at a higher rate.

Parameter SI Units Field Units

Tubing diameter, d 62.3 × 10–3 m 2.453 in.
Roughness, e 30 × 10–6 m 0.0012 in.
FBHP, pwf 29.0 × 106 Pa 4,206 psi
FTHP Fig. 3.4, ptf 19.2 × 106 Pa 2,785 psi
FTHP Fig. 3.5, ptf 1.5 × 106 Pa 218 psi
Rate Fig. 3.4, qg,sc −4.31 m 3/s −13.15 × 106 scf/D
Rate Fig. 3.5, qg,sc −8.62 m 3/s −26.27 × 106 scf/D
FBHT, Twf 120°C 248°F
FTHT, Ttf 30°C 86°F
Well depth, ztot 3,000 m 9,843 ft
Inclination, α 0 rad 0°
Density/gravity, ρg,sc/γ g 0.95 kg/m 3 0.77
Viscosity, μg Carr et al. (1954) correlation
Equation of state Standing and Katz (1942) Z factor
Table 3.1—Parameter values for Figs. 3.4 and 3.5.
Fig. 3.6—Tubing performance curve for the same well as in Fig. 3.4. The dot and the triangle correspond to
the low- and high-flow-rate cases as specified in Table 3.1.
Fig. 3.7—Tubing intake curves for the same well as in Fig. 3.4. The dot and the triangle correspond to the
low- and high-flow-rate cases as specified in Table 3.1.

3.6 Flow in an Annular Geometry

An approximate way to compute the pressure drop in an annular geometry (e.g.,
between casing and tubing) makes use of the hydraulic radius concept. The hydraulic
radius rh is defined as the cross-sectional area A, divided by the wetted perimeter Pw,
and it is assumed that conduits with identical hydraulic radii experience identical
pressure drops per unit length. For a circular pipe with inner diameter d the wetted
perimeter for internal flow is simply the perimeter πd, and the hydraulic radius
therefore becomes

For a casing/tubing annulus we find

where dc is the casing’s inner diameter and dt the tubing’s outer diameter. To
represent an annular casing/tubing geometry with an equivalent circular pipe, the
equivalent pipe diameter deq therefore follows from setting rh = rh,ann, resulting in

3.7 Questions

3.1 What is the physical meaning of the Reynolds number?

3.2 What are the dominant pressure drop mechanisms in (a) a vertical oil well and
(b) a horizontal gas pipeline?
3.3 Although we consider only steady-state flow (i.e., flow described with
variables that do not change over time), why may there still be an effect of
acceleration on the pressure drop?
3.4 Consider a gas well that produces at such a high rate that the acceleration
component cannot be completely neglected. How do the friction, gravity, and
acceleration components of the pressure drop in the well change with depth?
Sketch a typical profile. Explain the physical mechanisms.
3.5 Derive modified mass and momentum balance Eqs. 3.3 and 3.4 such that
they are valid for a pipe with a gradually changing cross-sectional area A.
3.6 Single-phase oil is pumped uphill through a pipeline under a 1.5° angle over a
length of 3 km. The oil has a density of 850 kg/m3, the pipeline has an inside
diameter of 232 mm and a roughness of 0.003 mm, the ambient temperature
is 45°C, and the pipeline pressure at the inlet is 10 bar. What is the outlet
pressure for a flow rate qo = −5000 m3/d? Hint: Use the Beggs and Robinson
(1975) dead oil correlation (Eq. B-10) to compute the viscosity and Eq. 3.31 to
compute the pressure drop. Choose the origin at the outlet of the pipeline.
3.7 The gas well of Fig. 3.5 (with parameter values given in Table 3.1) has friction
and Z factors that change only slightly over the height of the well. Their
average values are Zav = 0.96 and fav = 0.0166. Verify the numerical results of
Fig. 3.5 with the aid of the approximate analytical solution described in Section
3.4.
3.8 Fig. 3.8 schematically depicts a situation in which produced water is
reinjected by pumping it down the annulus between two intermediate casing
strings into a deep saltwater aquifer. The top of the perforations in the annulus
is at a depth of 3230 m. The sizes of the intermediate strings are
[i.e., 9⅝ -in diameter and 29.30 lbm/ft nominal weight (including couplings)]
and – 48#. The corresponding relevant outer diameter (OD) and inner
diameter (ID) are 0.2445 and 0.2815 m. The water density is 1070 kg/m3, the
viscosity 0.38 × 103 Pa · s, the temperature 20°C (neglect heating while
flowing down the annulus), and the casing roughness 0.04 mm. If it is required
 to maintain an injection rate of 1200 m3/d at an injection pressure of 40 MPa,
what is the pressure at the wellhead?

Questions 3.9 to 3.12 require the use of MATLAB. You may first want to complete the
MATLAB assignment in Section 3.8 to obtain experience in using the necessary function
files. Some guidance on the use of the required numerical integration routines is given
in Section D-2 of Appendix D.

3.9 Compare the results of Question 3.6 with the results from using the MATLAB m-
file pipe.m. Type help pipe or open the file in the editor to inspect the input
requirements. You may also inspect the file example_flowline.m to get
started.

Fig. 3.8—Produced water reinjection through the annulus between two intermediate casing strings. The
light gray areas represent layers of highly permeable rock. The shaded areas close to the well represent
cement. The thin horizontal lines represent perforations.

3.10 Refer to the last paragraph of Section D-2.2 in Appendix D. Use MATLAB file
example_well.m as a template and write a script file to check the absolute and
relative errors in the FBHP for the example of Fig. 3.5. Use the default
integration tolerance inside pipe.m, i.e., options = []. Repeat the exercise
using a higher integration tolerance, e.g., options = odeset(‘MaxStep’,
10,’RelTol’,1e-3). Do not forget to reset the tolerance after completing the
exercise to avoid a slow response of pipe.m during later use.
3.11 Consider a gas pipeline with four different segments as depicted in Fig. 3.9.
Use a series of four pipe elements, as shown in Fig. 3.10, to compute the
pressure drop over the pipeline for the parameter values given in Table 3.2.
 This can be done by repeated calling of pipe.m such that the input pressure of
an element is equal to the output pressure of the previous element. Choose
the coordinate s along the pipeline such that it runs from the gas plant to the
platform (Note: This is in line with the assumption that production rates are
negative.) Perform the integration backward from s = ltot to s = 0, where ltot is
the total pipeline length. Use a linear interpolation for the temperature. Plot the
pressure in the pipeline p as a function of distance s from the gas plant.

Fig. 3.9—Gas pipeline. All distances in km. Figure not to scale.

Fig. 3.10—Gas pipeline represented as four elements.

Property Value
Flow rate (m 3/d) −1 × 106
Inside diameter (m) 0.30
Roughness (mm) 0.03
Entrance pressure (MPa) 10
Entrance temperature (°C) 50
Exit temperature (°C) 10
Gas density (kg/m 3) 0.95
Table 3.2—Gas pipeline properties for Question 3.11.
 3.12 MATLAB file pipe.m computes the pressure drop over a pipe element for a given
flow rate. If the pressures at both ends of a pipe are given and the flow rate is
the unknown variable, an iterative procedure is required. Use the standard
MATLAB routine fzero to compute the single-phase gas flow rate through a
pipeline for parameters given in Table 3.3. Iterate on qg,sc until ,
where pout is the value as computed with pipe.m and is the value given in
Table 3.3. Consult Appendix D or the online MATLAB help functionality for
information on the use of fzero. Use low and high flow rates of 20 and 100
m3/s, respectively, to bracket the solution. Alternatively, write your own
Newton-Raphson routine to perform the iteration. Use numerical perturbations
to compute the derivatives. See Appendix D for details.

3.8 MATLAB Assignment: Single-Phase Gas Flow

3.8.1 Objectives

• Obtain a feeling for the order of magnitude of single-phase gas flow

parameters.
• Become familiar with gas PVT (pressure/volume/temperature) behavior and the
associated MATLAB functions.
• Learn to make wellbore pressure drop calculations and create a traverse with
MATLAB.
• Learn to use analytical approximations to check numerical results.

Property Value
d 0.40 m
e 50 × 10–6 m
5 × 106 Pa
4 × 106 Pa
Sin 0 m
Sout 8000 m
Tin 25°C
Tout 25°C
α π/2 rad
ρg,sc 0.95 kg/m 3
Table 3.3—Input data for flow rate calculation for Question 3.12.

3.8.2 Assignment. Consider a gas well with the properties listed in Table 3.4.
 Tasks
• Inspect the MATLAB functions pipe.m and gas_dpds.m.
• Copy the file example_well.m and save it under a new name. Use it to compute
the FBHP for the parameters given above. Use the MATLAB unit conversion
routines to convert the input from field to SI units.
• Keeping the FBHP fixed, compute a new FTHP for a rate that is six times
higher than the rate in the example.
• Expand the file to create a pressure traverse for the high flow rate. Refer to
script file example_traverse.m for guidance.
• Study Section B-3 Gas Correlations, in Appendix B. Inspect the MATLAB
functions pres_pseu_crit_Sutton, temp_pseu_crit_Sutton, Z_factor_DAK.m,
Z_factor_DAK_direct.m, gas_viscosity.m, Reynolds_number.m and
Moody_friction_factor.m.
• Expand your MATLAB routine and compute values for Z-factor, formation volume
factor, local density, local flow rate, viscosity, Reynolds number, and friction
factor at the top and the bottom of the tubing for the high-flow-rate case.
• Check the FBHP for the high-flow-rate case with the aid of an analytical
approximation, starting from the (new) FTHP and using average values fav,
Tav,abs , αav, and Zav and a quadratic pressure relationship (see Eqs. 3.37 through
3.44). Use top and bottom values from your numerical results to compute the
averages. The relative error between the analytical and the numerical result
should be approximately 5%.
• Now try to do the opposite; i.e., check the FTHP for the high-flow-rate case
analytically, starting from the FBHP. Why does this fail? (Or if it does not fail
because you made a small error in one of the previous steps, why is there a
good chance that it would fail in general?) Hint: Slightly increase (or decrease)
the starting value of the FBHP.

Parameter Field Units

Inclination α 30 degrees
Diameter d 2.453 in.
Roughness e 0.003 in.
FTHP ptf 2,900 psi
Rate qg,sc −3.0 × 106 scf/D
Gas gravity γg 0.85 –
FBHT Twf 250 °F
FTHT Ttf 115 °F
TVD Stot 12,124 ft
Table 3.4—Parameter values for MATLAB assignment 3.8.2.

Note that in the analytical approximation, the term “k1” should be positive for all wells,
whereas “k2” should be positive for producers and negative for injectors. Watch out
for the difference between α and θ the associated signs (θ = α − π/2).

3.8.3 Deliverables

• MATLAB program listing

• Pressure traverse
• Properties at top and bottom
• Analytical approximation of the FBHP for the high-flow-rate case
• Answer to question under last bullet
Chapter 4

Multiphase Flow in Wells and Pipelines

4.1 What Is Covered in This Chapter?

In this chapter we extend the pressure drop analysis of pipes to multiphase flow of
gas, oil, and water. A key aspect of such multiphase gas/liquid flow is the occurrence
of different flow regimes—i.e., different geometrical configurations of gas and liquid
mixtures such as gas bubbles or slugs in liquid, liquid droplets in gas, or segregated
liquid and gas layers. Another essential multiphase effect is the occurrence of slip
between the liquid and gas phases (i.e., of different liquid and gas velocities where
liquid is typically flowing slower than gas). Slip causes the pipe volume to be occupied
by a disproportionally large liquid fraction—an effect known as liquid holdup—which
may result in counterintuitive effects on the pressure drop in multiphase production
wells.
Quantitative models for the pressure drop resulting from multiphase flow through
pipes have been developed over the past half century in varying levels of detail. They
range from simple semiempirical correlations to complex mechanistic models that
take into account the details of the interaction between the gas and liquid phases for
all possible flow regimes. In Appendix E, we describe a few typical multiphase flow
models of medium complexity, which also have been programmed in the
accompanying MATLAB code. These enable the creation of traverses and performance
curves for use in nodal analysis of multiphase flow through (deviated) wells and
pipelines.

4.2 Multiphase Flow Concepts

4.2.1 Flow Regimes. The occurrence of radically different flow regimes (or flow
patterns) in multiphase (gas/liquid) flow depends on factors such as the gas/liquid
ratio, the gas and liquid velocities, the pipe inclination, and fluid properties. The flow
regimes for gas/liquid flow in horizontal pipelines are shown in Fig. 4.1. They are
generally known as follows, although some authors use a classification with more
categories:

• Single-phase liquid flow

• Bubble flow
• Slug flow
• Stratified flow
• Annular flow
 • Mist flow (fully dispersed liquid mist)

Fig. 4.1—Flow regimes in horizontal flow.

For vertical flow in a well, a similar flow pattern classification can be made. This is
shown in Fig. 4.2. The flow regimes are the same as those in horizontal flow except
for the absence of stratified flow and the occurrence of churn flow as an intermediate
regime between slug and annular flow. Furthermore, the slug flow regime is now
somewhat different, displaying bullet-shaped slugs known as Taylor bubbles, which
remain more or less centered in the wellbore. To describe flow in real pipelines or
wells, the inclination of the pipe has to be taken into account to give a full map of
multiphase effects. For a treatment of omniangle flow maps based on physical
principles, see, e.g., Hassan and Kabir (2002) or Shoham (2006). Several simpler,
but usually less accurate, approaches based on empirical correlations are discussed,
e.g., in Wallis (1969), Brill and Mukherjee (1999), and Govier and Aziz (2008).
In a vertical oil well, the pressure decreases as the oil flows from the bottom to the
top of the well. Thus, all the flow patterns shown in Fig. 4.2 may arise. Generally,
however, over most of their length most oil wells operate in the bubble flow and slug
flow regimes, whereas most gas wells operate in the annular flow regime. It is a
formidable task to try to solve the equations based on the laws of physics, which
govern these types of flow. There are numerical simulators that attempt this for
sensitive industrial processes that need very careful modeling. Within the oil industry,
a simpler approach is often adopted. Empirical correlations have been developed
based on extensive experiments. Some of these correlations have been published;
others remain proprietary to oil companies or service companies. These correlations
differ in complexity. Some are proposed as valid for all flow regimes, whereas others
have separate correlations for each different regime. Some methods try to include
basic physics, such as modeling the behavior of gas/liquid interfaces, whereas others
rely on a purely empirical approach. For an overview we refer again to Brill and
Mukherjee (1999) and for more in-depth treatments to Hassan and Kabir (2002),
Shoham (2006), and Govier and Aziz (2008). Many of these correlations are usually
built into modern well simulators. Care needs to be taken because correlations are
often suitable for only certain types of wells. Note that the correlations used for the oil
properties will affect the results and may contribute to the inaccuracy.

Fig. 4.2—Flow regimes in vertical flow.

4.2.2 Slip and Holdup. One of the complicating factors in the description of
multiphase flow is the difference in velocity between the phases. It is generally
assumed that water and oil travel at the same speed, known as the liquid velocity,
although in reality this is not always the case, in particular for stratified flow. However,
most computational methods do account for the difference between the liquid velocity
and the gas velocity, which is known as slip between the two phases. But before
considering phase velocities, it is useful to address phase flow rates. In the case of
two-phase liquid flow, and assuming that the liquids flow with equal velocities, the
local oil and water fractions are defined as
Here the word “local” refers to local pressure and temperature conditions (also known
as in-situ conditions). However, this does not imply that “local” refers to a very small
length scale—i.e., we are not interested in fluctuations in flow rates or velocities as a
result of small-scale flow features such as slugs or bubbles. Instead, all quantities
should be interpreted as averaged over a distance much larger than the small-scale
features (i.e., on the order of meters). Note that from Eqs. 4.1 and 4.2 it follows that
fo + fw = 1. The gas volume fraction and the liquid volume fraction are defined in the
same fashion, although we will indicate them with a λ instead of an f:

where the quantity qm = qg + gl = qg + qo + qw is known as the mixture flow rate or

total flow rate. Note that also λg + λl = 1. Because of slip the fractions of a unit
volume of pipe that are occupied by gas and liquid are generally not equal to the gas
and liquid fractions as given in Eqs. 4.3 and 4.4. In upward flow, as occurs in a
production well or an uphill pipeline, the gas usually travels faster than the liquid, and
so liquid holdup occurs. In downward flow, as occurs in a downhill pipeline, the liquid
may travel faster than the gas, in which case the gas is held up. The expression
“holdup” is also often used in the oil industry to indicate the volume fractions occupied
by gas and liquid, although in upward flow the gas is not actually held up, but to the
contrary is speeded up. The gas and liquid holdups Hg and Hl are defined as

where Vg and Vl are the fractions of a reference volume of pipe that are being
occupied by gas and liquid and V = Vg + Vl is the total reference volume. Similarly, Ag
and Al are the parts of the pipe’s cross-sectional area occupied by the gas and the
liquid, respectively, and A = Ag + Al is the total cross-sectional area. Note that
volumes and areas should be interpreted as quantities averaged over a length that is
sufficiently large to suppress the effect of small-scale flow features. Just as was the
case for the sum of phase volume fractions, the sum of phase holdups is equal to
unity. An alternative way to express the equations for phase volume fractions (Eqs.
4.3 and 4.4) and phase holdups (Eqs. 4.5 and 4.6) makes use of variables known as
the local, in-situ, or true phase velocities
the superficial phase velocities

and the mixture velocity

Substitution of these expressions in Eqs. 4.3 and 4.4 results in

and substitution in Eqs. 4.5 and 4.6 gives

Most computational methods for multiphase flow simulation make use of

experimental correlations for the liquid holdup expressed as functions of fluid
properties, flow rates, pipe diameter, and well inclination. Eqs. 4.9, 4.10, 4.14, and
4.15 can then be used to compute the gas and liquid velocities for given flow rates
according to

If there is no slip, the local phase velocities vg and vl are both identical to the mixture
velocity vm and therefore the holdups as expressed in Eqs. 4.14 and 4.15 become
identical to the phase volume fractions as expressed in Eqs. 4.12 and 4.13. Other
names for phase volume fraction are therefore no-slip holdup or no-slip volume
fraction. Alternatively, the expressions “phase content” or “input phase fractions” are
used in some publications to identify what we call phase volume fractions. Another
name for holdup is, somewhat confusingly, in-situ volume fraction, and for gas also
the term void fraction is found. In analogy to porous-media flow the term saturation
could also be applied. Nonetheless, we will stick to the oil industry convention and
speak of gas and liquid holdups. Other multiphase flow concepts used in literature are
slip velocity defined as

and gas and liquid mass fractions xg and xl, defined as

where and are the gas, oil, and water mass flow rates and where xg is also
known as the quality of the gas/liquid mixture.
To illustrate the effect of slip on the liquid (volume) fraction and the liquid holdup,
Fig. 4.3 gives an example of stratified flow where the liquid flow rate equals one-third
of the gas flow rate. In case of no slip between the phases, the liquid holdup is equal
to the liquid fraction, and 25% of the pipe’s cross-sectional area is occupied by liquid.
However, if the gas flows twice as fast as the liquid, the liquid fraction remains the
same but the liquid holdup increases such that 40% of the area is occupied by liquid.

4.3 Pressure Drop Analysis

4.3.1 Governing Equations. For single-phase flow we derived three governing

equations: a mass balance, a momentum balance, and an equation of state (EOS)
(see Eqs. 3.3 through 3.6 in Section 3.1). Therefore, it is to be expected that we can
derive six equations for the two-phase flow case: two mass balances, two momentum
balances, and two EOSs. Sometimes this approach can indeed be followed, in
particular for stratified flow, in which case the interface between gas and liquid is
relatively simple and the corresponding interface forces can be modeled in detail.
Usually, however, the interaction between gas and liquid is too complex to be taken
into account explicitly, and so a different approach is followed. Instead of modeling
the dynamics of the two phases individually, mass and momentum balances are
derived for the gas/liquid mixture, and two additional closure equations are specified
for the relationships between the mixture velocity and the phase velocities, and the
mixture density and the phase densities, respectively. The latter two equations are
usually empirical or semiempirical. The governing dynamic equations for two-phase
flow then become
Fig. 4.3—Illustration of the effect of slip between the gas and liquid phases on the liquid fraction λl (no
effect) and the liquid holdup Hl (increases for increasing slip velocity).

Here the EOS Eqs. 4.25 and 4.26 have been expressed in terms of volatile oil
properties with the aid of Eqs. 2.21 through 2.23. Moreover, we use a liquid-mixing
rule

where the weight factors fo and fw are the volume fractions defined in Eqs. 4.1 and
4.2. Just as for single-phase flow we can simplify the analysis considerably by
considering only steady-state flow, in which case Eqs. 4.21 and 4.22 can be
combined to give
where we have used the definitions of head and friction losses that were introduced in
Sections 3.2.2 and 3.2.3, but are now expressed in terms of mixture velocity vm and
mixture density ρm . Several slightly different formulations have been presented by
various authors, but they all contain a head loss, friction loss, and acceleration loss
component. The equation for the mixture density (Eq. 4.23) can in the simplest case
be written as an average of the phase densities weighted by the phase volume
fractions:

where the subscript mn indicates “mixture, no-slip” and reflects that this expression
assumes that there is no slip between the phases. The corresponding expression for
the mixture velocity (Eq. 4.24) then becomes

with vl = vo = vw; but because of the absence of slip, we also have

Normally, however, we assume that slip occurs, in which case Eq. 4.23 can be
written as

where the subscript ms indicates “mixture, slip” and where the gas and liquid holdups
are (semi)empirical functions of a large number of parameters such as inclination,
pipe diameter, flow rates, densities, viscosities, and surface tensions. The
corresponding equation for the mixture velocity is

as can be verified from Eqs. 4.11, 4.14 and 4.15. In analogy to Eqs. 4.29 and 4.32
we can also define no-slip and slip averaged versions for other fluid properties. For
example, for the mixture viscosity we obtain

with μl expressed in terms of μo and μw with the no-slip liquid-mixing rule

4.3.2 Holdup and Friction Correlations. Many empirical correlations and
semiempirical models have been developed for multiphase flow in pipes. Several of
these concern two-phase steam/water flow, especially for the nuclear industry.
Others are primarily meant for chemical engineering applications, while also a good
number are dedicated to the flow of hydrocarbons in wells and pipelines. We refer to
Brill and Mukherjee (1999), Hassan and Kabir (2002), and Shoham (2006) for
detailed overviews. In this book we present only three correlations in detail (in
Appendix E) as examples of the typical multiphase flow correlations. Brill and
Mukherjee (1999) distinguish three types of correlations of increasing complexity, in
addition to even more complex mechanistic models. The latter involve varying
amounts of first-principle physics but are never free from some form of empiricism,
which explains why they are also referred to as semiempirical models. The simplest
group of correlations consists of empirical expressions for the friction factor in terms
of no-slip mixture properties. The second category takes slip into account and
consists of empirical expressions for holdup and friction, while the third category also
considers different flow regimes. The latter two categories typically use a set of
dimensionless groups defined in terms of the essential variables to describe the
physics of multiphase flow.
In the first systematic dimensional analysis of multiphase flow in oil wells (Ros
1961), 13 variables were identified to describe two-phase flow in near-vertical pipes:

• pipe diameter d [L]

• pipe roughness e [L]
• acceleration of gravity g [L t−2]
• pressure p [L−1 m t −2]
• gas velocity vg [L t −1]
• liquid velocity vl [L t −1]
• well inclination α [-]
• pipe wall–gas wetting angle φ [-]
• gas viscosity μg [L −1 m t −1]
• liquid viscosity μl [L −1 m t −1]
• gas density ρg [L −3 m]
• liquid density ρl [L −3 m]
• liquid/gas interfacial tension σgl [m t −2]

Because these 13 variables are defined in terms of three dimensions (length L, mass
M, and time t), it is possible, according to the theory of dimensional analysis, to
derive a total of 13 – 3 = 10 dimensionless groups to completely describe the physics
of the problem. Note that the theory of dimensional analysis is described in most
textbooks on fluid mechanics—see, e.g. White (2016). For a discussion focused on
hydrocarbon pipe flow, see Shoham (2006).
 Ros (1961) chose to study the dimensionless pressure gradient (dp/ds)/(ρl g) as a
function of nine other dimensionless groups. Further analysis revealed that of those
nine, the following four were the most important:

Among the five groups that were determined to be less important was the wellbore
inclination α, because in those days nearly all oil wells were vertical. Clearly, for
deviated or horizontal wells the inclination must play an important role. Duns and Ros
(1963) used the dimensionless groups (Eqs. 4.37 through 4.40) to design an
experimental program that resulted in a set of early popular multiphase flow
correlations. Since then, other dimensionless groups have been defined to describe
multiphase flow correlations for oil and gas wells, but we will not discuss these and
refer instead to the references mentioned above for further information. That also
holds for the mechanistic models, which are of a complexity outside the scope of this
book.
In Appendix E we discuss three correlations. The first one, from Hagedorn and
Brown (1965), is an empirical correlation of the second category; i.e., it gives friction
and holdup values. It is valid only for near-vertical wells. The second one is the
Mukherjee and Brill (1985b) model. It is a correlation of the third category (i.e., it
takes into account flow regimes) and is valid for vertical, deviated, or horizontal wells
or pipes. Both these models are defined in terms of the dimensionless groups Eqs.
4.37 through 4.40, as well as, for the Mukherjee and Brill model, the wellbore
inclination α. The third model discussed in Appendix E is a so-called drift flux model,
which uses a particular approach to compute the holdup. The version discussed is a
second-category correlation; i.e., it takes into account slip but no distinct flow
regimes. It is a popular method for simplified multiphase flow computations inside
reservoir simulators because it is fast and results in smooth gradients without the
discontinuities that are often present in more complex models and may cause
numerical problems. The Hagedorn and Brown, and Mukherjee and Brill correlations
have been programmed in the MATLAB function files Hag_Brown_dpds.m and
Muk_Brill_dpds.m. Programming of the drift flux method is left as a MATLAB
assignment in Section 4.7. Another popular third-category correlation, by Beggs and
Brill (1973), is available in Beggs_Brill_dpds.m.
4.3.3 Element Equations. Just as in the single-phase case we can determine
element equations for multiphase well or pipeline flow, which can be written concisely
in the form of Eq. 1.4 as

or alternatively as

If we choose the boundary condition at one end of the element, (i.e., ), we

can compute the pressure pout at the other end through numerical integration with the
aid of the multiphase functions Hag_Brown_dpds.m, Muk_Brill_dpds.m, or
Beggs_Brill_dpds.m in conjunction with a standard MATLAB ordinary-differential
equation solver, such as ode4 5.m. This procedure has been implemented in the file
pipe.m., and examples of how to use these files can be found in
example_flowline.m, example_well.m, and example_traverse.m.
Fig. 4.4 displays the traverse of a vertical oil well with properties given in Table
4.1. The figure was created by fixing the flowing tubinghead pressure (FTHP) at 0.5
MPa and integrating from top to bottom, resulting in a flowing bottomhole pressure
(FBHP) of 23.9 MPa. Subsequently fixing the FBHP and integrating from bottom to
top results in Fig. 4.5. Both figures clearly show that the majority of the pressure
drop is caused by head losses, while friction losses play a much smaller role.
Acceleration losses can be completely neglected.
Fig. 4.6 displays the results for a deviated well with the same properties given in
Table 4.1 except for the FTHP, now taken as 5.4 MPa, and the well depth and the
inclination, now determined from survey file deviated_well_1.txt which was also
used in Appendix C on wellbore surveying. Fig. C-4, in Appendix C, illustrates that this
well has a horizontal section, which explains that in Fig. 4.6 the head losses between
measured depths of approximately 2700 and 4600 m are zero.

4.3.4 Gradient Curves. Before modern computers, the practice was to present
empirical correlations for wellbore pressure drop in the form of gradient curves,
which are just traverses for a given set of parameters. An example is given in Fig.
4.7, which was generated with the aid of the Mukherjee and Brill correlation for the
parameter values indicated in the chart. Although these curves are hardly used now,
they give some insight into the effects of the various parameters. After introduction of
the gradient curve concept by Gilbert (1954), large numbers of these curves have
been prepared for “typical” well parameters (see, e.g., Beggs 1991). They are valid
only for vertical wells. Their vertical axis represents depth measured from surface and
the horizontal axis the wellbore pressure computed for the minimum possible (i.e.,
atmospheric) FTHP. To use the curves for higher FTHPs, the vertical axis can be
interpreted as the difference in vertical depth between two points in the wellbore
rather than the absolute depth; and the horizontal axis, the corresponding pressure
difference (see Exercise 4.9).

Fig. 4.4—Traverse for a vertical multiphase well with parameters given in Table 4.1. Integration from top to
bottom.

4.4 Tubing Intake Curves

As discussed in Section 1.5, the multiphase flow equations for a wellbore element
specify a relation between the oil and gas flow rates qo,sc and qg,sc and the wellbore
pressure drop Δp = pin − pout (see Eq. 1.7). If we know the gas/oil ratio (GOR) Rgo,
we can also determine the flow rates from the pressure drop, although we need to do
so in an iterative fashion. However, usually it is one of the pressures that is unknown.
If the FBHP pwf is specified, then for a given GOR the FTHP ptf can be computed as a
function of qo,sc, resulting in a tubing performance curve (see also Section 3.5).
Alternatively, ptf may be specified, and pwf computed as a function of qo,sc, leading to a
tubing intake curve. Fig. 4.8 depicts a typical tubing intake curve for an oil well that
has been produced using the data listed in Table 4.1. In comparison to the tubing
intake curve for a gas well depicted in Fig. 3.7, there is a marked difference.
Whereas the gas curve is monotonously increasing with increasing flow rates, the
multiphase curve first decreases before going up. At first sight the gas curve may
seem to be right and the multiphase curve wrong, because increasing the flow rate
through a tubing requires an increase in pressure drop (i.e., an increase in FBHP for a
given FTHP).

Parameter SI Units Field Units

Tubing diameter d 62.3 × 10−3 m 2.453 in.
Roughness e 30 × 10−6 m 0.0012 in.
Water cut fw 0.2 – 0.2 –
FTHP ptf 0.5 × 106 Pa 72.5 psi
Gas/oil ratio Rgo 50 m 3/m 3 281 scf/STB
FTHT Ttf 30 °C 86 °F
FBHT Twf 120 °C 248 °F
Well depth Ztot 3,000 m 9,843 ft
Inclination α 0 rad 0 degree
Gas viscosity μg Carr et al. (1954) correlation
Oil viscosity μo Beggs and Robinson (1975), Vasquez and Beggs (1980)
Water viscosity μw 0.35 × 10−3 Pa s 0.35 cp
Gas density/gravity ρg,sc/γ g 0.95 kg/m 3 0.77 –
Oil density/gravity ρo,sc/γ API 850 kg/m 3 35 °API
Water density ρw,sc 1,050 kg/m 3 65.5 lbm/ft3
Gas/oil interf. tension σgo 0.008 N/m 8 dynes/cm
Gas/w. interf. tension σgw 0.04 N/m 40 dynes/cm
Equations of state Standing (1952), black oil; Standing and Katz (1942), Z factor
Multiphase flow correlation Mukherjee and Brill (1985b)
Table 4.1—Parameter values for Figs. 4.4 to 4.6 and 4.8 to 4.11 (except for deviations as indicated in the
figure captions).

To understand what causes the shape of the multiphase curve in Fig. 4.8, consider
Fig. 4.9, which is identical to Fig. 4.8 except for a split of the total pressure drop over
the tubing into two components: the head loss and the friction loss. It can be seen
that the friction loss behaves as expected; i.e., it increases with increasing flow rates.
However, the head loss shows a more complex behavior. For increasing flow rates, to
the right of the minimum of the curve, the head loss also increases. This is a result of
the increased compression of the gas/liquid mixture in the well at higher FBHPs
corresponding to higher flow rates. For decreasing flow rates, to the left of the
minimum, another effect takes place. The gas increasingly starts slipping through the
liquid, resulting in an increasing liquid holdup and therefore an increasing density of
the gas/liquid column.
Other types of tubing intake curves are obtained by fixing ptf and qo,sc, but varying
one of the process variables, such as the GOR or the tubing size. Some typical
examples of tubing intake curves have been displayed in Figs. 4.10 and 4.11. Fig.
4.10 shows the tubing intake curve that is generated when the oil production rate qo,sc
is held constant while the GOR Rgo is varied. At zero GOR, the well is producing only
liquid. Since the production rate is low, the friction is low, and the intake pressure is
close to the hydrostatic pressure of the fluid column. If gas is introduced, the liquid
column gets lighter, the hydrostatic pressure decreases, and hence the intake
pressure decreases. This effect continues as the GOR increases, but at the same
time the friction pressure drop increases because of the increased total mass flow of
oil and gas. At a certain point, the friction pressure drop starts to dominate, and then
the intake pressure starts to increase again. Apparently, for a well flowing a given
volume of liquid, there exists an optimum GOR at which the pressure drop over the
tubing is minimal. This effect plays an important role in gas lift optimization. In Fig.
4.11, where the numbers 1 to 5 correspond to tubing sizes given in Table 4.2, it can
be seen that there exists an optimum tubing size (for a fixed production rate). As
expected, to the left of this optimum, the pressure increases as the tubing diameter
decreases, given that it is more difficult for fluids to flow through a narrower tube. To
the right of the optimum, multiphase effects start to play a role. In a wider tubing it
becomes easier for the gas to slip past the liquid and the lifting is less efficient. The
downhole pressure required to maintain the flow rate therefore rises.

Fig. 4.5—Traverse for a vertical multiphase well with parameters given in Table 4.1. Integration from
bottom to top.

Fig. 4.6—Traverse for a deviated multiphase well with parameters given in Table 4.1, a FTHP of 5.4 MPa,
and survey file deviated_well_1.txt. Integration from top to bottom.
Fig. 4.7—Example of a gradient curve.
Fig. 4.8—Tubing intake curve for varying production rate.
Fig. 4.9—Tubing intake curve for varying production rate with contributions of head loss and friction loss.
Fig. 4.10—Tubing intake curve for varying GORs at a fixed flow rate of qo,sc = −4 × 10−3 m3/s (2,174 STB/D).
Fig. 4.11—Tubing intake curve for varying tubing inside diameters (ID) at a fixed flow rate of qo,sc = −1.5 ×
10−3 m3/s (815 STB/D) and a fixed GOR of Rgo = 50 m3/m3 (281 scf/STB).

4.5 Lift Tables

In numerical reservoir simulation it is necessary to specify the operating conditions of
the wells. This can be done in terms of total flow rates qt or FBHPs pwf. In reality,
however, it is usually not the FBHP that is controlled but the FTHP ptf. As discussed
above, the FTHP can be computed for given fluid properties, wellbore geometry,
phase flow rates; and FBHP. Conversely, the FBHP may be computed for a given
FTHP. The computation could be performed with the aid of a wellbore simulator.
However, these computations are generally too time consuming to perform every time
step of the reservoir simulator. Only for long horizontal wells or for complex well
configurations such as multilaterals, the use of a wellbore simulator may be
unavoidable, at least for the bottom part of the well. An alternative approach is to
perform a large number of wellbore flow simulations upfront to generate a
multidimensional table, known as a lift table or flow performance table, which can be
used as a look-up table by the reservoir simulator. Usually the four entries for a lift
table are the FTHP and the oil, gas, and water rates, all expressed at standard
conditions. (Note that the lift table entries can also be chosen as FTHP, oil rate,
gas/oil ratio, and water cut. Whatever the choice of the table entries, it is assumed
that the fluid properties at standard conditions and the wellbore geometry do not
change during the reservoir simulation.)
Typically, each of the entries is described with a small number of points, say, 5, in
which case the table has 54 = 625 points that correspond to the same number of
FBHPs. For intermediate values of the entries a linear or higher-order interpolation is
used to compute the corresponding FBHPs, which is much faster than performing a
full wellbore flow simulation. Sometimes a higher number of points is needed, at the
cost of a longer preprocessing time—e.g., to prevent convergence problems during
the numerical solution of the reservoir equations. Lift tables for the commercial
reservoir simulators Eclipse and Imex can be generated with MATLAB files
Eclipse_lift_table.m and Imex_lift_table.m. A typical aspect of lift tables is that
they are usually “regularized” such that there is no decreasing FBHP for increasing oil
flow rates; i.e., the corresponding tubing intake curves have no descending branch.
This is done to avoid numerical problems in the reservoir simulator, which usually
cannot cope with decreasing wellbore pressure drops for increasing flow rates. The
mechanism for regularization is to simply replace the FBHP values in the descending
branch of a pressure intake curve (i.e., the values to the left of the minimum) by the
minimum value.

Number OD (in.) OD (mm) Nominal Weight (lbm/ft) ID (mm)

1 2.375 60.3 4.70 50.67
2 2.875 73.0 6.50 62.00
3 3.500 88.9 9.30 76.00
4 4.000 101.6 11.00 88.29
5 4.500 114.3 12.75 100.53
Note: In the oil field literature, tubing dimensions are often referred to as e.g. 2-7/8” 6.5# tubing, meaning tubing with
an outside diameter of 2.865 in., and a nominal weight per unit length of 6.5 lbm/ft, where “nominal” means
averaged over the length, including the thicker couplings. For a given outside diameter, the inside diameter is
therefore a function of the nominal weigth.
Table 4.2—Tubing data for Fig. 4.11.

4.6 Questions

4.1 Is the liquid holdup in a production well typically larger or smaller than the
liquid fraction? Why?
4.2 What is the difference between superficial phase velocities and in-situ phase
velocities?
4.3 One-dimensional single-phase compressible flow with a given temperature
distribution can be described with three equations: a mass balance, a
momentum balance, and an EOS. Therefore it is to be expected that two-
phase flow can be described with six equations: two mass balances, two
 momentum balances, and two EOSs. Why is this approach usually not
possible? What is the usual alternative approach?
4.4 Why have the multiphase flow correlations described in Appendix E only a
limited predictive value for real oil wells?
4.5 A tubing intake curve for single-phase flow displays a monotonically increasing
relationship between flow rate and FBHP. This is intuitively correct: An
increase in flow rate, at a given FTHP, requires an increase in FBHP. However,
a tubing intake curve for multiphase flow typically displays a minimum. At low
flow rates, an increase in rate corresponds to a decrease in FBHP. Explain this
seemingly counterintuitive behavior.
4.6 Consider a gas/liquid mixture with ql = 0.3 qg and vg = 1.2 vl. What are the
liquid fraction and the liquid holdup?
4.7 A well is completed with a 0.122-m-ID tubing and produces a gas/oil/water
mixture with the following properties: qo,sc = 18.4 × 10−3 m3/s, Rgo = 238 m3/m3,
fw,sc = 0.23. You know that you are dealing with a black oil and that at a depth
of 400 m the formation volume factors and the solution GOR are given by Bg =
0.05, Bo = 1.15, Bw = 1.00, and Rs = 10.1, all expressed in m3/m3.
Furthermore, you know that the liquid holdup is 5% higher than the liquid
volume fraction. What are the superficial and local gas and liquid velocities?
4.8 Consider a deviated oil well with a trajectory as depicted in Fig. 4.12. The
well has a constant azimuth; and the “stepout,” displayed on the horizontal
axis, is therefore just the horizontal distance from the wellhead shor. Fig. 4.13
depicts the acceleration, friction, and gravity components of the pressure drop
over the well.
Fig. 4.12—Well trajectory for Question 4.8.
Fig. 4.13—Pressure drop components for Question 4.8.

(a) Indicate which of the curves correspond to which of the pressure drop
components.
(b) What is the magnitude of the FBHP? Sketch the total pressure drop.
(c) Consider the same well, producing oil with a much higher solution GOR.
Give a qualitative sketch of the corresponding traverses for the pressure
drop and its components (i.e., four curves) if the FBHP remains the
same. Hint: Pay attention to the shape of the friction and gravity curves
close to the surface.
(d) Which of the following methods could be used to compute the pressure
drop over this well: (a) Duns and Ros, (b) Hagedorn and Brown, (c) Brill
and Mukherjee, (d) Shi et al. (drift flux)?
4.9 A 3000-m deep vertical oil well is producing at a FTHP of 5 MPa with a GOR
of 50 m3/m3. Use the gradient curve in Fig. 4.7 to compute the FBHP.

Questions 4.10 and 4.11 require the use of MATLAB.

4.10 For the same well as in question 4.9 use MATLAB file pipe.m to compute the
FBHP. Use the operating conditions listed in Fig. 4.7, a tubing ID d = 0.10053
m, and a pipe roughness e = 30 × 10−6 m.
 4.11 Run MATLAB file example_intake_curve.m and verify that the results are
identical to those of Fig. 4.8, although in a slightly different layout.
(a) Rerun the file a few times with an increased FTHP—e.g., choose values
of 1, 3, and 5 MPa. (This is simply done by replacing the values for ptf in
line 32.) What do you observe? Explain the result. Reset the FTHP to 0.5
MPa.
(b) Make a copy of the file under a different name. Modify the copied file to
produce a tubing performance curve. Leave all well data unchanged. Run
the modified file for an FBHP of 26 MPa. Compare the curves to the
tubing performance curve for single-phase gas flow in Fig. 3.6. Explain
the difference in shape. Repeat the exercise for increasing values of the
FBHP—e.g., 28 and 30 MPa. What do you observe? Explain the result.

To answer questions 4.12 through 4.14 you should first study the multiphase flow
correlations in Appendix E. The correlations have been programmed in various MATLAB
routines (see Appendix H). However, the questions can be answered without
performing MATLAB computations.

4.12 For a vertical oil well we have Nlv = 17.6, Ngv = 237, Nd = 45.4, and Νμ =
0.131. What is the liquid holdup according to the Hagedorn-Brown correlation
at the point in the well where the pressure equals 15 MPa? (No need to check
if Hl > λl.)
4.13 Fig. 4.14 displays part of the flow chart for the Mukherjee and Brill multiphase
flow algorithm (see Appendix E). Enter the following text in the corresponding
boxes in the flow chart:

• uphill flow
• horizontal flow
• gas/liquid flow
• liquid-only flow
• injection well
• production well
• interpolate T
• compute vm
Fig. 4.14—Partial flow chart for Mukherjee and Brill algorithm.

4.14 Consider the following equation:

(a) What does this equation describe? (Hint: Check Appendix E.)
(b) What can you say about the velocity and concentration profiles if C0 is
equal to unity?
(c) Why is the ratio vms/vg,fld typically larger than unity for fully developed
annular flow?

4.7 MATLAB Assignment: Drift Flux

4.7.1 Objectives

• Gain insight into the typical contents of a multiphase flow correlation

• Learn to implement such a correlation in MATLAB
• Compare the results to those of another correlation to obtain a feeling for the
order of magnitude of the differences
4.7.2 Warning. This MATLAB exercise requires considerably more programming skills
than the MATLAB exercises in earlier chapters.

4.7.3 Assignment. The MATLAB function pipe.m has the option to call various
subfunctions to compute the pressure drop dp/ds over an inclined wellbore or
flowline. These include the multiphase flow models Hag_Brown_dpds.m,
Muk_Brill_dpds.m, and Beggs_Brill_dpds.m. The user can choose between these
submodels by specifying the value of the parameter fluid in one of the routines
calling pipe.m (see, e.g., the calling routine example_traverse.m). Program a new
correlation drift_flux_dpds.m and compare it against an existing routine.

Tasks

• Study the paper by Shi et al. (2005b) and Section E-3 (Drift Flux Models) in
Appendix E.
• Inspect the MATLAB functions pipe.m and Muk_Brill_dpds.m.
• Write a routine drift_flux_dpds.m to compute the pressure drop over an
inclined well using the Shi et al. drift flux correlation.
• Test the routine by comparing its results against those of Muk_Brill_dpds.m
when called by example_intake_curve.m for a well with a 60° inclination. Do not
change the other parameters. Verify the source of the difference in FBHPs by
plotting a traverse for the flow rate that corresponds to the largest difference
(near qo,sc = –0.001 m3/s). For this purpose modify example_traverse.m as
appropriate.

4.7.5 Notes

• The gas holdup has to be computed iteratively (see also Eqs. E-45 and E-46 in
Appendix E). Use Picard iteration and terminate it at a preset relative tolerance
(e.g., 10−3). Include an exit strategy to terminate the iteration if the computation
does not show convergence within a preset number of iterations. You may need
a damping factor in the iteration to avoid oscillations (see Eq. E-46).
• The Shi et al. correlation uses a different definition of the positive flow direction
than this book and the existing MATLAB files. Your file should be able to cope with
production in an uphill direction. There is no need to consider production in a
downhill direction or injection.
• For the pressure drop calculation, copy the code from Muk_Bril_dpds.m.

4.7.6 Deliverables

• MATLAB program listing

• A plot with traverses (with detailed contributions from gravity and friction) and a
plot with intake curves, each displaying the results for Muk_Brill_dpds.m and
drift_flux_dpds.m
Chapter 5

Flow Through Restrictions

5.1 What Is Covered in This Chapter?

In this chapter we discuss the steady-state pressure drop over restrictions to flow, in
particular those that are purposely designed to reduce the pressure in a part of the
production system. For single-phase liquid flow it is still possible to describe the
essential behavior by starting from mass and momentum balance equations and an
isothermal equation of state just as in pipe flow. However, unlike in pipe flow,
acceleration losses turn out to play a major role in the pressure drop, in particular in
the converging part of a restriction.
For single-phase gas flow, the assumption of isothermal flow is no longer valid,
and a more elaborate analysis is required to compute the pressure drop. Here we
use the classic assumption of isentropic (frictionless-adiabatic) flow to compute the
acceleration-dominated flow in the converging part. An essential result is that for an
increasing pressure drop the gas velocity may become so high that it reaches the
speed of sound in gas, a situation known as choked flow or critical flow. As a result,
the pressure downstream of such a choked restriction is no longer influencing the
pressure drop, a decoupling that has important practical implications and also causes
theoretical restrictions in nodal analysis.
For multiphase (gas/liquid) flow we develop a description analogous to the one for
single-phase gas, using the assumption of polytropic (between isothermal and
isentropic) flow. The resulting analytical expression for the pressure drop over a
restriction, as well as the expression for the critical pressure drop resulting in choked
flow, are now implicit and need to be solved iteratively. This is most easily done
numerically. Alternatively, simple semiempirical expressions may be used. Both
approaches result in (strongly) approximate results which are, however, good enough
for typical nodal analysis calculations for production engineering purposes.

5.2 Restrictions
At several points in the production system the flow may encounter restrictions, such
as valves or measurement devices. Usually the pressure drop in these components is
not desirable and deteriorates the performance of a production system, in particular
when a further reduction of flow area occurs because of deposition of solids such as
scale or asphaltenes. Sometimes, however, purposely designed restrictions are used
to create a pressure drop or to restrict the flow rate in a controlled fashion. These
devices, known as choke valves, chokes, or beans, occur in different forms. Fixed-
size chokes require a temporary interruption of the production stream if they need to
be changed out for a choke of a different size. Variable-size chokes can be operated
under flowing conditions and may be manually operated or remotely controlled. All
chokes work on the same principle of dissipating large amounts of energy over a
short distance.
Another reason to restrict the flow on purpose is to measure flow rates: A
restriction causes a local reduction in pressure that, for a given geometry of the
restriction, is a known function of the flow rate. Measuring the pressure drop, which is
relatively simple, can thus be used to indirectly measure the flow rate. Fig. 5.1
displays three schematic restriction geometries. In the bottom geometry the flowlines
converge between cross section 1, the entrance, and cross section 3, the throat;
thereafter they diverge again until an undisturbed flowline pattern is re-established at
cross section 4, the exit. The figure is not to scale, and generally it takes a distance
of at least 10 times the diameter reduction (d1 – d2) before the effects of the
restriction on the flowline pattern can no longer be noticed. The top of Fig. 5.1
represents an orifice plate, a simple circular plate with a circular hole with diameter
d2. The maximum flowline contraction in an orifice geometry occurs slightly
downstream from the orifice at cross section 3, a point known as the vena contracta.
The corresponding reduced diameter is d3.
The same effect occurs in other restrictions with a sudden change in flow area,
such as simple chokes, as shown schematically in the middle of Fig. 5.1. Both the
orifice and the choke geometries have a region downstream of the restriction where
strongly turbulent flow occurs, which may include large vortices or areas of flow
reversal. The turbulent flow results in dissipation of energy and a permanent reduction
in pressure between the upstream and the downstream side of the restriction. If such
a permanent pressure drop is not desired—e.g., when the restriction is meant only to
serve as an indirect flowmeter—a geometry as displayed at the bottom of Fig. 5.1
should be applied. A venturi, as this is known, allows for a gradual recovery of the
original flowline pattern with a minimum of turbulent dissipation and therefore only a
small permanent pressure drop. In contrast, choke valves usually have a geometry
that severely disturbs the flow and creates strong turbulence because they are
designed to dissipate energy and produce a permanent reduction in pressure.
Fig. 5.1—Schematic restriction geometries. The figure is not to scale.

5.3 Single-Phase Oil Flow

5.3.1 Reversible Pressure Drop. A simple semiempirical expression for the pressure
drop in incompressible single-phase liquid flow through a sudden restriction is given
by

where p1 and p4 are the pressures at the upstream and the downstream side of the
restriction, respectively; A2 = π (d2)2 /4 is the throat area; and Cd is an empirical
discharge coefficient. Eq. 5.1, which has been implemented in MATLAB routine
choke_oil.m, reflects that the pressure drop over a restriction is proportional to the
kinetic energy per unit volume: , where V is some representative volume
for the choke. This is in line with the fact that the flow is highly turbulent, which means
that the pressure drop is dominated by inertial forces rather than viscous forces.
A derivation of Eq. 5.1 can be obtained by taking a step back and starting from the
pressure drop equation for steady-state pipe flow in a pipe with a diameter that
varies along the pipe axis. From experiments it has been determined that the majority
of energy dissipation in a restriction occurs in the region of diverging streamlines,
whereas in the converging region the flow experiences only a small friction loss.
Therefore we can describe the flow between Points 1 and 3 by disregarding gravity
and friction. Following the same approach as in Section 3.1, the steady-state mass
and momentum balances can then be written as

where we introduced an extra term to account for the axial component of the force
resulting from pressure on the pipe wall. Expanding Eqs. 5.2 and 5.3 and dropping all
terms higher than first order in the differentials lead to

from which follows

Note that Eq. 5.6 is no longer dependent on A. Assuming a constant density, the
equation can be integrated from s1 to s3 according to

which results in

or
Eq. 5.9 is the Bernoulli equation for frictionless incompressible flow which can be
interpreted as a macroscopic mechanical energy balance (see Bird et al. 2002, §7.4).
The frictionless pressure drop over the contracting part of a sudden restriction follows
from Eq. 5.9 as

where A1 = π (d1)2 /4 is the pipe flow area, A3 = π (d3)2 /4 is the area of the vena
contracta, and β3 = A3 / A1. Note that p3 < p1 (i.e., an increase in velocity corresponds
to a decrease in pressure). The true pressure drop in the converging part of a
restriction will be somewhat larger than follows from Eq. 5.10 because frictionless
flow is an idealization, an effect that can be taken into account by replacing A3 in Eq.
5.10 by a slightly reduced surface area . Moreover, Point 3 corresponds to the
vena contracta because of the additional streamline contraction downstream of the
restriction. To express the pressure drop in terms of the known throat area A2 = π
(d2)2 /4, one can use a contraction coefficient, defined as for the frictionless
case or as for the case with friction. The pressure drop over the
converging part of a restriction (i.e., the maximum pressure drop), can now be
expressed as

where β2 = A2 / A1. (Note that some publications define the contraction coefficient as
the diameter ratio d3/d2 instead of a surface area ratio. Also, β is sometimes defined
as a ratio of diameters.) Alternatively, this effect is accounted for with a discharge
coefficient in which case Eq. 5.11 becomes

Comparison of Eqs. 5.11 and 5.12 shows that the contraction and discharge
coefficients are related as

For turbulent flow the value of is usually taken as 0.62 for a sharp edged orifice,
but it may become even smaller for other types of sudden restrictions. The
corresponding value of for an orifice is therefore typically smaller than 0.4. Note
that reflects only the losses in the converging part of the restriction, whereas Cd in
Eq. 5.1 reflected the losses over the entire restriction.
 Because orifices are often used as indirect flowmeters, it is useful to rewrite Eq.
5.12 in terms of the flow rate q as a function of pressure drop, resulting in an
expression known as the orifice equation:

where the approximation is valid for small values of β and where we have used a
minus sign under the assumption that the flowmeter is mounted in the production
stream. Alternatively, the orifice equation can be expressed in terms of mass flow
rate ṁ = ρq as

5.3.2 Permanent Pressure Drop. In theory, a large part of the pressure drop over
the converging part of a restriction is reversible because the flow is nearly frictionless.
However, in the diverging part there is usually a high amount of energy dissipation,
resulting in a limited pressure recovery and, therefore, a considerable permanent
pressure drop (see Fig. 5.2). An exception are venturis, which by design have very
small permanent losses. In chokes the energy dissipation is usually so high that there
is only a very small pressure recovery in the diverging section. For a well-defined and
simple geometry, such as an orifice, it is possible to determine the permanent
pressure drop theoretically using the Bernoulli equation with an additional loss term
(see, e.g., Bird et al. 2002).
For complex geometries, as in chokes, the permanent pressure drop needs to be
determined experimentally. The resulting permanent pressure drop Δpperm = p1 – p4
can then be expressed with the aid of Eq. 5.12 as
Fig. 5.2—Schematic pressure distribution in a choke. The figure is not to scale.

where fperm is a permanent pressure drop factor, which may have values ranging from
close to zero, for venturis, to close to unity, for complex-shaped chokes. Note,
however, that slightly different definitions may be encountered in the literature. For
example, Perry et al. (1997, 8–74) define the pressure recovery factor frec as the
square root of the permanent pressure drop factor.
Comparison with Eq. 5.1 shows that the permanent pressure drop can also be
expressed with a single discharge coefficient Cd as

such that

which accounts for (1) the effects of the term ; (2) the effect of irreversible
losses in the diverging part, as represented by fperm ; and (3) the combined effects of
irreversible losses and streamline contraction in the converging part, as represented
by either or . Note that because Δpperm < Δpmax it is necessary that

For most restriction types, the precise values of and fperm are not known.
However, values for Cd have been extensively published in the form of tables and
charts for various restriction geometries such as bends or valves. Typical values
range from 0.7 to 1.2. Note that in some publications, it is stated that the value of the
discharge coefficient can never exceed unity. This is indeed the case for because it
reflects only pressure losses not accounted for in the isentropic flow model. However,
Cd also reflects pressure recovery and may thus have a value larger than unity. Note
also that different definitions of the discharge coefficient and related variables are
used throughout the literature.
A related quantity is the valve coefficient Cv, also known as the flow coefficient,
which is sometimes used to quantify the capacity of a valve as determined
experimentally by pumping water through it at a known rate and measuring the
resulting pressure drop:

where is the flow rate in US gallons per minute, γw is the specific gravity of the liquid
flowing through the valve, and is the pressure drop in psi. Note that the dimensions
of Cv are gal/min-√psi, whereas Cd is dimensionless. The pressure drop in Pa over a
valve with valve coefficient Cv gal/min-√psi for single-phase flow of a liquid with
density ρ (kg/m3) at a flow rate of q (m3/s) can therefore be expressed as

from which follows

5.3.3 Choke Performance Curves. Fig. 5.3 depicts a choke performance curve—
i.e., a plot of tubinghead pressure vs. flow rate flow through a wellhead choke—for
four different choke sizes as computed with MATLAB file choke_oil.m. The
corresponding values of d2/d1 and β2 have been displayed in Table 5.1. The discharge
coefficient Cd = 0.7. In practice, single-phase oil flow through a choke seldom occurs
because usually the pressure at the wellhead is below the bubblepoint pressure.
However, the figure clearly illustrates the quadratic relationship between pressure
drop and flow rate for singlephase liquid flow. Moreover, single-phase liquid flow
through a restriction does occur in measurement devices such as orifices or venturis.
5.4 Single-Phase Gas Flow

5.4.1 Isentropic Flow. Unlike what we did for single-phase gas flow through pipes,
where we assumed isothermal conditions, we cannot neglect the change in
temperature for gas flow through restrictions. A full thermodynamic analysis would
require the use of a total energy balance, in addition to the mass balance equation
(Eq. 3.3) and the momentum balance equation (Eq. 3.4) derived in Section 3.1. We
refer to Bird et al. (2002) or a thermodynamics textbook such as Moran and Shapiro
(1998) for such a nonisothermal analysis. In the most general case this requires the
simultaneous solution of three coupled partial-differential equations together with an
algebraic equation of state (EOS) to obtain the unknown values of pressure, density,
velocity, and temperature as a function of time and space. Because of the rapid
compression and expansion of gas in a restriction, however, there is very little time
for energy exchange with the surroundings and the flow can thus be considered
adiabatic. Furthermore, we may neglect the friction forces when we consider the flow
in the converging section of the restriction, just as we did for single-phase oil. This
implies that the converging flow may be considered isentropic, a thermodynamic
condition also indicated as frictionless-adiabatic or reversible-adiabatic. Under
adiabatic conditions, there is no exchange of energy with the surroundings, whereas
reversibility implies that there is no change in entropy. A full derivation of the
thermodynamic relationship in this section is outside the scope of this text, and we
refer to thermodynamics textbooks instead; see, e.g., Moran and Shapiro (1998),
Chapter 6, for a detailed description of isentropic processes.
Fig. 5.3—Choke performance curves for single-phase oil flow through four different chokes with Cd = 0.7
and sizes 1–4 as specified in Table 5.1. The downstream pressure p4 = 0.3 MPa. The other properties have
been taken from Table 4.1.

Size (in.) Size (mm) d2/d1 β2

d1 all chokes 2.453 62.31 – –

d2 choke 1 8/64 3.18 0.051 0.002 6
d2 choke 2 16/64 6.35 0.102 0.010 4
d2 choke 3 24/64 9.53 0.153 0.023 4
d2 choke 4 32/64 12.70 0.204 0.041 6
d2 choke 5 40/64 15.88 0.255 0.064 9
d2 choke 6 48/64 19.05 0.306 0.093 5
Table 5.1—Choke sizes, corresponding flow diameter ratios d2/d1, and surface area ratios β2 = (d2/d1)2 for
figs. 5.2, 5.3, and 5.5.

In the case of isentropic flow it is not necessary to solve a differential equation to

obtain the change in temperature in the flowing gas. Instead we can use an algebraic
relationship between pressure and temperature, which for an ideal gas can be
expressed as
where C is a constant and where γ is the ratio of specific heat capacities at constant
pressure and constant volume:

The corresponding EOS for an ideal gas under isentropic conditions can be
expressed as

(Note that for isothermal conditions, the EOS for an ideal gas (Eq. 5.25) can simply
be expressed as p/ρ = C.)
With the aid of Eq. 5.23 we can also write that

and then use the definition of the gas formation volume factor (Eq. 2.34) to obtain the
relationship

where we used Z1 = Z3 = 1 because we restrict the derivation to ideal gases. (The

equivalent expressions for a nonideal gas are more complex and therefore outside the
scope of this book.) The value of γ is dependent on the gas composition and
pressure. For hydrocarbon gases under moderate pressures γ is usually chosen as a
constant with a value somewhere between 1.25 and 1.40.
Just as for single-phase liquid flow, we can now use Eq. 5.6 as a starting point for
the derivation of the pressure drop equation. After dividing both sides by ρ we obtain

Substitution of Eq. 5.25 in the left side along with integration from s1 to s3 results in
Integration of the right side of Eq. 5.28, and combining the result with Eq. 5.29, gives
us the isentropic Bernoulli equation for compressible flow:

In analogy to Eq. 5.11 we can now derive an expression for the pressure drop for gas
flow through an orifice as

or

where the coefficients β and B indicate surface area ratios and formation volume
factors, respectively. In analogy to Eq. 5.12 for single-phase oil flow, we can now
also write

where the discharge coefficient accounts for the contraction effect and for the
irreversible losses in the converging part of the restriction. The approximation
is justified, given the error already introduced by restricting the analysis
to ideal gases.
Although Eq. 5.33 resembles Eq. 5.12, it cannot be solved explicitly for one of the
pressures as a function of the other pressure and the flow rate because the formation
volume factors are functions of the unknown pressure. Therefore an iterative
procedure is needed to compute the maximum pressure drop Δpmax = p1 − p3. During
the iteration, Eqs. 5.26 and 5.27 should be used to express unknown downstream
values T3 and B3 in terms of known upstream values T1 and B1, or the other way
around. Note that we should still use the formation volume factor with the correct Z
value for one of the values B1 or B3 because the restriction to ideal gases that applies
to Eq. 5.27 is relevant only for the pressure drop over the choke. The iteration
procedure can be implemented in MATLAB with the aid of one of the methods described
in Appendix D. For example, if it is required to compute the upstream pressure p1 for
a given downstream pressure p3 and upstream temperature T1, we can combine Eqs.
5.33 and 5.27 as

and solve for the roots of this equation—i.e., the values of p1 for which f(p1) becomes
equal to zero—with the aid of one of the methods described in Appendix D. Although
the pressure drop needs to be computed iteratively, it is still possible to obtain an
explicit orifice equation (i.e., an expression for the flow rate qg,sc as a function of the
known pressures p1 and p3):

where the minus sign stems from our convention that production flow rates are
negative. With the aid of Eqs. 5.26 and 5.27, Eq. 5.35 can be rewritten as

or, in terms of the mass flow rate ṁg, as

which is the conventional representation in the gas engineering literature.

5.4.2 Critical Flow. Eq. 5.36 is a relation between the flow rate qg,sc and the pressure
ratio (p3/p1). It can be verified that for a given upstream pressure p1, qg,sc has a
maximum that occurs when the derivative d(qg,sc)/d(p3/p1) is equal to zero. To find the
corresponding value of the pressure ratio, it is sufficient to take the derivative of the
term between square brackets. (This is equivalent to computing the maximum of
instead of qg,sc, which removes the square root from the expression.) Starting from
Eq. 5.37 this results in

Eq. 5.38 can be solved for (p3/p1), resulting in

For typical values of γ between 1.25 and 1.40, we therefore find

or

The velocity corresponding to this critical pressure ratio that occurs in the throat of
the restriction is equal to the sonic velocity (i.e., the speed of sound) under isentropic
conditions. The corresponding expression for critical flow of an ideal gas can be
found by substitution of Eq. 5.39 into Eq. 5.36, which results in

Rewriting Eq. 5.42 in terms of p1 results in an expression for the upstream pressure
for ideal gas flow through a critical choke (i.e., a choke operating under critical
conditions):

which has to be solved iteratively because of the pressure dependence of Z1. Note
that the term between square brackets has a fixed value and is not dependent on p1.
Eq. 5.43 for critical choke flow forms an almost linear relationship between p1 and
qg,sc, the small deviation from linearity being caused by the pressure dependence of
Z1. With the aid of Eqs. 5.26 and 5.39 it follows that the throat temperature T3 under
critical flow conditions becomes

For typical values of γ between 1.25 and 1.40, this implies a temperature reduction
over the converging part of the choke in between T3,abs = 0.89 T1,abs and T3,abs = 0.83
T1,abs . A computation of the critical pressure and temperature ratios for real gases is
quite involved and beyond the scope of this book (see, e.g., Cornelius and Srinivas
2004).
The occurrence of critical flow in an orifice used for flow rate measurements is
usually avoided to limit the magnitude of the permanent pressure drop. In many other
restrictions, however, in particular in choke valves, critical flow frequently occurs.
Indeed, the name of such a valve stems from the fact that once the fluid flowing
through the valve reaches the critical pressure ratio, the flow is choked; i.e., the
maximum possible mass flow rate through the choke has been reached (for a given
upstream pressure). Another effect of reaching the speed of sound in a choke is that
disturbances can no longer travel upstream. The reason for this effect is the finite
propagation speed of disturbances. In a 1D conduit filled with a stationary fluid, a
small pressure disturbance will result in two waves, both traveling at the speed of
sound, away from the point of disturbance and in opposite directions. If the fluid is not
stationary but flows through the conduit, the two waves will be influenced such that
one will travel faster and the other slower. If the fluid moves so fast that it reaches
the sonic velocity, one of the waves will travel at twice the sonic velocity (relative to
the conduit) while the other wave will become stationary. As a consequence of this
effect any change in back pressure in a gas treatment facility will travel through the
flowlines in the direction of the reservoir but will not pass through any choke operating
at a critical condition. Although this description is simplistic, it captures the basic idea.
For a more detailed thermodynamic analysis of compressible flow in a restriction,
see, e.g., Cornelius and Srinavas (2004).
Recall that critical flow was encountered earlier, in Section 3.4, where an
isothermal expression for the speed of sound in a gas well was derived. In reality
such high flow velocities will not occur in gas wells, except in case of a blowout (i.e.,
an accidental loss of well control). However, in choke valves critical flow occurs
frequently.

5.4.3 Pressure Drop in Chokes. Sub-critical Flow. The maximum pressure drop
Δpmax = p1 – p3 during sub-critical single-phase gas flow in a choke can be computed
with the aid of Eq. 5.33, which needs to be solved iteratively because of the pressure
and temperature dependence of B1 and B3. In analogy to Eq. 5.16, the permanent
pressure drop during sub-critical flow could, in theory, be approximated with an
empirical permanent pressure drop factor fperm according to

where values of fperm for chokes probably range from 0.5 to 1.0 with most likely
values at the high end. However, values for fperm are usually unavailable, and a more
pragmatic solution is therefore often introduced through simply replacing the
discharge coefficient for the converging part of the choke in Eq. 5.33 by an overall
discharge coefficient Cd for the entire choke. Note that because Cd accounts for the
partial pressure recovery in the diverging part, it should always hold that . An
approximate solution for the pressure drop over the entire choke can therefore be
obtained by iteratively solving the equivalent of Eq. 5.33:

where the corresponding relationship between B4 and B1 follows from Eq. 5.27 as

Because of the significant friction losses in the diverging part of the choke, the
assumption of isentropic flow is no longer accurate and this relationship is therefore
now only a rough estimate. Note that either B1 or B4 should be computed in the usual
way, using the correct real-gas Z value, whereafter the other value should be
approximated with the ideal-gas isentropic relationship shown in Eq. 5.47 (see also
the comments after Eq. 5.33).
A more accurate analysis of the pressure recovery in the diverging part of a gas
choke would require thermodynamical computations beyond the level of this book. We
will also not pursue an analysis of the temperature recovery, but we do know that the
downstream temperature T4 will always be higher than T3 because of the dissipation
of mechanical energy in the form of heat.
Critical Flow. The critical flow boundary in terms of the downstream pressure p4
can, in theory, be obtained with the aid of Eq. 5.45 as

Although actual values of fperm are usually unknown, Table 5.2 gives an idea of the
possible range of critical pressure ratios for fperm values between 0.7 and 1.0 at two
different values of γ. Note that if we assume that fperm = 1, Eq. 5.48 reduces to the
equivalent of Eq. 5.39:

which seems to be the expression that is most often encountered in the petroleum
literature. To compute the upstream pressure in critical choke flow, we can follow the
same approach as for sub-critical flow and simply use an overall discharge coefficient
Cd. The equivalent of Eq. 5.43 then becomes

γ ≈ 1.25 γ ≈ 1.40
fperm (p4/p1)crit (p1/p4)crit (p4/p1)crit (p1/p4)crit
0.7 0.69 1.46 0.67 1.49
0.8 0.64 1.56 0.62 1.61
0.9 0.60 1.68 0.58 1.74
1.0 0.55 1.82 0.53 1.89
Table 5.2—Critical pressure ratios for single-phase gas chokes.

which has to be solved iteratively because of the pressure dependence of Z1. A more
accurate analysis of the pressure and temperature reversal during critical flow
requires thermodynamics theory outside the scope of this book. In simple terms, the
gas further accelerates after passing through the throat and reaches a supersonic
velocity, which corresponds to a further drop in pressure. The recovery to
downstream pressure p4 then occurs in a thin region where both pressure and
temperature increase abruptly in the form of a stationary shock wave. It may be
assumed that there is no wall friction and no exchange of heat with the surroundings
in the diverging part of a choke under critical conditions (i.e., that the flow is
isenthalpic). However, the change in thermodynamic state is no longer reversible, and
inside the shock there is a large conversion of mechanical energy to thermal energy,
which is indeed why a choke causes a considerable permanent pressure drop.
Computations. We can use the approximate equations for gas flow through a
choke, i.e., those based on an overall discharge coefficient Cd, in the following
scenarios:

1. To compute the gas flow rate for a given pressure drop. In this scenario it is
 necessary to first compute the effective pressure ratio according to

where (p4/p1)crit is given by Eq. 5.49, and then use the equivalents of Eqs.
5.36 and 5.42, which, using the effective pressure ratio, can be expressed
jointly as

2. To compute the pressure drop over a restriction for a given flow rate and a
given downstream pressure. In this scenario it is necessary to investigate both
critical and sub-critical flow. Assuming sub-critical flow, the upstream pressure
can be computed iteratively with the aid of Eq. 5.46. (Refer to the text just
after Eq. 5.33 for a discussion on how to perform the iteration in MATLAB.) If
the resulting pressure ratio (p4/p1) turns out to be smaller than the critical ratio
(Eq. 5.48), the flow is critical and the upstream pressure can be computed
iteratively with Eq. 5.50.
3. To compute the pressure drop over a restriction for a given flow rate and a
given upstream pressure, the same procedure may be followed as under
Point 2. In practice, this third scenario is of relevance only for sub-critical flow.
For critical flow we will typically find that the computed upstream pressure is
different from the given upstream pressure, which implies that the choke
model and/or any of the given parameters are inconsistent.

Option 2 has been implemented in the MATLAB routine choke_gas.m. Fig. 5.4 depicts
the choke performance curves for gas flow through four differently sized chokes,
each with a total discharge coefficient Cd = 0.7 and specific heat capacity ratio γ =
1.25. The horizontal dotted line indicates the boundary between sub-critical and
critical flow and has been computed with the aid of Eq. 5.49. According to Eq. 5.44,
the temperature at the throat of the choke during critical flow in this example drops to
T3 ≈ 0.89 × (30 + 273.15) − 273.15 = −3.3°C (i.e., to below the freezing point of
water). Note that choke temperatures below the freezing point of water may lead to
the formation of ice around the outside of the choke, or worse, to the formation of
gas hydrates at the inside. This is especially a problem in offshore installations where
the gas is already cooled by seawater during its flow through the riser.
Fig. 5.4—Choke performance curves for the flow of single-phase gas with specific heat capacity ratio γ =
1.25 through four chokes with discharge coefficient Cd = 0.7 and bean sizes 2–5 as specified in Table 5.1.
The downstream pressure p4 and the upstream temperature T1 are 12 MPa and 30°C respectively. The
other properties have been taken from Table 3.1.

5.5 Multiphase Flow: Empirical Models

5.5.1 Multiphase Choke Models. Just as for multiphase flow through pipes, there
are many models for multiphase flow through chokes, and they have strongly varying
levels of complexity. The simplest models are purely empirical and consist of
relationships that have been obtained through curve fitting of experimental data. They
are generally applicable to a limited range of choke geometries and fluid properties.
At the other end of the scale are detailed models that take into account flashing of
components, phase slip, and thermal interactions. These are based on
thermodynamics theory that is outside the scope of this book. For a good overview of
various choke models and an example of a detailed description see Schüller et al.
(2003). Here we will first address some simple empirical models for multiphase choke
flow, and thereafter some models at an intermediate level of complexity that have a
somewhat more theoretical basis.

5.5.2 Empirical Models for Critical Flow. We define the following variables: ql,sc is
liquid flow rate through the choke, m3/s (STB/D); p1 is pressure upstream of the
choke, Pa (psi); p3 is pressure at the throat of the choke, Pa (psi); and p4 is pressure
downstream of the choke, Pa (psi).
Just as for single-phase gas flow, for multiphase flow there exists a critical
pressure ratio (p3/p1)crit. Consider a situation with a constant upstream pressure p1
and a gradually decreasing downstream pressure p4 such that the pressure ratio
(p4/p1), and therefore also the pressure ratio (p3/p1), gradually decreases. As long as
(p3/p1) > (p3/p1)crit, the flow rate |ql,sc| increases. This is as expected; the larger the
pressure drop, the faster the flow. However, once p3/p1 < (p3/p1)crit, the flow rate
remains constant and a further reduction of p4 no longer has an effect. Note that it is
still possible to increase the flow rate, but only by increasing the upstream pressure
p1. The importance of critical choke flow for oil well performance is that pressure
disturbances downstream can no longer propagate through the choke to the
upstream side. Hence the flow behavior becomes independent of the throat pressure
p3 and therefore of the downstream pressure p4.
There are advantages in operating the choke at critical conditions. The
downstream pressure may vary for many reasons: There may be more wells entering
the same manifold, and one of these may be shut in; there may be fluctuations in the
processing system; the operating staff may change valve settings in the downstream
system; and so on. These effects will not influence the production rate of the well if
the choke is operating above critical conditions. Several expressions exist to predict
the occurrence of critical multiphase flow through a choke (see, e.g., Chapter 5 of
Brill and Mukherjee 1999).
As a rule of thumb, critical flow occurs when the pressure ratio p4/p1 is below the
critical ratio

but it should be kept in mind that this is a rather crude approximation of a very
complex phenomenon. At sub-critical conditions, the flow rate of a gas/liquid mixture
through a choke depends on the specific type of choke, the properties of the
multiphase mixture, and so on, and there is no simple pressure drop/flow rate
relationship. At critical conditions, there are a number of empirical correlations (see,
e.g., Gilbert 1954 or Ros 1960). Other correlations are connected to the names of
Baxendell and Achong (see Brill and Mukherjee 1999). Each of these has been
determined from a limited set of measurements for specific fluid properties and choke
types, and it is therefore not possible to make a general recommendation as to which
of them is preferred. The reason to list multiple correlations is, rather, to emphasize
the approximate nature of the underlying empirical models and illustrate their range of
uncertainty. Moreover, their formulation can serve as a template to develop a new
correlation for a specific field situation. All of them have the form
where Rgl is producing gas/liquid ratio, m3/m3 (scf/STB), dch is choke diameter, m
(1/64 in.), and A, B, C, and D are experimentally determined constants given in Table
5.3. Here, we have assumed that ql,sc has a negative value in line with our convention
that flow rates in production wells are negative. Note that in field units the diameter is
specified in 1/64 of an inch. For a given choke size, the choke performance curve is a
linear function of the flow rate (see Fig. 5.5). (Note that here we plotted just the oil
flow rate rather than the liquid flow rate.) For a fixed pressure, the flow rate is
approximately equal to the square of the choke diameter (i.e., the cross-sectional
area, as might be expected). For upstream pressures below approximately 1.7 times
the downstream pressure p4, these curves are of course invalid because then the
choke operates in the sub-critical regime.

5.5.3 Empirical Models for Sub-critical Flow. An extension of these empirical

models to the sub-critical regime can be obtained as follows. From Eq. 5.54 it follows
that the flow rate at the critical pressure ratio is given by

SI Units
Correlation A B C D E F
Gilbert 3.75 × 1010 0.546 1.89 1.01 × 105 5.61 2.52 × 103
Ros 6.52 × 1010 0.500 2.00 1.01 × 105 5.61 2.52 × 103
Baxendell 3.58 × 1010 0.546 1.93 1.01 × 105 5.61 2.52 × 103
Achong 1.43 × 1010 0.650 1.88 1.01 × 105 5.61 2.52 × 103
Field Units
Correlation A B C D E F
Gilbert 10.0 0.546 1.89 14.7 1.00 1.00
Ros 17.4 0.500 2.00 14.7 1.00 1.00
Baxendell 9.56 0.546 1.93 14.7 1.00 1.00
Achong 3.82 0.650 1.88 14.7 1.00 1.00
Table 5.3—Coefficients for different choke models.
Fig. 5.5—Choke performance curves for multiphase critical flow through four chokes with sizes 3–6 as
specified in Table 5.1 and fluid properties according to Table 4.1. The graphs have been obtained with the
aid of Eq. 5.54 and the Gilbert correlation. Note that they are valid only in the critical regime, which is
outside the gray area.

where, according to Eq. 5.53, p1,crit ≈ 1.7p4. We now look for an expression that fulfills
the following boundary conditions:

where
as follows from differentiation of Eq. 5.54. An appropriate expression that can fulfill
four boundary conditions is the third-order polynomial

where the four coefficients can be derived from Eqs. 5.56 through 5.59 as

Note that this sub-critical extension of the empirical critical choke model Eq. 5.54 is
not based on first-principle physics, but rather a pragmatic correction. Fig. 5.6
depicts the same choke performance curves as in Fig. 5.5, but now with
pragmatically corrected sub-critical parts of the curves. The qualitative behavior is the
same as for the single-phase gas curves in Fig. 5.4 (by design). The empirical model
for sub-critical and critical flow described in this section has been programmed in
MATLAB file choke_multiphase_simp.m.

5.6 Multiphase Flow: Theoretical Models

5.6.1 Polytropic Flow. Ros (1960), Sachdeva et al. (1986), Perkins (1993), and Al-
Safran and Kelkar (2009) presented multiphase choke models of intermediate
complexity. The main assumptions in these models are that during passage through
the choke

• Flow is 1D
• Acceleration losses form the dominant contribution to pressure drop
• Phase compositions (and therefore Rs) remain constant (a condition known as
“frozen” flow)
• The liquid phase density (and therefore Bl) remains constant
• The gas phase expands between isothermally and isentropically
• The gas/liquid mixture expands isentropically
• No slip exists between the gas and liquid phases, except in Al-Safran and
 Kelkar (2009), where this assumption is relaxed

In this section we follow a similar approach. Just as in the four papers listed
above, the equation of state for the gas is modeled as polytropic (i.e., between
isothermal and isentropic).

Fig. 5.6—Choke performance curves for multiphase critical flow through four chokes with sizes 3–6 as
specified in Table 5.1 and fluid properties according to Table 4.1. The graphs have been obtained with the
aid of Eqs. 5.54 and 5.61 and the Gilbert correlation.

This is to account for the fact that the temperature of the expanding gas in multiphase
flow drops less than that in single-phase gas flow because it rapidly takes up heat
from the surrounding liquid. The relationship between pressure and gas density under
polytropic conditions is given by

where n is the polytropic expansion coefficient. Eq. 5.66 strongly resembles Eq. 5.25
for isentropic expansion, the only difference is that n replaces γ. The polytropic
expansion coefficient n is defined as
where xg and xl are the gas and liquid mass fractions defined in Eqs. 4.19 and 4.20;
cp and cv are, as before, the specific heat capacities for gas at constant pressure and
volume, respectively; and cl is the specific heat capacity for liquid at constant volume.
It can be verified that for λg = 1 and λl = 0 (i.e., for single-phase gas flow); n
becomes equal to γ, corresponding to isentropic flow; whereas for λg = 0 and λl = 1,
n becomes equal to unity, corresponding to isothermal flow. For typical hydrocarbon
mixtures the value of n is therefore between 1 and 1.3 depending on the gas/liquid
ratio. For a derivation of Eq. 5.67 and a detailed description of the physical
mechanisms involved, see Ros (1960).
Just as was the case for isentropic expansion, for polytropic expansion we have an
algebraic relationship between pressure and temperature given by

such that we do not have to solve a differential equation to take account of the
conservation of total energy over the choke. In analogy to Eqs. 5.26 and 5.27 we can
also write

and

where we used Z1 = Z3 = 1 because we restrict the derivation to ideal gases. To

derive the multiphase choke equations, we can now proceed along the same lines as
we did for gas flow and start from a momentum balance that considers only
acceleration and pressure forces (and no friction or gravity effects):

The no-slip mixture velocity vmn and the inverse mixture density 1/ρmn can be written
with the aid of Eqs. 4.11, 4.19, 4.20, and 4.29 as
(Note that for mixture flow rates, as opposed to mixture velocities, there is no need to
make a distinction between slip and no-slip conditions. Therefore we can simply use
the subscript m to indicate the mixture flow rate qm .)
Substitution of Eqs. 5.66 and 5.73 in the left side of Eq. 5.71, along with
integration from s1 to s3, results in

Integrating the right side of Eq. 5.71 and then combining the result with Eq. 5.74 give
us the polytropic Bernoulli equation for gas/liquid flow:

which with the aid of Eqs. 4.19, 4.20, and 5.75 and the equality qgρg + qlρl = qm,sc
ρm,sc can be rewritten as

Using the approximation and then introducing a discharge coefficient ,

to account for the contraction effect and for the irreversible losses in the converging
part of the restriction, results in

which is analogous to Eq. 5.33 for the pressure drop over a single-phase gas choke
as derived in Section 5.4. Continuing the analogy, we can solve Eq. 5.77 for qm,sc,
resulting in
It can be verified that Eq. 5.78 reduces to its single-phase oil counterpart (Eq. 5.14)
for volume fraction values λg,sc = 0 and λl,sc = 1. Similarly, it reduces to its single-phase
gas counterpart (Eq. 5.35) for λg,sc = 1 and λl,sc = 0.

5.6.2 Critical Flow. To further continue the analogy, we need to define various
formation volume factor ratios that are a bit more complicated than for the single-
phase gas case. For example, we find that

and, using similar reasoning, that

and that

The parameter ε can be interpreted as the local liquid/gas ratio:

where we used the assumption that Bl remains constant during passage through the
choke. With the aid of Eqs. 5.79 through 5.81, Eq. 5.78 can be rewritten as
where

For ql,sc = 0 (i.e., for ε = 0), this equation reduces to its gas equivalent, Eq. 5.36,
whereas for qg,sc = 0 (i.e., for ε → ∞), it reduces to the single-phase oil expression,
Eq. 5.14. Just like the gas flow rate qg,sc in Eq. 5.36, the mixture flow rate qm,sc in Eq.
5.83 obtains a maximum for a critical pressure ratio (p3/p1)crit. However, unlike for gas
flow, it is not possible to obtain a closed-form expression for (p3/p1)crit, and so we
need to search numerically for the maximum of Eq. 5.83 or, equivalently, for the
maximum of Θ. In addition, it is necessary to iterate on p1 because of the pressure
dependence of n and of the formation volume factors Bg,1 and Bl,1, which appear in
the definition of ε. Rewriting Eq. 5.83 in terms of p1 results in the upstream pressure
for multiphase flow through a critical choke:

where Θcrit is the value of Θ corresponding to (p3/p1) = (p3/p1)crit. Eq. 5.85 needs to be
solved iteratively because of the pressure dependence of Bm,1 (or, choosing the last
term, the pressure dependence of ρmn,1 and qm,1). Although not immediately obvious, it
forms a near-linear relationship between the upstream pressure and the flow rate,
just as its single-phase gas counterpart, Eq. 5.43.

5.6.3 Pressure Drop in Chokes. Just as for single-phase oil and single-phase gas
chokes, the effect of pressure recovery in the diverging part of a multiphase choke
can be accounted for pragmatically by replacing the discharge coefficient for the
converging part with an overall discharge coefficient . The discussion in Section
5.4.3 for single-phase gas chokes is therefore also valid for multiphase chokes, the
only difference being that it is no longer possible to compute the critical pressure ratio
in closed form. We will not repeat the discussion except for the most frequently used
scenario in which it is required to compute the pressure drop over a choke for a given
flow rate and a given downstream pressure. Assuming sub-critical flow, the upstream
pressure can then be computed iteratively with the equivalent of Eq. 5.85:
where Θ is now defined as

(Refer to the text just after Eq. 5.33 for a discussion on how to perform the iteration
in MATLAB and how to treat the formation volume factor calculations.) If the resulting
pressure ratio (p4/p1) turns out to be smaller than (p4/p1)crit, the flow is critical and the
upstream pressure can be computed with the aid of the same Eqs. 5.86 and 5.87
after replacing Θ with Θcrit, and (p4/p1) with (p4/p1)crit, respectively. This scenario has
been implemented in the MATLAB routine choke_multi_phase.m.
Fig. 5.7 depicts the choke performance curves for multiphase flow through four
differently sized chokes. The dashed lines depict the corresponding empirical “Gilbert”
lines for critical flow that were shown earlier in Fig. 5.6. In this example, the
correspondence between the results of the numerical solution and those of the
empirical expression happens to be quite good. However, a much worse
correspondence may be encountered for other parameter values. The horizontal
dotted line indicates the boundary between sub-critical and critical flow, which, with
the aid of the MATLAB file choke_multi_phase_boundary.m, has been computed
numerically. An example explaining how to plot choke performance curves for either
single-phase oil flow, single-phase gas flow, or multiphase flow is given in the MATLAB
file example_choke_performance curve.m.
Fig. 5.7—Choke performance curves for multiphase flow with specific heat capacities cl = 0.80, cp = 0.25
and cv = 0.20, through four chokes with discharge coefficient Cd = 0.7 and bean sizes 3–6 as specified in
Table 5.1. The downstream pressure p4 and the upstream temperature T1 are 0.7 MPa and 30°C,
respectively (other properties have been taken from Table 4.1, and the dashed lines represent the
extended Gilbert curves that were depicted earlier in Fig. 5.6).

5.7 Questions

5.1 Why is the Bernoulli equation suitable for modeling the flow in the converging
part of a choke, but not in the diverging part?
5.2 Which physical mechanisms are causing the maximum and the permanent
pressure drop over an orifice plate? What kind of a device would you choose
to obtain flow rate measurements with a lower pressure drop than with an
orifice plate?
5.3 Why is the temperature at the throat of a gas choke lower than at the
entrance?
5.4 Why is it often desirable to operate a wellhead choke in the critical regime?
5.5 Why does critical choke flow not occur for single-phase oil?
5.6 Why are the expressions for critical flow over a choke much more
approximate than for critical flow over an orifice?
 5.7 Why do we need the assumption of polytropic conditions instead of isentropic
conditions to describe the flow in the converging part of a multiphase choke?
5.8 It is sometimes stated that the value of the discharge coefficient Cd of a
choke can never exceed unity. Is that correct? Why?
5.9 A manufacturer of orifice-type flow measurement devices for single-phase
liquid flow provides you with the following specifications: orifice size is 23 mm,
to be fitted in a 48-mm pipe (inner diameter); contraction coefficient is 0.62.
For a flow rate of −300 m3/d and oil with a density of 876 kg/m3, estimate the
maximum pressure drop.
5.10 If a restriction has a valve coefficient Cv = 40 gal/min-√psi, what is the
pressure drop over the restriction for flow of single-phase oil with a density of
830 kg/m3 at a rate of −0.8 m3/min?
5.11 Consider flow through an orifice-type dry gas meter. The relevant data are
given in Table 5.4.
(a) Is the flow critical or noncritical?
(b) What is the flow rate?
5.12 Consider multiphase flow through a choke. The relevant data are given in
Table 5.5.
(a) What is the upstream pressure, according to the Gilbert correlation?
(b) Is the key assumption for validity of the Gilbert correlation satisfied?

Property Value Units

A 4.91 × 10-4 m2
0.35 –
cp 0.26 J/kg K
cv 0.20 J/kg K
p1 4.3 MPa
p3 3.2 MPa
T1 65 °C
Z1 0.94 –
ρg,sc 0.80 kg/m 3
Table 5.4—Input data for dry gas orifice measurement for Question 5.11.

Property Value Units

Cd 0.9 –
cl 0.82 J/kg K
cp 0.26 J/kg K
 cv 0.20 J/kg K
dch 0.012 7 m
qo,sc –120 m 3/d
fw 0.70 –
Rgo 40 m 3/m 3
p4 1.2 MPa
T1 50 °C
ρg,sc 0.87 kg/m 3
ρo,sc 865 kg/m 3
ρw,sc 1030 kg/m 3
Table 5.5—Input data for multiphase choke flow for Question 5.12.

Questions 5.13 and 5.14 require the use of MATLAB.

5.13 Consider the same choke as in Question 5.12.

(a) Compute the upstream pressure with the aid of MATLAB function
choke_multi_phase.m. Is the result identical to your answer for Question
5.12?
(b) Compute the upstream pressure with the aid of MATLAB function
choke_multi_phase.m. Choose Standing black oil correlations. Is the
result identical to your answer for Question 5.12? What is the difference
and what are the consequences?
5.14 Consider a flowline/choke/tubing combination as depicted in Fig. 5.8. For the
choke and fluid properties, see Table 5.5; for the tubular properties, manifold
pressure, and temperatures, see Table 5.6.
(a) Make copies of example_choke_performance_curve.m and
example_intake_curve.m, and merge and modify the copied files to
create a choke performance curve and a tubing intake curve for the
flowline/choke/tubing combination. Use flow rates between 0 and –150
m3/d. Explain the difference in shape.
(b) Expand the file to make a plot of the critical pressure ratio (pfl / ptf)crit and
the actual pressure ratio (pf / ptf) as a function of flow rate. For which flow
rates is the choke operating in the critical regime? Why is the critical
pressure ratio slightly changing with flow rate (unlike in Fig. 5.7, where it
is constant)?
Fig. 5.8—Wellhead choke in between a flowline and a tubing.

Property Value Units

dt 0.07 m
dfl 0.10 m
et = efl 30 × 10-6 m
Lt 3300 m
Lfl 4800 m
Pmf 1.2 MPa
Tmf 40 °C
Twf 110 °C
αt 0 degree
αfl 90 degree
Table 5.6—Tubing, flowline, and manifold properties for Queston 5.14.

5.8 MATLAB Assignment: Choke Flow

5.8.1 Objectives

• Gain insight into the theory behind the multiphase choke model described in this
book and implemented in MATLAB functions choke_multi_phase.m and
choke_multi_phase_boundary.m.
• Gain insight into the effect of slip in multiphase choke flow through modifying
the existing MATLAB functions and testing the effect of the modifications.
• Demonstrate the similarities and/or differences between the (modified)
multiphase choke model described in this book and other models in the open
literature.
5.8.2 Warning. The second optional activity of this MATLAB exercise requires
considerably more theoretical analysis than do the MATLAB exercises in the other
chapters.

5.8.3 Assignment. The theory behind the multiphase choke model described in
Sections 5.6.1 through 5.6.3 and the associated MATLAB functions is related to several
choke models presented in the literature. However, one element present in the most
recent of these other models is absent: the effect of slip between liquid and gas in the
choke throat. You are requested to test the effect of adding slip models and,
optionally, to theoretically demonstrate the similarities and/or differences between the
modified model and the earlier models.

Tasks
• Read through the papers by Perkins (1993), Sachdeva et al. (1986), Schüller
et al. (2003, 2006), and Al-Safran and Kelkar (2009) without studying them in-
depth.
• Study Sections 5.6.1 to 5.6.3, dealing with multiphase choke modeling.
• Inspect the MATLAB functions choke_multi_phase.m and
choke_multi_phase_boundary.m. Make a flow chart that illustrates their
computational flow.
• Following the approach of Al-Safran and Kelkar (2009), modify the files to
implement the Schüller et al. (2006) slip model for critical flow. Study the
relevant parts of the papers in depth.
• Test the effect of the addition of slip by running
example_choke_performance_curve.m for various flow rates, choke diameters,
and discharge coefficients. In particular attempt to reproduce the test case
used by Al-Safran and Kelkar (2009).
• Optional Activity 1: Also add the “Grolmes and Leung” slip model for sub-
critical flow as described in Eq. 6 of Al-Safran and Kelkar (2009), and test the
effect.
• Optional Activity 2: Demonstrate theoretically the similarities and/or differences
between the multiphase choke model as used in this book (after addition of the
slip models) and the models proposed by Perkins (1993), Sachdeva et al.
(1986), and Al-Safran and Kelkar (2009).

5.8.4 Deliverables

• Flow chart
• MATLAB program listing
• Plots with choke performance curves illustrating the effect of adding the
Schüller et al. (2006) slip model for critical flow
• Optional Activity 1: plots with choke performance curves illustrating the effect
of adding the “Grolmes and Leung” slip model for sub-critical flow
• Optional Activity 2: a theoretical analysis demonstrating the similarities and/or
differences between the various multiphase choke models
Chapter 6

Inflow Performance: The Basics

6.1 What Is Covered in This Chapter?

Inflow performance is a quantitative measure of the effect of flow resistance in the
near-well reservoir and the completion on the production rate of a well. In this chapter
we discuss the functional relationship between inflow performance and reservoir
properties. Although we restrict the analysis to a simple circular reservoir produced
by a vertical well, many of the findings will remain valid for the more-complex well
configurations discussed in the next chapter. We start by deriving the governing
equations for radial single-phase porous media flow and apply these to describe the
inflow performance of single-phase oil and gas wells. In oil wells the relationship
between flow rate and drawdown (i.e., pressure drop over the near-well reservoir) is
linear and can therefore be expressed with a single productivity index, while for gas
wells it becomes nonlinear. However, in both situations we can usually apply relatively
simple analytical expressions to obtain rough estimates of the inflow performance of a
well for given reservoir and fluid properties (which typically are known only to a limited
extent). More-accurate values of a well’s single-phase inflow performance can
therefore be obtained only experimentally. To an increased extent, the same holds for
multiphase reservoir flow, and so we present some empirical relationships to describe
the two-phase (oil/gas) inflow performance of a well producing from an oil reservoir at
pressures below the bubblepoint pressure. Finally, we treat the inflow performance of
wells producing from multiple reservoirs.

6.2 The Importance of Inflow Performance

This section concerns the relationship between flow rate and pressure in the near-
wellbore region. The difference between the reservoir pressure and the flowing
bottomhole pressure (FBHP) of a well is the driving force for inflow into the wellbore.
Resistance to well inflow depends on reservoir rock properties, fluid properties, well
trajectory in the reservoir, details of the well completion, and sometimes the late
effects of drilling and workover activities. In combination, these factors determine the
inflow performance or deliverability of the well. Because all fluids entering the
wellbore have to pass through the narrow area around the wellbore, the highest flow
rates in the reservoir occur just there, and any increased resistance to flow has a
large effect on the well performance. Because inflow performance plays such an
important role, it should be regularly measured through well testing—i.e., by
connecting the well to a test separator and determining the gas, oil, and water flow
rates as functions of different wellbore pressures.
 Note that different types of well tests can be distinguished such as production
tests, which are mainly aimed at measuring the phase rates in a well at a given
flowing tubinghead pressure (FTHP), and pressure transient tests, primarily aimed at
inferring reservoir properties. The latter can be further classified depending on the
nature of the well operation: e.g., pressure buildup tests, in which a well is closed in
and the resulting increase in BHP is measured, or pressure falloff tests, in which a
closed-in well is suddenly opened and the resulting increase in flow rate and
decrease in FBHP is measured. (For further information on well testing see, e.g.,
Kamal 2009.)
The pressure in the near-wellbore region has to be measured at the bottom of the
hole with either a permanent downhole gauge (PDG) or a dedicated wireline tool.
This regular testing will indicate when a well is producing less than expected because
of impairment (i.e., blockage of the pores in the near-wellbore region). Remedial
measures, such as hydraulic fracturing of the formation through high-pressure
pumping of liquids, or stimulation with acids, can then be taken. The results of the well
tests can be incorporated in one of the models for inflow performance given below. It
is important to realize that these are only models and thus the actual downhole well
data must be respected. (For details of inflow performance measurement see, e.g.,
Golan and Whitson 1991 and Economides et al. 2013.)
As an example of a well’s inflow performance, consider a vertical oil well with
either an openhole producing zone or a perforated zone. The production performance
of this zone is usually described by an inflow performance relationship (IPR)
between the oil flow rate qo,sc and the BHP pwf. In practice, it is found that the IPR is
often a near-linear relationship between qo,sc and pwf, at least when pwf is above the
bubblepoint pressure pb. In that case the IPR can be expressed as a productivity
index (PI) J defined as the ratio between −qo,sc and the drawdown Δp, which is the
difference between the closed-in BHP pws (also known as the static BHP) and the
FBHP pwf (also known as the dynamic BHP), both measured at the top of the zone or
at the top of the perforations (although sometimes the middle of the perforated
interval is taken as reference level). If we assume that the static BHP equals the
reservoir pressure pR, we can write

where we adopt the convention that a positive flow rate qo,sc implies injection into the
reservoir (with pwf > pR), and a negative flow rate implies production into the well (with
pwf < pR). The PI therefore always has a positive value. The units of the PI are
m3/s·Pa (“strict” SI units), m3/d·kPa (“allowable” SI units), or STB/D-psi (field units).
The reservoir pressure is the pressure at the boundary of the drainage area of the
well. Alternatively, the PI can be defined in terms of the average reservoir pressure
pR,av in the drainage area of the well, which results in a somewhat higher value of the
PI for the same flow rate. For injection wells it is customary to use the injectivity
index (II) as an indication of the injection performance. The definition of the II is
completely analogous to that of the PI, with the drawdown now being referred to as
the falloff. Because both the drawdown/falloff and the flow rate switch signs, the II
always has a positive value as well.
Fig. 6.1 depicts the linear IPR for a single-phase oil well. The corresponding
reservoir parameters have been given in Table 6.1, and their relationship to the IPR is
discussed in detail in Section 6.5. At a flow rate qo,sc = 0 the FBHP pwf equals the
static BHP pR. In the theoretical case of zero pressure at the bottomhole, the flow
rate would reach a value known as the absolute open-flowing potential (AOFP) of the
well. The assumption of a linear PI is justified for single-phase oil or liquid (oil/water)
flow and realistic drawdowns. (Note that typical values for drawdown in an oil well are
between 0.01 and 5 MPa. At much higher values the pressure dependency of the
viscosity will, in theory, cause a nonlinear IPR for single-phase oil. Usually, such
higher drawdowns will correspond to a FBHP below the bubblepoint pressure, in
which case the IPR becomes nonlinear anyway because of multiphase effects.)
For gas wells, or for oil wells producing from a reservoir below bubblepoint
pressure, the IPR is a nonlinear function of the flow rate and can no longer be
represented with a straightline PI.

6.3 Well Operation and Reservoir Flow Stages

During the primary recovery phase of an oil reservoir—i.e., the period when oil is
produced because of reservoir pressure depletion only—wells can be produced at
either a constant flow rate and a gradually decreasing FBHP, a constant FBHP and a
gradually decreasing rate, or a combination of decreasing rate and FBHP (Walsh and
Lake 2003). Constant-rate operation (“plateau production”) is typical for the early life
of a well, when surface facility constraints are often limiting the flow rate. Because of
reservoir pressure depletion, maintaining the plateau requires a gradual decrease of
the FBHP. After reaching the minimum allowable FBHP—i.e., the FBHP required to lift
the wellbore contents to surface—a constant-FBHP operation results. During the
secondary recovery phase, when either water or gas is injected to maintain the
reservoir pressure, it is also possible to operate a well at a (near-)constant rate and
FBHP.
Inside the reservoir, the following different stages of flow can be distinguished
during primary recovery (Walsh and Lake 2003):
Fig. 6.1—Straightline inflow performance relationship for an oil reservoir with parameters given in Table
6.1. The operating point is indicated with a dot. The corresponding FBHP and flow rate are 4.00 × 106 Pa
(580 psi) and 1.09 × 10−3 m3/s (593 STB/D), respectively. The AOFP is 2.18 × 10−3 m3/s (1187 STB/D). The
drawdown Δp = 8.00 × 106 − 4.00 × 106 = 4.00 × 106 Pa (540 psi). The productivity index J = 2.18 × 10−3/8.00
× 106 = 2.76 × 10−10 m3/s Pa (1.03 STB/D-psi).

Parameter SI Units Field Units

Reservoir height h 20 m 65.6 ft
Reservoir radius re 500 m 1,640 ft
Well radius rw 0.20 m 0.65 ft
Permeability (Fig. 6.1) ko 0.35 × 10−13 m2 35 mD
(Other figures) 1.00 × 10−13 m2 101 mD
Viscosity (Fig. 6.1) μo 1.9 × 10−3 Pa·s 1.9 cp
(Other figures, analytical)* 5.8 × 10−3 Pa·s 5.8 cp
(Other figures, numerical) Beggs and Robinson (1975); Vasquez and Beggs (1980)
Gas density ρg,sc/γ g 0.95 kg/m 3 0.77
Oil density ρo,sc/γ API 850 kg/m 3 35 °API
Sol. gas/oil ratio Rsb 0 m 3/m 3 0 scf/STB
Oil FVF (Fig. 6.1) Bo 1 m 3/m 3 0 bbl/STB
(Other figures, analytical) 1.03 m 3/m 3 1.03 bbl/STB
Reservoir pressure pR 8 × 106 Pa 1,160 psi
Res. temp. (Fig. 6.1) TR 120 °C 248 °F
(Other figures) 60 °C 140 °F
Skin S 0 – 0 –
Flow regime Steady-state
* Computed with the aid of the same correlations as used for the numerical solution with p = pR and T = TR.
Table 6.1—Parameter values for several figures in Chapters 6 and 7.

1. Transient flow, also known as infinite-acting flow. After opening up a well, a

pressure wave will radially travel into the reservoir. Transient flow occurs until
the wave approaches either a reservoir boundary or a pressure wave of a
neighboring well. This stage typically takes several hours to tens of days for
oil reservoirs, and up to several months for gas reservoirs.
2. Transitional flow. This occurs once the first boundary effects are felt by the
pressure wave, until the effects of all boundaries surrounding the drainage
area around a well have exerted their influence on the transient flow. This
stage is typically also relatively short (in the order of days) for oil wells and
much longer for gas wells.
3. Semisteady-state flow. This next stage may last for many years. True
semisteady-state flow requires a constant production rate and no inflow
through the boundaries of the drainage area. That is, there should be no
aquifer influx, and all neighboring wells should be produced at constant rates
such that the drainage area of each well remains stationary. In that case, the
pressure drops uniformly over the entire reservoir while the flow rates remain
constant.
4. Depletion flow. This stage starts when the FBHP reaches its minimum
allowed pressure and the well starts to be operated at a constant FBHP. As a
result both the reservoir pressure, away from the well, and the well flow rates
gradually decrease.

During secondary recovery, or in the case of primary recovery in the presence of a

very strong aquifer, the reservoir pressure may remain (near-)constant, in which case
it as also possible to obtain a fifth stage of flow:

5. Steady-state flow. Depending on the properties of the different phases there

may occur slow changes over time in the reservoir pressure, the FBHPs, and
the flow rates in the wells and the reservoir, corresponding to the movement
of the oil/ water and/or oil/gas fronts through the reservoir. In this case we
may consider the reservoir to be in near-steady-state condition with slowly
varying parameters.
It should be noted that these classifications are highly idealized and in reality a
reservoir is nearly always in a continuous pressure transient situation. The closing in
and opening up of wells for maintenance, drilling new wells or sidetracks, and the
effects of oil/ water or oil/gas fronts that slowly travel through the reservoir, all result
in a pressure distribution that continuously changes in space and time. Nevertheless,
the analysis of (semi)steady-state flow around a wellbore provides much insight and
is often a sufficient approximation to form the basis for useful production engineering
models.

6.4 Governing Equations

6.4.1 Mass Balance, Momentum Balance, and Equation of State. In this section
we derive the equations for single-phase fluid flow in the near-wellbore region using
the same approach used to describe pipe flow in Chapter 3. Consider the classic
textbook case of a single vertical well, either openhole or perforated over the entire
reservoir height, producing from a circular reservoir (see Fig. 6.2). (The first book
covering this material in depth is the truly classic text by Muskat (1937). It develops
analytical solutions for a wide variety of well configurations and, despite its age,
anyone interested in analytical expressions to describe inflow performance should
consult this book.) It is assumed that the height of a reservoir is small compared to its
lateral dimensions such that the flow can be approximated to be 2D. Moreover, the
radial symmetry of the configuration allows for the use of polar coordinates. We can
then write the mass balance per unit time through a wedge-shaped control volume as
(see Fig. 6.3):

Fig. 6.2—Well in a circular reservoir (with vertical dimension strongly exaggerated).

Fig. 6.3—Control volume and pressure forces in polar coordinates.

where A = hrdψ is the cross-sectional area of the control volume in radial direction,
m2; h is reservoir height, m; r is the radial coordinate, m; ψ is the tangential
coordinate, radians; ρ is fluid density, kg/m3; v = q/A is superficial radial fluid velocity,
m/s; q is the radial flow rate, m3/s; ϕ is porosity, –; and t is time, s. Note that the true
fluid velocity v is in general much higher than the superficial velocity because the
cross-sectional area A is only partially open to flow, depending on the porosity ϕ.
Furthermore, the true fluid velocity field is 3D because the pores form a tortuous 3D
network.
A positive velocity implies flow in the positive coordinate direction and therefore
corresponds to injection from the well into the reservoir. Maintaining the analogy with
pipe flow, the momentum balance can formally be written as

where the components of the pressure term have been illustrated in Fig. 6.3 and
where p is pressure, Pa; Fg(ρ,s) is gravity force per unit length, N/m; Ff (ρ,μ,v) is
friction force per unit length, N/m; and μ is dynamic viscosity, Pa·s. However, in flow
through porous media the velocities v are usually so small that the momentum terms
at the left side, which depend on v2, play no role. Furthermore, it also can be shown
that the momentum term at the right side is negligible and that therefore only the
pressure, gravity, and friction terms need to be taken into account (see, e.g., Bear
1972). In our case, we can furthermore disregard gravity because we consider
horizontal flow only. The nature of the friction force Ff (ρ,μ,v) is discussed in more
detail in Section 6.4.2.
Just as in the case of pipe flow, we can complete the set of governing equations
with the aid of the equation of state (EOS) for the fluid (i.e., Eq. 2.9 for gas or Eq.
2.18 for oil). If we expand Eqs. 6.2 and 6.3, substitute A = hrdψ, disregard
momentum and gravity, drop all terms higher than first order in the differentials, and
simplify the results, we can write the three equations as

where the oil compressibility co and the gas deviation factor Z are known functions of
p and T. (We will not treat the case of single-phase water flow separately because it
follows directly from the single-phase oil case by replacing the fluid properties for oil
by those for water. Note that unlike in pipe flow, a two-phase liquid mixture—i.e.,
oil/water—in a porous medium cannot be described with averaged liquid properties; in
porous media, the presence of water has a considerable effect on the flow of oil and
vice versa.)
The temperature T (and therefore also Tabs ) can generally be taken as constant
because the large heat capacity of the reservoir is usually sufficient to guarantee
isothermal conditions. Only in high-rate gas wells may some cooling because of
expansion of the gas occur in the near-wellbore region, an effect known as the Joule-
Thomson cooling (see, e.g., Moran and Shapiro 1998).

6.4.2 Friction Force: Darcy’s Law. The friction loss for single-phase liquid flow in
porous media is described by the experimental relationship known as Darcy’s law,
which can be written in polar coordinates as
or, with an eye on Eq. 6.5, as

where k is rock permeability, expressed in m2 in SI units or millidarcy (md) in field

units. For single-phase liquid flow and a homogeneous reservoir, k is a constant.
More generally, the permeability may have different values for flow in different
directions (see Section 7.2.5). However, for multiphase mixtures, the permeability for
each phase will usually be a nonlinear function of the phase saturations, i.e., the
proportion of each phase present in the pores). Moreover, the permeabilities may be
path dependent (i.e., they may depend on the saturation history in the pores. (The
nature of this nonlinear behavior is discussed in more detail in Appendix F.) For gas
flow, we can use a similar relationship as for liquid flow except for high velocities such
as occur in the near-wellbore region of high-rate gas wells. In that case we have to
replace Darcy’s law with a nonlinear equivalent for which several forms have been
proposed (see Bear 1972). A frequently used relationship is given by

where β is Forchheimer’s coefficient with dimension L–1 (Forchheimer 1901). It

represents the inertia effects experienced by gas when it is accelerated and
decelerated during its flow through the pore throats. It is also referred to as inertia
coefficient or turbulence coefficient. The latter name is not entirely correct because
the inertia effect can be noticed at much lower velocities than the velocity that
corresponds to the onset of turbulence in the pores (see, e.g., Bear 1972). The
values of k and β should be determined experimentally, either directly through
measurements on cores or indirectly from well tests. The relationship between β and
k is typically of the form (Dake 1978)

with k expressed in m2 and β in 1/m (which implies that the dimensional constant A
has units m2B−1).

6.5 Single-Phase Oil Flow

6.5.1 Steady-State Flow. At steady-state conditions the right side of the mass
conservation equation (Eq. 6.4) vanishes, which reduces it to
This trivial first-order differential equation requires one boundary condition, which can
be obtained from the known density and velocity at the wellbore radius:

where we used Eqs. 2.28 and 2.31 with Rs = Rsb . Following an approach similar to
the one used for single-phase pipe flow in Section 3.3, we can now solve Eq. 6.12
and determine the integration constant with the aid of the boundary condition (Eq.
6.13); combine Eqs. 6.5, 6.7, and 6.9; and use Eqs. 2.28 and 2.31 (with Rs = Rsb )
and Eq. 2.38 to arrive at the following set of equations for single-phase steady-state
oil flow in the near-wellbore region:

where we have dropped some of the subscripts o to improve the readability. This set
of differential algebraic equations strongly resembles Eqs. 3.24 through 3.26 as
defined in Section 3.3 to describe the steady-state flow of oil in pipes. The equations
are nonlinear because ρ and μ are functions of the unknown pressure p. The
differential equation (Eq. 6.14) requires one boundary condition, for which we can use

which represents a constant pressure pR at the external boundary of the circular

drainage area. This situation can, with some imagination, be interpreted as a
reservoir with constant pressure support in the form of a strong aquifer.
Just as was done for pipe flow, the equations can be solved with the aid of a
standard numerical integration routine in MATLAB, and a numerical implementation can
be found in res.m and res_oil_dpdr. An example of how to use these files to plot
pressure vs. radial distance has been programmed in example_res_pres.m. A typical
result is shown in Fig. 6.4. It displays a steep decrease in pressure close to the
wellbore, caused by a radial convergence of the streamlines, which require an
increasingly large pressure drop dp/dr to force the oil through an increasingly smaller
surface area 2πhr.

6.5.2 Analytical Solution. For oil above the bubblepoint pressure, for realistic
drawdowns, it is reasonable to assume that Bo and μ are constants (with a typical
value of Bo slightly below Bob ; see Fig. 2.6). Eqs. 6.14 through 6.16 can then be
combined to give the classic linear differential equation for steady-state single-phase
radial oil flow near a well (Muskat 1937):

Fig. 6.4—Pressure as function of radial distance from a well in a reservoir with parameters given in Table
6.1 and a flow rate qo,sc = 1.09 × 10−3 m3/s (593 STB/D). The numerical result for the FBHP is 4.43 MPa (643
psi); the analytical result 3.95 MPa (572 psi).

Eq. 6.18 can be integrated analytically to give

The value of the unknown integration constant C can be found with the aid of the
boundary condition (Eq. 6.17), and substitution in Eq. 6.19 then gives us the
expression for p as a function of r under steady-state flow conditions:
Recall that qo,sc ≤ 0, such that p(r) ≤ pR.
The FBHP for a given flow rate now simply follows by substituting r for rw and p for
pwf. Reorganizing the result such that it becomes an expression for the drawdown as
a function of the oil flow rate gives

Note that because of our definition of the positive flow direction, a positive drawdown
corresponds to a negative flow rate (i.e., to flow toward the well as occurs in a
production well). We can now also define the PI, relating the flow rate to the
drawdown as

A numerical implementation of Eq. 6.22 can be found in res_oil_simp.m. Just like its
numerical counterpart, res.m, it can be used to create a plot of pressure vs. radial
distance from the well (see example_res_pres.m). The corresponding result has
been plotted as a dashed line in Fig. 6.4.
Although at first sight the analytical and numerical results appear to be very close,
there is an increasing discrepancy with increasing proximity to the wellbore,
culminating in a difference of approximately 12% in the FBHP. The reason is the
pressure dependency of the density and viscosity, which has not been taken into
account in the analytical model. However, the drawdown in this example is at the high
end of what could be expected in real life. In nearly all circumstances it is fully justified
to use the analytical solution to predict a well’s inflow performance because
uncertainties resulting from reservoir heterogeneity have a much larger influence on
the inaccuracy of the prediction than the choice of the solution method.

6.5.3 Average Reservoir Pressure. Alternatively, we may want to express the IPR
in terms of the volume-averaged reservoir pressure pR,av defined as

Substitution of Eq. 6.20 in Eq. 6.23 gives

where the integral can be solved through integration by parts as follows:

Substitution of this result in Eq. 6.24 gives

where the approximation holds for rw << re, which is the usual situation. (Recall that
qo,sc is negative and therefore pR,av < pR, as expected.) Often the difference between
pR,av and pR is relatively small. Solving for pR from Eq. 6.26, substituting in Eq. 6.23,
and then rearranging the result gives an expression for the drawdown in terms of
average pressure:

6.5.4 Semisteady-State Flow. Most reservoirs are initially produced through

pressure depletion (primary recovery). Moreover, even during secondary recovery the
injected volumes may not be sufficient to maintain a constant pressure, such that it
gradually drops over time. Such a gradual pressure depletion scenario can be
schematically represented by a circular reservoir with a boundary condition:

which implies that there is no pressure gradient and therefore no driving force for flow
at the external boundary. This type of no-flow condition typically occurs when a large
number of vertical wells producing at constant rates are used to drain a reservoir in a
regular pattern. The drainage areas can then be approximated reasonably well by
circular cylindrical volumes. A refined approximation can be obtained with the aid of
shape factors or equivalent outer radii to account for the fact that the drainage areas
are not exactly circular (see Dietz 1965). As a consequence of the absence of flow
through the outer boundaries, and of constant oil production rates from the wells, the
pressure in the reservoir will steadily decrease, a situation known as semisteady
state (also known as pseudosteady state).
To analyze this situation, we can start from the mass balance (Eq. 6.4), which can
be rewritten with the aid of the EOS (Eq. 6.7) as follows:
Under semisteady-state conditions the pressure derivative ∂p/∂t should remain
constant— say, equal to an unknown constant C1. The mass balance equation
therefore reduces to

If we maintain the assumption that ρ is constant, the first-order differential (Eq. 6.30)
can be integrated analytically to give

We know from the boundary condition (Eq. 6.28) that v = 0 at r = re, which allows us
to express C1 in terms of C2. With the aid of the other boundary condition (i.e., Eq.
6.13), which was also used for the steady-state solution, we can now derive

Similarly to what we did in the steady-state situation, we may combine Eqs. 6.5, 6.9,
and 6.32 to arrive at the differential equation for semisteady-state radial oil flow:

Integration of the equation results in

We can use the boundary condition (Eq. 6.17) (but now for a slowly decreasing
pressure pR) to solve for the integration constant C3, and substitution of the result in
Eq. 6.34 gives an expression for p as a function of r under semisteady-state flow
conditions:

In particular, the drawdown under semisteady-state conditions can now be expressed

as
 The difficulty with Eqs. 6.35 and 6.36 is that the pressure pR at the external
boundary, which is gradually decreasing, just as the pressure in any point of the
reservoir is doing, can usually not be determined. However, the average pressure
pR,av can often be determined from the pressure response in a well after shut-in, a
well-testing procedure known as pressure transient analysis (see, e.g., Dake 1978
or Kamal 2009). To express Eq. 6.36 in terms of the average reservoir pressure, we
can use the same approach as in Section 6.5.3, resulting in

(For a derivation of this expression, see Question 6.7 in Section 6.11 and its answer
in Appendix G.)

6.5.5 Combined Expression. Eqs. 6.21, 6.27, 6.36, and 6.37 can be combined in a
single expression:

where fR and pR,ref are given by

The corresponding combined expression for the PI becomes

For many practical situations, the value of ln(re/rw) in Eqs. 6.38 and 6.40 will be
between 5 and 10, and the difference between steady state or semisteady state, and
between drawdown with respect to reservoir pressure or average reservoir pressure,
becomes unimportant.
Another noteworthy feature of these equations is that the radii rw and re occur
within a logarithm. For typical parameter values the sensitivity of J to changes in the
radii is therefore relatively small compared to, e.g., changes in k. This is particularly
relevant for the sensitivity to uncertainties in the outer radius re, because circular
reservoirs do not exist and there is usually considerable uncertainty about the shape
and “equivalent” radius of the outer boundary. Fortunately, changes in the value of re
do not strongly influence the results. A downside of this feature is that increasing rw
(i.e., drilling a larger-diameter well) only results in a relatively small increase in
production (see also Question 6.6).

6.6 The Skin Factor

6.6.1 Formation Damage. The IPRs derived above assume that the radial
permeability is everywhere constant. In practice, this is not the case. In addition to
geological heterogeneity, the well may be impaired. During the drilling of the well
there is penetration of alien fluids into the reservoir rock, which may reduce the
permeability of the rock around the well and therefore reduce the rate of oil inflow.
This reduction in permeability is called formation damage or impairment. We
currently have a good understanding of the fundamental causes of formation damage,
thanks to experimental and theoretical research over the past years. (For an
extensive overview, see Civan 2000.) From the moment the drill bit first penetrates
the reservoir section until the well is put production, the reservoir rock is exposed to a
series of operations that can cause damage:

• Mechanical interference. The drilling itself can create mechanical damage, with
pore collapse and particle rearrangement.
• Solids. Solids come into contact with the rock formation, such as drill cuttings,
solid material added to the drilling mud, or metal debris. If small enough, the
solids can be swept into the formation and block the pores. If larger, the solid
particles cannot enter into the rock pores but are deposited at the borehole
wall. Some of these solids will be swept away again when the well is put on
production, but not all; Fig. 6.5 shows a thin layer of residual mud solids on the
borehole wall.
• Fluids. Fluids used in well construction can also cause formation damage. Such
fluids are composed of water, oils, salts, acids, surfactants, and many other
chemicals. They may interact with the reservoir rock and fluids, causing
detachment of fine particles, flocculation, wettability change, precipitation,
emulsion formation, or fluid saturation changes. In particular, the pores may be
lined with clay, which may swell disastrously, completely blocking the pores.
• Phase changes. Changes in pressure and temperature in the oil and water
may result in phase changes, with precipitation of waxes, asphaltenes, or
 scale, which are deposited in the pores.
• Microbial damage. Last, microbes introduced into the well, or possibly
indigenous in the reservoir in a dormant state, may multiply, forming deposits in
the pores.

The effect of all these damage mechanisms is to reduce the permeability of the
reservoir rock over a relatively small region around the wellbore. This small damaged
region is called the skin of the well (Van Everdingen 1953). This skin gives rise to an
additional pressure drop, as shown in Fig. 6.6, so that the well produces less than
expected. The additional pressure drop can be taken into account in the IPR as
follows. Eq. 6.38 can now be modified to give

6.6.2 Combined Expression. Introducing the dimensionless skin factor S defined

through

Fig. 6.5—Residual mud solids.

Fig. 6.6—The additional pressure drop caused by skin, at a given flow rate.

we can rewrite the combined expression for the drawdown (Eq. 6.38) as

where

The corresponding combined expression for the PI becomes

A numerical implementation of the general single-phase oil pressure drop Eq. 6.43
can be found in res_oil_simp.m.

6.6.3 Thick Skin. Alternatively, the skin can be taken into account by simulating the
behavior of two or more rings of the near-well reservoir in series, using different
permeabilities for the damaged and the undamaged zones (Muskat 1937; Hawkins
1956). Fig. 6.7 represents an example using two rings. The corresponding pressure
drop can be expressed as

where we used the analytical approximation for steady-state oil flow (Eq. 6.21),
where ks is the reduced permeability, rs is the external radius of the damaged zone,
and ps is the pressure at rs. Using the definition for the skin S (Eq. 6.42), we can
derive

which is sometimes referred to as the finite skin or thick skin approach. The value of
the skin S can be determined from transient well tests (see, e.g., Dake 1978,
Economides et al. 2013, or more specialized books on well testing such as Kamal
2009). If the value of the skin is high, then remedial measures may be required—e.g.,
stimulating the well with acid to repair the damage. If a well is tested, it may appear
that the skin S is nonzero. However, this may not be a result of formation damage but
of the completion. If the well is gravel packed, the permeability of the gravel is likely
to be greater than that of the reservoir rock. Thus the gravel pack may give less
pressure drop, resulting in negative skin. On the other hand, the gravel pack itself can
be heavily impaired during installation or subsequent production, and positive skin
could result. Perforations can give rise to negative skin if they provide an effective
path for the oil to flow into the well. Often, they contain debris from shooting the
perforations and have a zone of crushed rock around them, both effects contributing
to positive skin. Fractures, whether natural or produced by hydraulic fracturing, will
result in easier inflow and thus negative skin.

6.7 Single-Phase Gas Flow

6.7.1 Numerical Solution. In the case of single-phase gas flow the mass
conservation (Eq. 6.4) can be rewritten for steady-state conditions as
Fig. 6.7—Schematic representation of a damaged near-well reservoir (left) as two pressure drop elements
in series (right).

The corresponding boundary condition at the wellbore radius becomes

We can now solve Eq. 6.48 and determine the integration constant with the aid of the
boundary condition (Eq. 6.49) in the usual fashion. Next we can combine Eqs. 6.5 and
6.10, and together with the EOS for gas (Eq. 6.7), we arrive at the following set of
equations to describe single-phase gas flow in the near-wellbore region:

where we have dropped most of the subscripts g to improve the readability. This set
of equations resembles Eqs. 3.33 through 3.35 as defined in Chapter 3 to describe
the flow of gas in pipes. The equations are nonlinear because ρ, μ, and Z are
functions of the unknown pressure p and because the pressure drop is a quadratic
function of the velocity. The differential equation (Eq. 6.50) requires one boundary
condition, for which we can use Eq. 6.17. As was the case for single-phase oil flow,
the equations can be solved with the aid of a standard numerical integration routine in
MATLAB, and a numerical implementation can be found in res.m. and res_gas_dpdr.m.
The IPR can then be obtained by plotting pwf as a function of qg,sc for a given pR (see
example_res_pres.m). The solid and dotted lines in Fig. 6.8 depict an example based
on the data in Table 6.2 for values of the Forchheimer coefficient β = 0 and β = 3.5 ×
1010 1/m, respectively.

6.7.2 Analytical Approximations. Alternatively, the equations can be linearized

through the use of a real-gas pseudopressure defined as

where pref is an arbitrary reference pressure. In this formulation the pseudopressure

has the dimension of pressure. Often other formulations are used, without the scaling
factor , such that the pseudopressure has a different dimension. (See Al-
Hussainy et al. 1966 for the original paper and Hagoort 1988 for practical details.)
The linearization can be obtained as follows. First we combine Eqs. 6.50 through
6.52, while dropping the Forchheimer term, to obtain the pressure drop equation

Fig. 6.8—Nonlinear IPR for a gas reservoir with parameters given in Table 6.2.

Parameter SI Units Field Units

Reservoir height h 20 m 65.6 ft
Reservoir radius re 500 m 1,640 ft
Well radius rw 0.2 m 0.65 ft
Permeability kg 2.0 χ 10−15 m2 2.0 mD
Forchheimer coeff. β 3.5 × 1010 1/m 8.9 × 107 1/in.
Viscosity μg Carr et al. (1954) correlation
Density/gravity ρg,sc/γ g 0.95 kg/m 3 0.77 –
Reservoir pressure pR 40 × 106 Pa 5,802 psi
Reservoir temperature TR 120 °C 248 °F
Skin S 0 – 0 –
Flow regime Steady-state
Table 6.2—Parameter values for Fig. 6.8.

which is nonlinear because of the dependence of μ and Z (and therefore of Bg) on p.

Next we use the relationship

and Eq. 6.53 to obtain

and therefore

Eq. 6.54 can now be rewritten as

which is now a linear equation in terms of the pseudopressure m(p). Here, in line with
assumptions made before, we assume that the reservoir temperature Tabs remains
constant.
For a given gas composition and the corresponding relationships for μg and Z as
functions of p, we can determine a one-to-one relationship between p and m through
numerically integrating Eq. 6.53. We can then simply use all the results that were
obtained for single-phase oil flow, replacing p with m and changing the constant oil
properties to the corresponding constant gas properties. In other words, the general
single-phase oil equation (Eq. 6.43) can now be used to obtain a corresponding
single-phase gas equation:

with fR and pR,ref given by Eq. 6.39. The value of the arbitrary reference pressure pref
— not to be confused with pR,ref—is not relevant to the computation of the
pseudopressure because we are interested only in pressure differences, and not in
absolute pressures.
Fig. 6.9 depicts an example of the relationship between pressures and
pseudopressures. The graph was produced with the aid of MATLAB files
gas_pseu_pres.m and gas_dmdp.m. We refer to Dake (1978) or Hagoort (1988) for
further information about the use of pseudopressures for gas engineering
calculations. For an extension of the concept of pseudopressures to two-phase
(gas/oil) flow, see Walsh and Lake (2003).
If the difference between the reservoir pressure pR and the FBHP pwf is not too
large, we can use constant average values μav and Zav, evaluated at an average
reservoir pressure pR,av, in the definition of the pseudopressure. If we now choose pref
= pR,av, integration of Eq. 6.53 shows that the pseudopressure reduces to

Formally, the average viscosity is generally not equal to the viscosity evaluated at the
average pressure because μ(p) is a nonlinear function. However, given the
approximate nature of the model, the simplification μav ≈ μ(pav) is justified. The same
holds for the average compressibility factor Zav ≈ Z(pav).
Substitution of Eq. 6.60 in the general single-phase gas expression (Eq. 6.59)
results in

or

Eq. 6.61 can also be obtained without the use of pseudopressures (see, e.g., Russell
et al. 1966). Note that we still have the freedom to choose pR,ref as either pR or pR,av,
with corresponding values of fR as given in Eq. 6.39. The choice of pR,ref = pR,av will be
useful if the average reservoir pressure can be obtained from a well test. If we
choose pR,ref = pR, the average reservoir pressure can be approximated with the aid
of Eqs. 6.26 and 6.60 as

which can then also be used to compute the average values for the gas properties.
(The full expression is obtained by solving the quadratic equation in pR,av. The
approximation then follows by using a Taylor expansion for the square-root term.) A
few iterations may be required to determine the average properties to sufficient
accuracy.

Fig. 6.9—Pseudopressure as a function of pressure for hydrocarbon gas with density ρg,sc = 0.95 kg/m3 (γg
= 0.77), at temperature T = 75°C (167°F) and pressures 0.1 < p < 30 MPa (14.7 < p < 4,351 psi).

Note that this result is valid only for a Forchheimer coefficient β equal to zero,
because the nonlinear term βρν2 was not taken into account in the derivation. For high
flow rates (i.e., when β cannot be neglected) we need to add an additional pressure
drop term (for details see, e.g., Dake 1978). This approximate inflow relationship
(Eq. 6.61) has been implemented in MATLAB file res_gas_simp.m, which was used to
create the approximate IPR indicated with a dashed line in Fig. 6.8. The difference
with the numerical result, determined with res.m and indicated with a solid line, is not
very large even though the drawdown is considerable. However, at higher flow rates
(for | qg,sc | > 9 m3/s) the iteration became problematic and no solutions were found. A
semi-empirical generalization of Eq. 6.61, known as the backpressure equation, can
be written as

where the constants C and n have to be determined from measurements of gas flow
rates and bottomhole pressures during a multirate production test. A typical value for
n is between 0.5 and 1.

6.8 Multiphase Flow: Empirical Models

Starting from a detailed geological model and measured reservoir and fluid
properties, and using numerical reservoir simulation, one could, in theory, obtain a
detailed time-varying prediction of the pressure and oil, gas, and water distributions
around a well. In practice, predicting the inflow performance of a well under
multiphase flow conditions is possible only to a limited extent. However, after a well
has been put on stream, we may use data collected during multirate well tests. These
can then be used to fit a PI or a nonlinear IPR, which may have some predictive
capacity, although extrapolations outside the range of water cuts and gas/oil ratios
(GORs) actually observed are still of limited value. Here we first present some
examples of such semiempirical IPRs. In Chapter 7, we present a somewhat more
physics-based approach to illustrate the effects of the presence of gas in the near-
wellbore region around an oil well.

6.8.1 Solution Gas Drive. For primary production of a reservoir below the
bubblepoint pressure, several empirical relationships have been established, directly
expressed in terms of oil flow rate and FBHP. They are similar to the backpressure
equation (Eq. 6.64) for gas wells (which is also sometimes used for oil wells). We
mention the ones proposed by Vogel (1968) and Fetkovich (1973), which are valid for
gas/oil flow from a reservoir produced by solution gas drive:

Here, qo,sc,max is the AOFP, i.e., the value of qo,sc when pwf is zero (see Fig. 6.10, left).
Note that this zero pressure can normally not be achieved in practice. The reference
reservoir pressure pR,ref may be chosen as either pR or pR,av. Eqs. 6.65 and 6.66 both
contain two parameters (α in Vogel’s, n in Fetkovich’s, and qsc,max in both) and
therefore need at least two sets of pressure and flow rate measurements from a
multirate well test. Based on a large number of numerical simulations, Vogel (1968)
recommends choosing α = 0.2. In that case at least one single-rate well test is
required to estimate qsc,max. Fetkovitch (1973) reported values of n between 0.57 and
1.00, as observed in a large number of well tests. The shapes of the IPRs
corresponding to Eqs. 6.65 and 6.66 are similar to those for gas flow depicted in Fig.
6.8. We refer to Golan and Whitson (1991) and Beggs (1991) for examples of how to
determine empirical IPRs from production tests.

6.8.2 Predictive Mode. To use Eqs. 6.65 and 6.66 in a predictive mode, it is
necessary to guess the two unknown parameters (i.e., α and qsc,max, or n and qsc,max).
Alternatively, we can use a theoretical estimate of the PI for small flow rates to
eliminate one of the two unknowns and use a best guess for the remaining one. For
this approach we make use of the observation that for small flow rates the IPR is
close to linear such that we can define an approximate initial PI as

which for the Vogel relationship (Eq. 6.65) leads to

such that we can express the AOFP as

Fig. 6.10—Schematic solution gas drive IPRs for two oils with bubblepoint pressures above the (average)
reservoir pressure (left) or below it (right).
(Recall that in our convention qo,sc,max < 0.) At the start of production from a solution
gas drive reservoir, the value of J* for pressures at the bubblepoint follows from Eq.
6.45, where the permeability to oil should be taken as . After production has
resulted in a drop in reservoir pressure and thus an increase in gas saturation and a
reduction in oil permeability, the value of J* may be approximated as

where ko,ref is the (saturation-dependent) permeability to oil, and μo,ref and Bo,ref are the
(pressure-dependent) oil viscosity and the oil formation volume factor, respectively
(Standing 1971). The pressure-dependent parameters μo,ref and Bo,ref can be
approximated by using their values evaluated at pR,ref. A crude estimate of the
saturation-dependent oil permeability ko,ref can be obtained by assuming that the oil
saturation decreases linearly with decreasing reservoir pressure such that

where Sob = 1−Swi is the oil saturation at (and above) the bubblepoint pressure. An
approximate predictive expression for pwf as a function of qo,sc under primary two-
phase (gas/oil) production now follows by solving for pwf from Eq. 6.65, which leads to

where qo,sc,max should be determined with the aid of Eqs. 6.69 and 6.45 or 6.70, and
where, following the original publication of Vogel (1968), the value of α should be
taken as 0.2. As before, the reference reservoir pressure pR,ref may be chosen as
either pR or pR,av.

6.8.3 Initial Pressure Above Bubblepoint. For a reservoir that has a reservoir
pressure above the bubblepoint pressure, but for which the FBHP drops below it, we
can derive a modified Vogel expression with the aid of Fig. 6.10 (right) as
where qob ,sc = J (pb − pR,ref) is the (negative-valued) flow rate at the bubblepoint
pressure, with J given by Eq. 6.45, and where the AOFP should now be determined
as

Eq. 6.73 can be rewritten in a similar form as Eq. 6.72, which leads to

with qo,sc,max given by Eq. 6.74. Eq. 6.75 has been implemented in MATLAB file res
Vogel.m.
Fig. 6.11 displays two typical Vogel-type IPRs for oils with different solution GORs
and reservoir parameters given in Table 6.3.
Fig. 6.11—Vogel-type IPRs of an oil well with parameters given in Table 6.3 for two different values of the
solution GOR as indicated in the legend. The corresponding bubblepoint pressures are 9.00 MPa (1,306
psi) and 22.7 MPa (3,288 psi) respectively (arrow indicates point at which the low-GOR curve drops below
bubblepoint pressure).

Parameter SI Units Field Units

Water cut fw 0 – 0 –
Reservoir height h 20 m 65.6 ft
Reservoir radius re 500 m 1,640 ft
Well radius rw 0.2 m 0.65 ft
Permeability k 1.11 × 10−13 m2 101 mD
Forchheimer coeff. β 3.5 × 1010 1/m 8.9 × 107 1/in.
Endpoint gas 0.7 – 0.7 –
Endpoint oil 0.9 – 0.9 –
Endpoint water 0.5 – 0.5 –
Corey gas ng 3 – 3 –
Corey oil/gas nog 3 – 3 –
Corey oil/water now 3 – 3 –
Corey water nw 3 – 3 –
Critical gas sat. s gc 0.15 – 0.15 –
Residual oil sat. Sor 0.10 – 0.10 –
Immobile water Swi 0.15 – 0.15 –
saturation
Oil rel. perm. k ro Modified Stone II model; Aziz & Settari (1979)
Gas viscosity μg Carr et al. (1954) correlation
Oil viscosity μo Beggs and Robinson (1975), Vasquez and Beggs (1980)
Gas density/gravity ρg,sc/ γ g 0.95 kg/m 3 0.77 –
Oil density/gravity ρo,sc/ γ API 850 kg/m 3 35 °API
Water density/gravity ρw,sc/ γ w 1050 kg/m 3 1.05 –
Equations of state Standing (1952), black oil; Standing and Katz (1942) Z factor
Reservoir pressure pR 15.0 × 106 Pa 1,378 psi
Reservoir temp. TR 90 194 °F
Skin S 0 – 0 –
Flow regime Steady-state
Table 6.3—Parameter values for Figs. 6.11 and 7.21.

6.9 Multilayer Inflow Performance

If a well is completed in more than one layer, the production from the well is called
commingled. If the layers are separated by impermeable flow barriers, and if they
have different permeabilities, differential depletion may occur, resulting in gradually
increasing differences in the reservoir pressures of the layers. (Here, we disregard
the differences in hydrostatic pressure resulting from depth under the assumption that
the differences resulting from differential depletion are much larger.) In that case the
combined inflow performance can readily be calculated provided that the individual
IPRs are both linear.
As shown in Fig. 6.12, for a given value of pwf, the total production rate qo,sc can be
determined by calculating the individual contributions qo,sc,1 and qo,sc,2, and adding them
as follows:

Thus the production rate still varies linearly with pwf. Note however that this formula
applies only to values of pwf lower than the lowest of the two zone pressures. Above
this pressure, part of the production from one zone will be injected into the other
zone. This phenomenon, which is often referred to as crossflow, is illustrated in Fig.
6.13. If the well is closed in at surface, a steady-state situation will develop in which
equal amounts are produced from and injected into the respective reservoir units.
Crossflow can seriously impair the reservoir into which the injection takes place. A
similar reasoning can be applied in case of nonlinear IPRs (see Question 6.10). For
an example of controlled, commingled production from a stacked reservoir, see the
MATLAB assignment in Section 6.12.

6.10 Related Topics Not Considered in This Chapter

Many aspects influencing near-wellbore flow have not been considered in this
chapter, such as the effect on inflow performance of well deviation, permeability
anisotropy, partially penetrating wells, hydraulic fractures, perforation patterns,
washed-out well sections, gravel packs, slotted liners, and wire-wrapped screens.
We also did not discuss methods used to compute transient reservoir flow, which is
important during the early stages of production. Some of these topics will be
discussed in Chapter 7, and for some others we refer to the literature. Many
publications are available describing pressure drop over specific completion elements
such as perforations, slotted liners, or screens. Nearly always the pressure drop over
these elements can be considered to be in series with the near-well reservoir
pressure drop, such that they may simply be added. For further reading on the topic
of inflow performance, see, e.g., the textbooks of Brown (1984), Beggs (1991),
Golan and Whitson (1991), and Economides et al. (1998, 2013) as well as the more
specialized references mentioned in Chapter 7.
Fig. 6.12—Two-layer inflow performance.

Fig. 6.13—Crossflow between two reservoir units in a closed-in well.

6.11 Questions

6.1 Under which conditions does a linear PI provide a good model for well inflow
behavior?
6.2 What is the difference between steady-state and pseudosteady-state
 reservoir flow?
6.3 What are the units of the skin factor S in a well inflow equation?
6.4 What is the most common reason for crossflow between two zones after
closing in a well that has been completed in multiple reservoir zones?
6.5 Consider single-phase oil flow under semisteady-state conditions toward a
well in the center of a circular reservoir with parameters given in Table 6.4.
What is the PI expressed in SI units and in field units?
6.6 Analyze the sensitivity of the solution to Question 6.5 to changes in k, rw, and
re by recomputing the PI for different values of these parameters (e.g., 0.5,
0.75, 1.5, and 2.0 times their original values) and plotting the results in a
spider plot—i.e., a plot with the relative change in a parameter value on the
horizontal axis and the corresponding relative change in the PI on the vertical
axis. What do you conclude?
6.7 Starting from the expression for semisteady-state flow toward a vertical well
(Eq. 6.35), derive exact and approximate expressions for the average
reservoir pressure and for the corresponding drawdown (i.e., Eq. 6.37).

Property Value Units

Bo 1.12 –
H 35 ft
ko 84 mD
pR,av 5,600 psi
re 2,100 ft
rw 0.33 ft
Rsb 145 scf/STB
S 0 –
TR 230 °F
γ API 29 °API
γg 0.78 –
μo 2.54 cp
Table 6.4—Input data for well inflow performance for Question 6.5.

6.8 A frequently used equation for the inflow performance of a horizontal well was
derived by Joshi (1988) and is derived in detail in Chapter 7. He modeled
horizontal flow toward a horizontal well of length L oriented along the major
axis of an elliptical drainage domain with major axis length ae and constant
height h. He chose the ends of the well to coincide with the foci of the ellipse,
and he assumed a vertical constant pressure boundary at the elliptical edge of
the domain, as well as two horizontal no-flow boundaries at the top and the
bottom with the well midway between. In addition he approximated the flow
 resistance in the vertical plane to account for the radial convergence of the
flowlines close to the well. The combined effects result in a PI that can, for an
oil well, be expressed as

Compute the production rate for a horizontal gas well with fluid properties
given in Table 6.2, well length L = 500 m, reservoir dimensions h = 20 m and
ae = 700 m, and FBHP pwf = 20 MPa. Set the Forchheimer term equal to zero,
and use the approximate analytical solution technique with quadratic pressures
described in Section 6.8.2. For the average gas properties use μg,av = 23 × 106
Pa·s and Bg,av = 4.4 × 103. Compute the average reservoir pressure iteratively
with Eq. 6.63; start with a first guess—e.g., pR,av = (pwf + pR)/2—and perform a
single iteration.
6.9 A vertical oil well in a reservoir above bubblepoint pressure has a productivity
index J = 4 × 109 m3/s·Pa (15.0 STB/D-psi). The initial average reservoir
pressure is 30 MPa (4351 psi) and the bubblepoint pressure 28 MPa (4,061
psi). Compute the flow rate when the FBHP is lowered to 25 MPa (3,626 psi).
Use the Vogel equation with α = 0.2.
6.10 A vertical gas well’s production is commingled from two reservoir layers with
different permeabilities. Because of differential depletion, the reservoir
pressure in the upper layer has dropped to 9 MPa, whereas the reservoir
pressure in the lower layer is still 10 MPa. Fig. 6.14 depicts the IPRs
corresponding to the two layers, which are close enough vertically to neglect
the effects of gravity. What is the CBHP of the well? Will there be crossflow
between the layers? If yes, how much? How much does the well produce at a
FBHP of 6 MPa?

Questions 6.11 through 6.14 require the use of MATLAB.

6.11 Recompute the answer to Question 6.5 with the aid of MATLAB function res oil
simp.m.
6.12 Inspect and run the MATLAB file example_IPR.m. Investigate the effect of a skin
value S = 5. What would be a more effective measure to counteract this
amount of wellbore damage for future wells in the same area: (a) performing
an acid job to return the skin value to zero again or (b) drilling a well with twice
the diameter?
6.13 Recompute the answer to Question 6.8 with the aid of pseudopressures. Hint:
Inspect MATLAB function example_gas_pseu_pres.m. Make a copy and modify it
to compute m(pR), m(pwf), μg,ref, and Bg,ref for use in a slightly adapted version
of Eq. 6.77. What is causing the small difference with respect to the answer to
 Question 6.8? Which of the answers is more accurate?

Fig. 6.14—IPRs for a vertical gas well with commingled production.

6.14 Have another look at Question 6.10. The corresponding parameters are given
in Table 6.5. Use the standard MATLAB routine fzero to iteratively compute the
closed-in bottomhole pressure (CBHP). Iterate on qg,sc,1 (= −qg,sc,2) until pwf,1 –
pwf,2 = 0 (see Fig. 6.15). Consult Appendix D for information on the use of
fzero. You may also want to review Question 3.12 and its solution as an
example of how to implement an iterative procedure.

Property Value Units

h 20 m
k1 1.0 × 10−14 m2
k2 0.5 × 10−14 m2
9.0 × 106 Pa
10.0 × 106 Pa
re 400 m
rw 0.20 m
 TR,1 = TR,2 80 °C
β 4.0 × 1010 1/m
ρg,sc 0.93 kg/m 3
Table 6.5—Input data for CBHP calculation for Question 6.14.

Fig. 6.15—Schematic representation of two crossflowing layers in a closed-in well.

6.12 MATLAB Assignment: Commingled Production With a Smart Well

6.12.1 Objectives

• Become familiar with the concept of smart wells and their use to control
commingled production
• Become familiar with the MATLAB function fsolve to perform multivariable root
finding (as an extension to single-variable root finding with function fzero)

6.12.2 Assignment. Introduction. In many oil provinces, hydrocarbons are present

in stacked reservoirs—i.e., a series of (near-)horizontal permeable layers separated
by impermeable boundaries, typically in the form of sand/shale sequences. Most
governments have regulations against commingled production from such reservoirs
because of the risk of crossflow during well shut-in. Crossflow may cause reservoir
impairment and ineffective use of reservoir pressure and therefore a lower ultimate
recovery. However, since the beginning of this century the advent of smart wells (i.e.,
wells equipped with sensors and valves to control inflow from each layer) has opened
up possibilities for controlled commingled production and even for optimizing the
production by balancing the effects of oil, gas, and water production from each layer
in a controlled fashion (see, e.g., Naus et al. 2006).
Fig. 6.16 schematically depicts a vertical well producing commingled from three
horizontal reservoir layers (also known as zones) with the aid of inflow control valves
(ICVs). ICVs can be operated hydraulically or electrically and can be continuously
variable or have a number of discrete settings (valve openings) with the extreme of
being either on or off. Here we consider continuously variable ICVs. Usually smart
wells are equipped with pressure sensors inside the tubing and in the annular areas
outside each ICV such that it is possible to estimate the (mixture) flow rates from
each zone by combining measurements and models for the pressure drop over each
of the valves. The combined use of measurements and models of pressure drop to
estimate flow rates is known as soft sensing. Alternatively, downhole flowmeters are
occasionally employed, and it is also possible to estimate zonal inflow through
selective production testing of (combinations of) individual zones.

Fig. 6.16—Schematic representation of a smart well producing from three zones (dotted line indicates flow
from the near-well reservoir through the perforations and the ICV into the tubing; lateral dimension is
exaggerated).

Requirements. This assignment is focused on modeling. Compute the flow rates

in the three zones for given ICV settings, reservoir pressures, and the FBHP at the
middle of Zone A. The system schematic is given in Fig. 6.17. The largest
contributions to the system pressure drop originate from flow through the reservoirs
and the ICVs. Pressure drop over the perforations is taken into account with a small
positive skin factor. Assume that the zones produce single-phase dry oil from circular
drainage areas. All relevant data are given in Table 6.6.
Note. In this assignment you will have to use the black-box MATLAB function fsolve,
which forms part of the MATLAB Optimization Toolbox. (For more information, see
Section D-1.5 in Appendix D.) This is not necessarily the most efficient way to solve
the problem, but more-advanced multivariable optimization approaches and the
underlying theory are outside the scope of this book.
Fig. 6.17—Nodal analysis configuration.

Variable Symbol Unit Value

Reservoir data
Height of zone A hA m 13
Height of zone B hB m 15
Height of zone C hC m 7
Permeability of zone A kA m2 8 × 10−13
Permeability of zone B kB m2 4 × 10−13
Permeability of zone C kC m2 3 × 10−13
Drainage radius re m 300
Skin S – 1
Reservoir temperature TR ºC 120
Flow regime Steady-state
Fluid data
Solution GOR at bubblepoint Rsb m 3/m 3 5
Oil viscosity Beggs and Robinson (1975), Vasquez and Beggs (1980)
Gas density at standard conditions ρg,sc kg/m 3 0.85
Oil density at standard conditions ρo,sc kg/m 3 885
Water density at standard conditions ρw,sc kg/m 3 1200
Equations of state Standing (1952), black oil
Water cut fw,sc – 0.0
Well data
Tubing length between compl. top and zone A Ltop m 15
Tubing length between centers zones A and B LAB m 16
Tubing length between centers zones B and C LBC m 12.5
Wellbore inclination α rad 0
Tubing inside diameter d m 0.076
Tubing roughness e m 30 × 10−6
Well diameter (for reservoir flow) rw m 0.20
ICV data
Discharge coefficient Cd 0.9
Maximum ICV opening dICV,max m 0.05
Pressure data
FBHP at top of completion ρwf,top MPa 28.0
Reservoir pressure at middle of zone A ρR,A MPa 44.0
Reservoir pressure at middle of zone B ρR,B MPa 44.5
Reservoir pressure at middle of zone C ρR,C MPa 39.0
Table 6.6—Data for commingled production with a smart well.

Tasks and Questions

• Check the data in Table 6.6. What is the thickness of the impermeable layers
between the zones?
• Inspect the reservoir pressures. Is the system in hydrostatic equilibrium? If not,
in which zones do you expect crossflow to occur when the well is shut in?
• Use MATLAB functions res_oil_simp.m, choke_oil.m, and pipe.m to model the
system, and then perform the following steps:

1. Start with a fully open ICV for Zone A and closed ICVs for the other two, and
iterate on qo,sc,A until the difference between pwf,top as computed and as
specified becomes zero. (Because this is a single-variable root-finding
 problem, it may be done with either fzero or fsolve.) Repeat this step for
the other two zones, every time just producing from one zone. The following
MATLAB code gives an example of how to compute the flow rates for zone B
with fsolve:
q_o_sc_B = fsolve(@(q_o_sc_B)…
p_wf_top - cascade_ICV(alpha,av,C_d,d,…
d_ICV_B,dir_p,e,fluid,h_B,oil,k_B,p_R_B,…
[0,q_o_sc_B,0],r_e,r_w,rho_sc,semi,
L_top+L_AB,0,S,T_R),q_o_sc_start);
Here, the function cascade_ICV is defined as an anonymous function with a
single argument (in this case q_o_sc_B). It generically computes the pressure
drop over a cascade of elements (reservoir, ICV, and tubing) for flow from a
single zone (in this case zone B). It is specified as
function p_out = cascade_ICV(alpha,av,C_d,d,…
d_ICV,dir_p,e,fluid,h,oil,k,p_R,q_sc,r_e,…
r_w,rho_sc,semi,s_in,s_out,S,T_R)
% Computes the pressure drop over a cascade of
% elements consisting of the reservoir, an ICV
% and a section of tubing.
p_ann = res_oil_simp(av,h,k,oil,p_R,q_sc,r_e,…
r_w,rho_sc,semi,S,T_R); % annulus pr., Pa
p_wf_int = choke_oil(C_d,d_ICV,dir_p,oil,p_ann,…
q_sc,rho_sc,T_R); % intermediate pr., Pa
p out = pipe(alpha,d,e,fluid,oil,p wf int,q sc,…
rho_sc,s_in,s_out,T_R,T_R); % output pr., Pa
2. Next perform multivariable root finding with the aid of fsolve and iteratively
compute the required flow rates qo,ssc,i, i = A, B, C to produce the well for
given values of the reservoir pressures pR,i and the FBHP at the top of the
completion, pfw,top. The corresponding MATLAB code becomes
q_o_sc_vect_out = fsolve(@(q_o_sc_vect)…
system_ICV(alpha,av,C_d,d,d_ICV_A,d_ICV_B,…
d_ICV_C,dir_p,e,fluid,h_A,h_B,h_C,L_AB,L_BC,…
L_top,oil,k_A,k_B,k_C,p_R_A,p_R_B,p_R_C,…
p_wf_top,q_o_sc_vect,r_e,r_w,rho_sc,semi,…
S,T_R),q_o_sc_start);
Here, the variables q_o_sc_vect, q_o_sc_vect_out, and q_o_sc_start are
vectors of three elements (i.e., the three zonal flow rates) each. The
anonymous function system_ICV is partly specified as follows (code the
missing parts indicated with “Etc., etc.” yourself):
function p_out = system_ICV(alpha,av,C_d,d,
d_ICV_A,d_ICV_B,d_ICV_C,dir_p,e,fluid,
h_A,h_B,h_C,L_AB,L_BC,L_top,oil,k_A,k_B,
k_C,p_R_A,p_R_B,p_R_C,p_wf_top,
q_o_sc_vec,r_e,r_w,rho_sc,semi,S,T_R)
% Computes various pressure drops in a system of
% elements consisting of three reservoirs,
% three ICVs, and three sections of tubing.
% Assemble flow rate vectors:
q_sc_A = [0,q_o_sc_vec(1),0];
q_sc_B = [0,q_o_sc_vec(2),0];
 q_sc_C = [0,q_o_sc_vec(3),0];
q_sc_BC = q_sc_B + q_sc_C;
q_sc_ABC = q_sc_A + q_sc_B + q_sc_C;
% Compute horizontal pressure drop over zone C
% to tubing plus vertical pressure drop over
% bottom of completion:
p_ann_C = res_oil_simp(av,h_C,k_C,oil,p_R_C,…
q_sc_C,r_e,r_w,rho_sc,semi,S,T_R);
p_wf_C_C = choke_oil(C_d,d_ICV_C,dir_p,oil,…
p_ann_C,q_sc_C,rho_sc,T_R);
p_wf_B_C = pipe(alpha,d,e,fluid,oil,p_wf_C_C,…
q_sc_C,rho_sc,L_top+L_AB+L_BC,L_top+L_AB,…
T_R,T_R);
% Compute horizontal pressure drop over zone B
% to tubing:
p_ann_B = res_oil_simp(av,h_B,k_B,oil,p_R_B,…
q_sc_B,r_e,r_w,rho_sc,semi,S,T_R);
p_wf_B_B = choke_oil(C_d,d_ICV_B,dir_p,oil,…
p_ann_B,q_sc_B,rho_sc,T_R);
% Compute vertical pressure drop over middle
% of completion:
% Etc., etc.
% Define vector of pressure differences that
% should all be zero:
p_out = [p_wf_top - p_wf_top_ABC;…
p_wf_A_A - p_wf_A_BC;…
p_wf_B_B - p_wf_B_C];
Use the results obtained under Step 1 as starting values. How large (or small)
is the difference with the result obtained under Step 1? What happens if you
reduce the tubing size? Explain your answers.

• The sand in the reservoir is loosely consolidated, and therefore the zonal oil
production needs to be restricted to avoid sand production. Typical sand
production constraints specify either a maximum rate or a maximum drawdown.
Constrained optimization is not an option in fsolve. Therefore manually iterate
on the three valve settings dICV,i to achieve a maximum production without
exceeding a zonal rate of 1,500 STB/D. What are the resulting valve openings?

6.12.3 Deliverables

• MATLAB program listing

• Pressures and flow rates in field units
• Answers to all questions
Chapter 7

Inflow Performance: Further Topics

7.1 What Is Covered in This Chapter?

In this chapter we extend the basic treatment of inflow performance of Chapter 6 to
more complex well configurations. Starting with analytical methods to describe single-
phase flow toward groups of vertical wells and wells near reservoir boundaries, we
progress to discuss various methods to approximate the inflow performance of
horizontal wells and multilateral wells. Next we discuss a semianalytical technique to
approximate multiphase (gas/oil/water) near-well reservoir flow. Another aspect of
complex-well flow is the effect of pressure drop inside the reservoir section of the
wellbore, which can become especially significant for high-rate horizontal wells in low-
permeability reservoirs. We treat a simplified approach to this coupled
reservoir/wellbore-flow problem in the assignment at the end of the chapter. In
general, the detailed prediction of multiphase well inflow performance in complex well
configurations requires numerical reservoir simulation methods beyond the scope of
this book; moreover, we typically need field measurements (production test data) to
accurately account for effects that cannot be modeled a priori because of lack of
information.

7.2 Two-Dimensional Reservoir Flow

7.2.1 The Laplace Equation. Until now we have mainly discussed the inflow
performance of vertical wells. One of the underlying assumptions was that the height
of a reservoir is usually small compared to its lateral dimensions such that the flow
can be approximated to behave two-dimensionally. Moreover, thanks to radial
symmetry, the governing differential equations for a single vertical well could be
expressed in polar coordinates, resulting in a 1D problem. (Note that only the radial
coordinate r plays a role in the differential equations. The angular coordinate ψ is just
a constant parameter.) For horizontal wells, we may often maintain the assumption of
relatively small reservoir height, but we can no longer make use of the symmetry
argument; therefore, we need to express the governing equations in two spatial
dimensions. Moreover, in a configuration of multiple vertical wells, the flow field also
becomes 2D.
For the control volume depicted in Fig. 7.1 the mass balance equation for single-
phase flow can be expressed as
Fig. 7.1—Two-dimensional control volume with dimensions dx and dy in the x and y coordinate directions.

(Note the resemblance of Eq. 7.1 to the 1D and radial cases represented in Eqs. 3.3
and 6.4.)
Darcy’s law (i.e., the degenerated momentum balance) can be expressed in two
dimensions as

where we have assumed that the reservoir is homogeneous with scalar permeability
k. (See Section 7.2.5 for discussion of a more general expression for nonscalar
permeabilities). Substituting Eqs. 7.2 and 7.3 and simplifying the results using the
definition of single-phase oil compressibility (Eq. 2.13) lead to

which is known as the diffusivity equation. If we restrict the analysis to steady-state

flow, the right side of Eq. 7.1 vanishes, resulting in

known as the Laplace equation. (Note that the Laplace equation is often written as 2
p = 0, where is the gradient operator.) It is a time-independent, linear,
homogeneous, second-order, partial-differential equation in two spatial dimensions,
and it therefore requires four boundary conditions: two for each coordinate direction.
Exactly the same equation describes other physical phenomena such as the steady-
state flow of heat or electric current, in which case p should be replaced by the
temperature or the electric potential, respectively. Many single-phase reservoir flow
problems can therefore be solved through analogy by using published results from
these domains (see, e.g., Morse and Feshbach 1953 or Carslaw and Jaeger 1959).

7.2.2 Polar Coordinates. It is often possible to find a transformation from

rectangular coordinates (x, y) to another coordinate system that better suits the
geometry of the problem at hand. In Section 6.4 we already used polar coordinates r
and ψ, which can be related to rectangular coordinates x and y according to

where 0 ≤ r < ∞, 0 ≤ ψ < 2π. The inverse transformation is given by

Lines of constant r form circles, and lines of constant ψ are radial lines (see Fig. 7.2).
The Laplace equation in polar coordinates can be determined as
Fig. 7.2—Polar coordinate system with 0 ≤ r ≤ 0.8, 0 ≤ ψ < 2 π.

This is now the equation for single-phase incompressible flow in polar coordinates.
[To verify this result, start from 2p = ∂2 p/∂x2 + ∂2 p/∂y2. Use the chain rule to rewrite
the first right-hand-side term as ∂2 p/∂x2 = ∂/∂x (∂p/∂x) = ∂/∂x(∂p/∂r ∂r/∂x + ∂p/∂ψ ∂ψ/
∂x). Repeat the chain rule for the second derivative, use the same approach for the
term ∂2 p/∂y2, compute the derivatives from Eqs. 7.6 and 7.7, and simplify the
results.]
In Section 6.5 we expressed the same flow equation as two first-order differential
equations and, with the aid of boundary conditions at r = rw and r = re, obtained the
analytical solution for radial flow given in Eq. 6.20. It can be simply verified by
substitution that this expression is also a solution of Eq. 7.10 if we express the
boundary conditions as
 Note that in Section 6.5 we never considered the tangential coordinate ψ because
we tacitly assumed that radial symmetry implies that ∂p/∂ψ = 0 everywhere. In that
case Eq. 7.10 reduces to

which can be twice integrated to give the radial pressure distribution

where C1 and C2 are integration constants that have to be determined from the radial
boundary conditions (Eqs. 7.11 and 7.12). Using the first boundary condition leads to

a result identical to Eq. 6.19, which is sometimes referred to as the general radial
flow solution. Using the second boundary condition then leads to

which is identical to Eq. 6.20. Alternatively, we may chose the radial boundary
conditions as

in which case the solution becomes

The total flow rate from the reservoir into the well can now be computed with the aid
of Darcy’s law according to
which is identical to the result obtained in Eq. 6.21.
The lines of constant r in Fig. 7.2 (i.e., the circles) can be interpreted as contours
of constant pressure, called equipotential lines, and the lines of constant ψ (i.e., the
radii) as streamlines. Note that in Fig. 7.2 the coordinates r and ψ have been
selected equidistantly but that the corresponding pressure contours will not be
equidistant in magnitude because the pressure increases logarithmically with
increasing r. In contrast, if we plot pressure contours of equidistant magnitude, the
radial coordinates will be logarithmically distributed (see the example in Fig. 7.3).

7.2.3 Linear Superposition. Because the Laplace equation is a linear, homogeneous

differential equation, any linear combination of solutions will also be a solution, a
property known as the principle of linear superposition. Note that this can easily be
verified by substitution. For a proof, see any introductory text to differential equations
(e.g., Boyce and DiPrima 2012). In general, we can therefore generate new solutions
of the form

where the coefficients c0, c1, …, cN may be chosen such that specific (new) boundary
conditions are met. Linear superposition is a valuable property with which to generate
solutions for the flow field originating from a number of isolated wells. For example, if
we combine two radial solutions, as given in Eq. 6.20, for a doublet (i.e., a pair of
wells consisting of an injector and a producer) we obtain
Fig. 7.3—Equipotential lines and streamlines for radial flow with parameters given in Table 6.1 (values for
“other figures”) and a flow rate of −1.09 × 10−3 m3/s (−593 STB/D) (numbers in the plot indicate pressures
in MPa).

Here D is the interwell distance which has been chosen along the x-axis. We have
used n = 2, c0 = pR, c1 = 1, and c2 = −1; moreover, each of the two solutions was
shifted over a distance ± D/2 before addition. Fig. 7.4 displays the corresponding
equipotential lines and streamlines for parameters given in Table 6.1, and a value of D
equal to 500 m. Note that the outer boundary, re, has dropped out of the equation.
Thanks to the choice of c0, the far-field pressure is equal to pR. Furthermore, because
of symmetry the average reservoir pressure is equal to the far-field reservoir
pressure while we have only steady-state flow because the equal injection and
production rates result in voidage replacement and, thus, a constant reservoir
pressure.
 For (x, y) = (−D/2 + rw, 0), Eq. 7.24 gives the pressure in the producer. (Note that
the superposition of pressures results in pressure values that are not exactly identical
for all points at the wellbore radius. However, this small effect is completely negligible
for our purpose.) The steady-state PI for a producer in a doublet therefore becomes

Fig. 7.4—Equipotential lines and streamlines around a doublet with parameters given in Table 6.1 and an
inter-well distance D = 500 m and flow rates qo,sc = −1.09 × 10−3 m3/s (−593 STB/D) (numbers in the plot
indicate pressures in MPa).

where the approximation holds for rw << D. As usual, a skin term may be added if
required, just as in the standard productivity-index (PI) equation (Eq. 6.45).

7.2.4 Image Wells. Another example of linear superposition of the solution equation
(Eq. 6.20) is the case of four producers in a rectangular configuration with interwell
distances Dx and Dy along the x- and y-axis, respectively. Together, they result in a
pressure distribution:
where we now used n = 4, c0 = −3pR, and c1 = c2 = c3= c4 = 1, and where the four
solutions were shifted in different combinations in the x- and y-direction over
distances ± Dx / 2 and ± Dy / 2 before addition. Note that for this case, with only
producers, the external radius re still appears in the equation. Fig. 7.5 displays one
quadrant of the resulting double-symmetric configuration of equipotential lines and
streamlines. The image can be interpreted as representing flow toward a well near
the corner of two no-flow boundaries (along the x- and y-axes) at right angles.

Fig. 7.5—Equipotential and streamlines around a well near a no-flow corner with parameters given in Table
6.1 and a flow rate qo,sc = −1.09 × 10−3 m3/s (−593 STB/D) (numbers in the plot indicate pressures in MPa).

This example illustrates that we can represent no-flow boundaries with the aid of
image wells. This property offers many opportunities to create complex flow
configurations, especially because we may select any combination of image wells,
even if the pressure field outside the domain of interest becomes physically
unrealistic. For several other examples of the use of image wells to model subsurface
flow, we refer to Muskat (1937), Verruijt (1970), and Bear (1972). These authors also
extensively discuss the use of conformal mapping, which is an even more powerful
method for obtaining solutions of the Laplace equation in a wide variety of spatial
domains. (A conformal map is a transformation that preserves angles. It can be seen
in Figs. 7.2 to 7.5 that the angles between the equipotential lines and the streamlines
are always 90°, just like those between the x- and y-axis in rectangular coordinates.)
The method of conformal mapping makes use of transformations in the complex
domain (i.e., using imaginary numbers), which is outside the scope of our book.
Just as in the previous example we can directly compute the PI for this flow field.
For , Eq. 7.26 gives the pressure in the producer in the
top-right quadrant, and the steady-state PI for a producer near a no-flow corner
therefore becomes

where the approximation holds for rw << Dx << re and rw << Dy << re. It can be verified
that the complete expression has the correct limiting behavior: When Dx and Dy
approach zero (i.e., when the well approaches the corner), the PI approaches just
one-quarter of the single-well PI given in Eq. 6.22. The approximation breaks down
for (very) small values of Dx and Dy but gives completely acceptable results for
practical purposes, especially in light of the strongly approximate nature of the entire
PI concept.

7.2.5 Anisotropic Permeability. Until now we considered homogeneous reservoirs

(i.e., having properties that do not vary spatially, with an isotropic permeability, i.e., a
permeability that has equal magnitude for flow in any direction). All realistic reservoirs
are heterogeneous (i.e., nonhomogeneous). An accurate representation of the
governing equations for flow through a heterogeneous reservoir typically requires
numerical techniques such as finite volume or finite element methods, which are
outside the scope of this book. However, it is often useful to describe reservoirs as
homogeneous but anisotropic—i.e., with different values of permeability for flow in
different directions. In case of 2D flow this requires that we change the definition of
Darcy’s law as given in Eqs. 7.2 and 7.3 to

where kxx, kxy, kyx, and kyy are the elements of the symmetric 2D permeability tensor,

in which kxy = kyx. The presence of off-diagonal elements in a permeability tensor

implies that a pressure gradient in the x-direction causes flow in the x- and y-
direction. It can be shown with physical arguments that a pressure gradient in the y-
direction will then also cause flow in the x- and y-directions and that, moreover, the
off-diagonal terms kxy and kyx must have the same magnitude. As a result, the
permeability tensor is symmetric. Substitution of Eqs. 7.28 and 7.29 in the mass
balance (Eq. 7.1) and then simplifying the results using the definition of single-phase
oil compressibility (Eq. 2.13) will lead to an anisotropic form of the diffusivity equation
(refer to Eq. 7.4):

which can be more compactly written in vector notation as

or

where is the gradient operator. This expression also holds for 3D flow, in which
case K has nine elements (of which, because of symmetry, only six are independent).
Often the permeability tensor is diagonal with elements of different magnitude:
This implies that the flow is aligned with the direction of the pressure gradient, which
typically occurs when the coordinate axes are chosen such that they are in alignment
with the geological layering of the reservoir.
For the case of a homogeneous anisotropic reservoir it is usually possible to
realign the coordinate system through rotation over an angle such that the
permeability tensor becomes diagonal. (We will not discuss this any further but for
an-in depth description refer to Bear 1972.) Moreover, it is usually possible to obtain
an equivalent isotropic system through scaling the coordinate axes (Vreedenburgh
1936; Muskat 1937; Bear 1972). Several choices for scaling are available, and here
we choose

where

Substitution of Eqs. 7.35 through 7.37 in Eq. 7.31 with kxx = kx, kxy = kyx = 0, and kyy =
ky results in

from which follows

which demonstrates that as defined in Eq. 7.37 is just the equivalent isotropic
permeability for the diffusivity equation in scaled coordinates and . By restricting
the analysis to steady-state flow, Eq. 7.39 reduces to the isotropic Laplace equation
in scaled coordinates:

When applying this scaling method to an inflow equation, it is necessary that all
coefficients and boundary conditions that are a function of the reservoir dimensions
be scaled accordingly. For circular boundaries, such as a well or the boundary of a
circular drainage area, this implies that the circle deforms to an ellipse with (half)
axes equal to

and using a simple arithmetic average, we obtain the scaled equivalent radius

As an example of the use of coordinate scaling to represent reservoir anisotropy,

consider Eq. 7.25 for the PI in a doublet with an interwell distance D along the x-axis.
If we want to apply this equation to a reservoir with anisotropic permeabilities kx and
ky we should use

where

is known as the anisotropy index.

7.3 Horizontal Wells, Part 1

7.3.1 Box-Shaped Reservoir. Horizontal wells have become ubiquitous since the
1980s, although their first use dates from much longer back (for overviews see, e.g.,
Joshi 1991 and Butler 1994). The first expressions to describe horizontal well inflow
considered a box-shaped reservoir geometry as schematically depicted in Fig. 7.6.
The reservoir dimensions are specified as length L, height h, and width w. The well is
fully penetrating— i.e., it runs from the front to the back of the reservoir—and is
operated at a constant flow rate qo,sc. The x−y planes at z = ±h/2 (i.e., the top and
bottom) and the y−z planes at x = 0 and x = L (i.e., the front and back) are no-flow
boundaries, while the x−z planes at y = ±w/2 (i.e., the sides) are constant pressure
boundaries with pressure p = pR. If pressure drop along the wellbore is disregarded,
the flow will be perpendicular toward the well; i.e., the flow can be described by
considering a slab in the y-z plane with a unit thickness along the x-axis (see Fig.
7.7). If we consider single-phase incompressible steady-state flow and disregard
gravity, it seems reasonable to expect that the flow will be primarily horizontal with
some streamline convergence close to the well.

7.3.2 Horizontal Flow. As a precursor to this problem, we first consider a simpler

case in which the horizontal well is replaced by a vertical fracture over the entire
height and length of the reservoir (see Figs. 7.8 and 7.9). In this configuration the oil
flows purely horizontally from the vertical boundaries to the vertical fracture. For such
a linear flow pattern the steady-state mass balance and Darcy’s law can be written
as

Fig. 7.6—Schematic representation of a box-shaped reservoir with a central horizontal well (with vertical
dimension strongly exaggerated).
Fig. 7.7—Vertical two-dimensional plane with a central horizontal well, no-flow boundaries at top and
bottom, and constant pressure boundaries at the sides (dotted lines represent schematic streamlines).

Fig. 7.8—Schematic representation of a box-shaped reservoir with a central vertical fracture.

Fig. 7.9—Vertical two-dimensional plane with a central vertical fracture, no-flow boundaries at top and
bottom, and constant pressure boundaries at the sides (dotted lines represent streamlines).

and the boundary conditions as

Note that qo,sc represents the flow rate originating from reservoir inflow from both
sides. Differentiation of Eq. 7.47 and substitution of Eq. 7.46 results in the 1D
Laplace equation

which can be integrated twice to obtain

where the constants C1 and C2 can obtained with the aid of boundary conditions (Eqs.
7.48 and 7.49), resulting in

The fracture pressure for a given flow rate follows by substituting 0 for y and pfrac for
p. Reorganizing the results, we obtain the PI for steady-state linear flow toward a
vertical fracture:

7.3.3 Near-Well Radial Flow. Returning to the horizontal well problem, we may
expect a similar PI but with a slightly lower value because of increased resistance
due to radial flow and the associated streamline convergence near the well. The
solution for this 2D flow problem in the y−z plane can be obtained using linear
superposition with image wells. The effect of top and bottom no-flow boundaries can
be represented by image wells, but unfortunately each of these disturbs the pressure
field of the other. These disturbances can, in turn, be compensated by new image
wells, leading to an infinite array of image wells stretching in the upward and
downward direction (see Fig. 7.10).
Because the infinite row of wells stretches beyond the external reservoir boundary
re, which is present in the conventional radial pressure solution (Eq. 6.20), we start
from the general radial flow solution (Eq. 6.19) and will introduce a relevant external
reference pressure later. The corresponding expression for the pressure distribution
has first been given by Muskat (1937). Using the notation introduced in Eq. 7.23 we
can write
where we have left out the constant C under the summation because the constant c0,
which is as yet undetermined, will be used to fix the pressure. Eq. 7.54 can, after
considerable algebraic manipulation, be rewritten as

Fig. 7.10—Vertical two-dimensional plane with an infinite array of image wells to represent no-flow
boundaries at top and bottom.

where we have neglected several constant terms because their effect will be
annihilated once we choose the constant c0 to fix the reservoir pressure. (The full
derivation of Eq. 7.55 is given in question 7.6 in Section 7.8 and its answer in
Appendix G). If we choose to fix the pressure at the external boundaries at the
vertical level of the well, we have
which leads to a constant

and therefore to a solution

As an example, consider a horizontal well in a box-shaped reservoir with

parameters given in Table 7.1. Figs. 7.11 and 7.12 depict the equipotential lines and
streamlines in the near-well area and in a larger part of the reservoir, respectively.
The PI for the well follows by specifying the wellbore pressure as

Property Symbol Value Units

Reservoir height h 25 m
Reservoir length L 5000 m
Reservoir width w 500 m
Permeability k 2.0 × 10−12 m2
Well diameter d 0.20 m
Oil viscosity μ 20 × 10−3 Pa · s
Oil formation volume factor Bo 1 –
Reservoir pressure pR 33.5 MPa
Oil flow rate qo,sc −2.28 × 10−4 m 3/s
Skin Sv 0 –
Table 7.1—Parameter values for horizontal well in box-shaped reservoir.
Fig. 7.11—Equipotential and streamlines around a horizontal well with parameters given in Table 7.1. Note
that the figure covers the entire reservoir height (25 m) but only a small part (33.3 m) of the reservoir width
(500 m) (numbers in the plot indicate pressures in MPa).

Fig. 7.12—Equipotential and streamlines around a horizontal well with parameters given in Table 7.1. Note
that the figure covers the entire reservoir height of 25 m but only 100 m of the 500 m of reservoir width.
which leads to

For large values of w/h we can write

and for small values of wR/hR

and therefore we can derive that

Eq. 7.63 can be interpreted as the harmonic average of two productivity indices:

where Jfrac is just the PI for linear flow toward a vertical fracture as given in Eq. 7.53,
and Jconv accounts for the additional pressure drop because of streamline
convergence near the well:

Eq. 7.65 can be interpreted as the PI for radial flow toward a well of length L in a
circular reservoir with an equivalent external radius req = h/(2π) (refer to Eq. 6.22).
Fig. 7.12 illustrates that the flow in the majority of the reservoir is indeed linear,
whereas streamline convergence occurs only close to the well, corresponding to a
more radial flow pattern.
These results were published before by Giger (1987) and Butler (1994), who also
derived similar expressions for the case of an eccentric horizontal well—i.e., for a
situation where the well is positioned above or below the center of the reservoir. Both
Giger (1987) and Butler (1994) refer to an earlier Russian publication, notably Borisov
(1964), which has apparently been translated into English but which, unfortunately,
could not be located during the preparation of this book. Moreover, according to
Butler, the paper by Borisov (1964) refers to other Russian papers, namely, Borisov
(1951), Shchurov (1952), and Merkulov (1958), none of which, however, could be
found.

7.3.4 Skin and Geometric Skin. Just as in vertical wells, an additional pressure drop
may occur because of formation damage. Following the approach of Section 6.6, this
effect can be described with a skin factor, and in analogy to Eq. 6.45 we can then
rewrite Eq. 7.63 as

Here, we have added a subscript h to the skin factor S to indicate that it is defined for
a horizontal well. Alternatively, we could use the same definition as used for a vertical
well (see Eq. 6.42). This would make sense for a situation in which the skin value is
first determined experimentally in a vertical well—e.g., through pressure transient
analysis in an exploration or appraisal well—and subsequently used for the design of
a horizontal production well. In that case we can rewrite Eq. 7.63 as

where

The additional pressure drop corresponding to these two definitions follows as

In addition, the effect of near-well streamline convergence can be interpreted as a

geometric skin factor defined as

with a corresponding additional pressure drop

Using this approach, the PI for steady-state flow toward a horizontal well in a
homogeneous box-shaped reservoir becomes

or, alternatively,

7.3.5 Average Reservoir Pressure. In Eqs. 7.72 and 7.73 we tacitly used a
reference pressure pref equal to the reservoir pressure pR. Just as for vertical wells, it
may be more convenient to define the PI with respect to a reference pressure equal
to the average reservoir pressure pR,av. To do so we can start from Eq. 7.52 for the
linear part of the pressure drop:

An expression for the average reservoir pressure then follows as

with which we can derive

7.3.6 Semisteady-State Flow. Alternatively, we may consider semisteady-state flow.

Starting the derivation again for horizontal flow toward a vertical fracture, the
appropriate boundary conditions are now

Following the same reasoning as in Section 6.5.4, and using boundary conditions
(Eqs. 7.77 and 7.78), we can derive the differential equation for semisteady-state
linear flow:

Integration of the equation results in

Just as in Section 6.5.4, we can use the steady-state boundary condition (Eq. 7.56)
(but now for a slowly decreasing pressure pR) to solve for the integration constant C,
and back substitution of the result in Eq. 7.80 then gives an expression for p as a
function of y under semisteady-state flow conditions:

The PI for semisteady-state linear flow from two sides toward a vertical fracture
follows as

which is twice as high as the results for steady-state flow (see Eq. 7.53). The PI for
semisteady state flow toward a horizontal well can now be obtained by adding skin
factors to represent the effects of near-well streamline convergence and formation
damage:

where the reference pressure is the reservoir pressure.

Just as for steady-state flow, we can also express the PI with respect to the
average reservoir pressure. (For the derivation, see Question 7.8 in Section 7.8 and
its answer in Appendix G.) The corresponding expression for the PI follows as

7.3.7 Anisotropic Permeability. As discussed in Section 7.2.5, we can model

reservoir anisotropy with the aid of scaled reservoir dimensions. In horizontal wells
the difference between horizontal and vertical permeabilities is often significant
because of the nature of geological layering. For example, the presence of horizontal
layers with low permeability in an otherwise high-permeability reservoir (as in, e.g., a
sand/shale sequence) can result in especially strong anisotropy. If we restrict the
analysis to a situation in which the coordinate axes are aligned with the geological
layers and consider 2D flow in the vertical y−z plane, we can write

where kv and kh are the vertical and horizontal permeabilities. Scaling of the
coordinate axes and the relevant parameters and boundary conditions can now be
carried out using Eqs. 7.35 through 7.37 and Eqs. 7.43 and 7.45 with an appropriate
change in variables. Note that in expressions that combine 1D and 2D flow, only the
2D part requires scaling. As an example, consider the expression for the PI of a
horizontal well in an isotropic box-shaped reservoir under steady-state conditions,
which contains a linear and a radial flow term in the denominator (Eq. 7.63). The
equivalent expression for the anisotropic case becomes

where the auxiliary variables and are given by Eqs. 7.43 and 7.35 (replacing x by
h and ky by kh), and where the equivalent isotropic permeability , the anisotropy
index Ivh and the scaled geometric skin are now defined as

7.3.8 Combined Expression. Just as we did for vertical wells in Eqs. 6.43 through
6.45, we can combine the various equations for the drawdown of a horizontal well in a
box-shaped reservoir into a single expression. Including the effect of kv/kh anisotropy,
we obtain
where was already defined in Eq. 7.89, and where

with Ivh as given in Eq. 7.88, while

such that the combined expression for the PI becomes

Eqs. 7.90 through 7.93 have been programmed in MATLAB file res_oil_box.m.
Examples of how to use this file are given in example_res_pres_oil_hor.m and
example_PI_oil_hor.m.

7.4 Horizontal Wells, Part 2

7.4.1 Partially Penetrating Wells. The horizontal well in a box-shaped reservoir

considered in Section 7.3 was assumed to be fully penetrating (i.e., running over the
entire length of the reservoir with no-flow boundaries at the heel and the toe).
However, in reality horizontal wells are nearly always partially penetrating, resulting in
significant additional inflow at the heel and the toe because of the absence of flow
barriers.

7.4.2 Elliptic Coordinates. A classic approach to modeling the 2D steady-state

pressure distribution around a partially penetrating line source was published by
Schlichter (1889, p. 344–347) for groundwater flow and by Muskat (1937) for oil flow.
It makes use of the following coordinate transformation:

where 0 ≤ ξ < ∞, 0 ≤ η < 2π are known as elliptic coordinates. It can be verified by

substitution that the inverse transformation is given by
where

with the plus sign referring to Hξ and the minus sign to Hη. Lines of constant ξ form
ellipses, and lines of constant η form hyperbolas (see Fig. 7.13). The parameter L is
the distance between the foci of the ellipses. The equation for the line through the foci
of the ellipses (the x-axis in Fig. 7.13) is ξ = 0 for the part of the line between the foci
(the thick black line in Fig. 7.13), and η = 0 and η = π for the parts to the right and to
the left of the foci respectively. The lengths of the long and short axes of the ellipses
are given by

Fig. 7.13— Elliptic coordinate system with 0 ≤ ξ ≤ 0.8, 0 ≤ η ≤ 2π and L = 1. The endpoints of the thick
horizontal line coincide with the foci of the ellipses.
where a, b, and L are related according to

while the equation for the ellipse in rectangular coordinates is given by

2
With some algebra it can be shown that the expression for p in elliptic coordinates
now becomes

[To prove this result, use the chain rule to rewrite the first term between brackets in
Eq. 7.103 as ∂2 p/∂ξ2 = ∂/∂ξ(∂p/∂ξ) = ∂/∂ξ(∂p/∂x ∂x/∂ξ + ∂p/∂y ∂y/∂ξ), repeat the chain
rule for the second-order derivative, use the same approach for the term ∂2 p/∂η2,
compute the derivatives from Eqs. 7.94 and 7.95, and simplify the results using the
equality cosh2(ξ) − sinh2(ξ) = 1.]
From Eq. 7.103, the Laplace equation in elliptic coordinates follows as

7.4.3 Horizontal Flow. Just as we did in the case of a boxed-shaped reservoir, we

will first consider flow toward a vertical fracture with length L. The corresponding
boundary conditions are

The first condition prescribes the pressure in the fracture. The second condition
prescribes the reservoir pressure at the boundary of an elliptical drainage area with
long and short axes ae and be as given in Eqs. 7.99 and 7.100. The last two boundary
conditions follow from the argument that the pressure field should be symmetric
around the x-axis. (Note that we could just as well have chosen these boundary
conditions at η = π/2 and η = 3π/2 because the pressure field should also be
symmetric around the y-axis.)
It can be verified by substitution in Eq. 7.103 that the solution to this problem is
simply

The lines of constant ξ (the ellipses) in Fig. 7.14 are equipotential lines, and the lines
of constant η (the hyperbolas) are streamlines. Note that for increasing values of ξ
the ellipses start increasingly to look like circles; i.e., at a large distance from the well
the pressure field becomes nearly identical to the pressure field around a vertical
well. The total flow rate from the reservoir into the fracture can be computed with the
aid of Darcy’s law according to

where the latter two equalities have been obtained with the aid of Eqs. 7.99 and
7.100 and the equivalent expressions listed in Table 7.2. The corresponding PI
becomes

Function Equivalent Derivative Equivalent

sinh x = (ex − e−x)/2 cosh x = (ex + e−x)/2
cosh x = (ex + e−x)/2 sinh x = (ex − e−x)/2
asinh x =
acosh x =

Table 7.2—Derivatives of some hyperbolic and inverse hyperbolic functions and their equivalent
expressions.

7.4.4 Near-Well Radial Flow, Skin, and Anisotropy. To use this result for flow
toward a horizontal well, we have to add the additional effect of near-well streamline
convergence in the vertical plane. As before, this can be done conveniently with the
aid of a geometric skin factor as defined in Eq. 7.70. If we also add the regular skin
factor and account for reservoir anisotropy in the vertical plane with the aid of Eqs.
7.87, 7.88, and 7.90, the result becomes

Note that scaling is required only for the terms related to flow in the vertical plane
(because of streamline convergence). The horizontal permeability remains isotropic
and is equal to kh.
An almost identical expression of Eq. 7.112 is often referred to as the Joshi
equation. Derived by Joshi (1988) and later slightly modified by Economides et al.
(1991), the only difference is the factor 1/π inside the logarithm of the geometric skin
factor (refer to Eq. 7.89):

Eq. 7.112 has been programmed in MATLAB file res_oil_ellipse.m. Examples of how
to use this file are given in example_res_pres_oil_hor.m and example_PI_oil_hor.m.

7.4.5 Specific PI. Often, the inflow to a horizontal well is expressed in terms of the
specific flow rate (i.e., the flow rate per unit length):

For a fully penetrating well in a box-shaped reservoir, and neglecting pressure drop
inside the well, the specific flow rate is constant along the well. However, for a well in
an elliptical domain the specific flow rate varies as a function of distance along the
well. Neglecting the effects of permeability anisotropy and streamline convergence in
the ver-tical plane, we can derive

where the third equality is obtained with the aid of Eqs. 7.109 and 7.95, and Table
7.2; the fourth one with Eq. 7.94 for ξ = 0; and the fifth one with the trigoniometric
identity . The inflow is not uniform along the well, with a minimum at x
= 0. However, the maximum inflow, at x = 0 ± 0.5L, goes to infinity. To overcome this
singularity we require the flowing bottomhole pressure (FBHP) to occur at the
wellbore radius—i.e., at y = rw instead of at y = 0. (Here, we tacitly assume that rw <<
ξe such that Eq. 7.109 remains approximately valid.) Eq. 7.115 then has to be
replaced by

The specific PI (i.e., the PI per unit length) now follows as

The total flow rate from the reservoir into the well can, as before, be computed
with the aid of Darcy’s law according to

Because asin(1) = π/2, it follows that Eq. 7.118 reduces to the third equality in Eq.
7.110 when rw is chosen to be equal to zero. Fig. 7.14 depicts the specific PI as a
function of distance along the well for a case where (2rw/L) = 0.0001, which
corresponds, e.g., to a well 1800 m (5,906 ft) long with a diameter of 0.18 m (7 in.).
It can be seen that a disproportionate amount of the inflow occurs at the ends of the
well. In reality the nonlinear effects of friction in the near-wellbore region and the
completion will smoothen the inflow profile and the exact values at the ends are
therefore not particularly relevant. Moreover, the pressure drop caused by friction
inside the horizontal part of the well will result in a higher drawdown at the heel of the
well than at the toe, such that the profile becomes asymmetric. Because the pressure
drop along the well results in a higher drawdown at the heel than at the toe, water or
gas breakthrough is more likely to occur there than at the toe. In long (say, more than
500 m) horizontal wells in highly permeable reservoirs, these effects can be
substantial (see Dikken 1990).
Fig. 7.14—Scaled specific PI J’D = J’ μBo,scξeL/(4kh) as a function of dimensionless distance xD = 2x /L
along the horizontal well for a dimensionless wellbore radius rw,D = 2rw /L = 0.0001. Note: The vertical axis
of the figure has been truncated. The largest value of J’D occurs at xD = ±1 and has magnitude J’D = 5000.

7.4.6 Other Horizontal Well Models. Many other horizontal well models have been
proposed in the literature. One that is often referred to is by Babu and Odeh (1988),
who present an expression for semisteady-state flow toward a partially penetrating
well in a box-shaped reservoir. However, as pointed out by Butler (1994), their
solution assumes a constant specific inflow that therefore results in a somewhat
unrealistically varying wellbore pressure along the well. Another frequently referenced
model is the one by Dikken (1990), who included pressure drop inside the wellbore,
which may become important for long wells in low-permeability reservoirs where high
flow rates result from relatively small drawdowns (i.e., relative to the pressure drop
over the well). This Dikken model is the topic of the MATLAB assignment at the end of
this chapter. In line with the objectives of our book, we intend not to give a full
catalogue of horizontal well models, but to discuss some of the essential modeling
aspects as was performed in the previous sections. For extensive discussions and
further references on horizontal well inflow performance we refer to Joshi (1991),
Butler (1994), and Economides et al. (1998, 2013).

7.5 Semi-Analytical Approach

7.5.1 Complex Well Configurations. After the proliferation of horizontal drilling
during the last two decades of the 20th century, the next innovation in drilling
technology was the introduction of multilateral wells (i.e., of wells with two or more
branches). Fig. 7.15 depicts two examples of typical configurations. For a detailed
overview of the various types of multilateral wells and completions, see Hill et al.
(2008), which also contains a variety of methods to compute inflow performance
relationships (IPRs).

7.5.2 Point Sources. Generally, the pressure field around complex well
configurations can no longer be described with closed-form analytical solutions as
was done for straight vertical and horizontal wells. A detailed description can be
obtained with numerical reservoir simulators which, in addition to allowing for a
complex well configuration, also allow for incorporating reservoir complexity in the
form of layering, fractures, high-permeability streaks, or other geologic
heterogeneities. Although such fully numerical reservoir simulation is outside the
scope of our book, we next briefly discuss an intermediate, semi-analytical technique
that is capable of simulating the pressure around complex wells in homogeneous
reservoirs. The basic element of this approach is the use of linear superposition to
represent wells in the form of a series of point sources or sinks. (Note that because a
sink is just a negative source, from now on we will refer only to sources.)
To illustrate the method, consider again the flow toward a horizontal well, just as in
Subsection 7.3.2. The horizontal flow toward the well is, as before, schematized as
flow toward a vertical fracture with height h and length L, now oriented along the x-
axis between x = 0 and x = L. Using a similar superposition approach as in
Subsections 7.2.3 and 7.2.4, we can represent the pressure field around the fracture
with the aid of N point sources as follows:
Fig. 7.15—Examples of multilateral well configurations: left, dual-lateral well with opposed branches; right,
herringbone well with multiple branches.

where xi and yi are the positions of the point sources, which in this case are given by

According to the notation introduced in Section 7.2.3, we have n = N, c0 = pR, and, as

yet, unspecified values of ci, i = 1,2,…, N, while each of the N sources is shifted over
a distance ΔL. If we would choose equal weights for the coefficients ci, the flow rate
in each of the point sources would have the same value, but the pressure in each of
the sources would be different. For most wells, however, a near-uniform wellbore
pressure is the more likely situation. (Note that the pressure drop along a horizontal
well may become important at high flow rates. For an analysis of this effect, see the
MATLAB assignment at the end of this chapter. Here, we assume that such pressure
drop effects are absent.)
If we require the pressure in each of the sources to be identical, this results in
unequal flow rates in the sources, in line with the uneven specific inflow distribution
along a horizontal well discussed in Subsection 7.4.5. To implement the latter
approach, we can use Eqs. 7.119 through 7.121 to generate the following set of N
equations:

where (xp,i,yp,i), i = 1,2,…,N, are the coordinates of the points where we fix the
pressures. Note that these fixed-pressure points are almost equal to the source
points but have been shifted slightly to be at the wellbore radius—i.e., at (xp,i, yp,i) =
[(i − 1)ΔL, rw ]—to avoid singularities. (Because we arbitrarily chose the fixed-
pressure points to be at the wellbore radius in the positive y-direction, the pressure
distribution becomes slightly asymmetric around the x-axis. However, the effect is so
small that it may be neglected.)
If we require the pressure in all N sources to be equal to the wellbore pressure pwf,
the set of equations (Eq. 7.122) can be rewritten as

where we introduced matrix-vector notation with matrix coefficients Aij given by

Note that the first line of Eq. 7.124 is valid for any configuration of point sources,
whereas the second line is specific for our example of flow toward a well or fracture
oriented along the x-axis. Eq. 7.123 constitutes a system of N linear equations for the
N unknown vector elements qi = ci qo,sc, i = 1, …, N, which can formally be solved as

or, computationally more efficiently, by solving the linear system for q directly. The
total flow rate and the PI then follow as

and

Using MATLAB we can simply use the backslash operator to obtain the result as
q = A \ Delta_p;
q_o_sc_tot = sum(q);
J = −q_o_sc_tot / (p_R − p_wf)
7.5.3 Near-Well Radial Flow and Skin. To include the effects of skin resulting from
near-well streamline convergence and formation damage, we can use Eqs. 7.69 and
7.71 to develop

where Sconv and Sv are given by Eqs. 7.70 and 7.68, respectively; and then apply this
result to modify each of the lines in Eq. 7.122 as

The matrix coefficients Aij in Eq. 7.123 now become

and the resulting equations can be solved as described in the previous section. As an
example, Figs. 7.16 and 7.17 depict the pressure field around a horizontal well with
length L = 500 m, wellbore pressure pwf = 7 MPa, external radii re = 1000 m, and
other properties as given in Table 6.1, represented with 5 and 80 point sources,
respectively. Fig. 7.18 depicts the error in the PI relative to the analytical solution for
an increasing number of point sources. The analytical result obtained is from Eq.
7.112 with be = 2re.
Fig. 7.16—Steady-state pressure field in the horizontal plane (top view) around a horizontal well with
length L = 500 m (indicated with a fat horizontal line), wellbore pressure pwf = 7 MPa, external radii re =
1000 m, and other properties as given in Table 6.1, represented with five point sources (numbers in the
plot indicate pressures in MPa).
Fig. 7.17—Steady-state pressure field around the same horizontal well as depicted in Fig. 7.16 but now
represented with 80 point sources (numbers in the plot indicate pressures in MPa).

7.5.4 Multilateral Wells. As a somewhat more complex example, consider a

multilateral well of the herringbone type, like the one depicted in Fig. 7.15. We use
the same approach as for the horizontal well in the previous example but with a more
elaborate description of the location of the point sources (i.e., of the matrix
coefficients Aij in Eq. 7.124). The result is plotted in Fig. 7.19, which represents the
same horizontal well as in Figs. 7.16 and 7.17, but now equipped with two 150-m
lateral branches. As a result, the PI has increased from 0.95 × 10−9 to 1.02 × 10−9
m3/Pa·s (3.56 to 3.83 B/D-psi)— i.e., an increase of just 5% even though the total
well length has increased 60%. This relatively small increase demonstrates the effect
of pressure interference between the branches. (Note that this result relates to
steady-state flow. Early during the transient flow period of a multilateral well, a much
higher increase of the PI can be observed.) Figs. 7.16, 7.17, and 7.19 can be
reproduced with the MATLAB file example_point_sources.m, which can also be edited
to represent other configurations.
Note that unlike other system element models previously considered, here we
compute the flow rate for a given pressure drop (drawdown) rather than the other
way around. Of course, it is still possible to create an inflow performance relationship
(i.e., a plot of pwf vs. qo,sc) that will be a straight line, just as in Fig. 6.1, because the
entire theory of superposition is based on the assumption of linearity of the governing
equations. If it is necessary to compute the drawdown as a function of flow rate—
e.g., when using a multilateral reservoir model as an element in a cascade of
pressure drop elements—it will be required to use an iterative procedure, in which the
linear nature of the equations allows for convergence in one iteration.

Fig. 7.18—Error in the numerically computed PI, relative to the analytical solution, for a number of point
sources increasing from 2 to 80.
Fig. 7.19—Steady-state pressure field around a multilateral well with a backbone that is identical to the
horizontal well depicted in Fig. 7.16 and equipped with two branches. The well is represented with 121
point sources (numbers in the plot indicate pressures in MPa).

7.5.5 Extended Functionality. Linear reservoir behavior is a good approximation for

single-phase oil flow. Reservoir anisotropy can be included with some additional
programming and using the scaled equations discussed in Sections 7.2.5 and 7.3.7.
Moreover, it is relatively straightforward to extend the semianalytical approach with
point sources for use in single-phase gas flow with the aid of the pseudopressure
formulation that was discussed in Section 6.7.2. The extension to transient or
semisteady-state flow is more elaborate, although conceptually the same approach
can be used starting from the diffusivity equation rather than the Laplace equation.
The next step in complexity is the inclusion of pressure drop in either the branches or
the main wellbore. Finally, with some creativity it is possible to approximate two-
phase flow under solution gas drive conditions by including Vogel- or Fetkovitch-type
relationships as were discussed in Section 6.8.1. However, doing this makes the
equations nonlinear and will therefore require an iterative solution strategy—e.g.,
multivariable Newton-Raphson iteration. We will not discuss these extensions but
refer to Ouyang and Aziz (2001) and Hill et al. (2008) for further information.
7.6 Multiphase Flow: Theoretical Models

7.6.1 Relative Permeabilities. As mentioned briefly in Section 6.4.2, in multiphase

flow through porous media the permeability for each phase is generally a nonlinear,
path-dependent function of the phase saturations. The gas, oil, and water
saturations, Sg, So, and Sw, are defined as the fraction of the pore space occupied by
the corresponding phase, such that by definition Sg + So + Sw = 1. The saturations are
equivalent to the holdups or in-situ volume fractions used in multiphase pipe flow.
Also, the viscosities of the phases differ, as well as the mobilities (i.e., the quotients
of permeability and viscosity for each phase). In the case of three-phase
gas/oil/water flow we need to replace Darcy’s law for single-phase flow, Eq. 6.9, by

where k is the absolute permeability (also known as the homogeneous permeability)

governed by rock properties only, and 0 ≤ krg ≤ 1, 0 ≤ kro ≤ 1, and 0 ≤ krw ≤ 1 are the
relative permeabilities (“rel perms” for short), which are nonlinear functions of the
phase saturations. They represent the reduction of permeability to one phase
because of the presence of the other phases. (See Appendix F for a brief description
of the underlying physics and some semiempirical expressions to describe two-phase
and three-phase rel perms.) We can now modify the radial wellbore flow treated in
the previous sections by introducing relative permeabilities for the phases while
maintaining the assumption of steady-state flow. Starting from Eqs. 6.14 to 6.16,
Eqs. 6.50 to 6.52, and their equivalents for water flow, we can write the multiphase
Darcy equations as

where, for radial flow,

7.6.2 Pressure Gradient. Approximate solutions to Eqs. 7.134 through 7.136 were
obtained by Evinger and Muskat (1942) and Wiggins et. al (1992), and the following
numerical procedure is inspired by their approach.
Introducing the phase mobilities

Eqs. 7.134 and 7.136 can be simplified to

where F is the Forchheimer term

and where and are the in-situ gas/oil and water/oil ratios

Moreover, introducing the total mobility

Eq. 7.135 (i.e., the radial pressure gradient in the oil) can be rewritten as
and therefore as

where we indicated the dependency of the various terms on po(r), Sg(r), and Sw(r).
Eq. 7.150 can be integrated numerically, and the pressure-dependent terms F, , ,
and vo can simply be taken into account. However, the coefficients λt, and F are also
dependent on the phase saturations Sg and Sw, which have yet to be determined.

7.6.3 Phase Saturations. In general, phase saturations in reservoir flow are

nonstationary. After all, it is usually the intention to reduce the hydrocarbon
saturations in a reservoir to the lowest possible value. In some special cases it is
possible to derive relationships between pressure and saturation in a given reservoir
geometry and under given production conditions, known as saturation-pressure paths
(Walsh and Lake 2003). Here we will consider the derivation of a steady-state path,
which is valid for a circular reservoir with steady-state well production and steady-
state boundary inflow of fluids with a constant composition. Such a steady-state
relationship between pressure and saturation can be obtained from Eqs. 7.134
through 7.136 by considering that, disregarding capillary pressures, the pressures pg,
po, and pw have to be equal, and therefore also the gradients dpg /dr, dpo /dr, and dpw
/dr. We therefore aim to find the saturations Sg and Sw at each point r that result in

The system of nonlinear equations (Eqs. 7.151 and 7.152) cannot be solved
explicitly for Sg and Sw because the expressions for the relative permeabilities are too
complex. Instead, we can use an iterative method such as the multivariable Newton-
Raphson scheme described in Appendix D. Using matrix notation, a single Newton-
Raphson step can be expressed as
where k is the iteration counter. Solution of Eq. 7.153 requires solving a linear system
of two equations for the two error terms ΔSg and ΔSw, after which Eq. 7.154 gives us
the new estimates (Sg)k+1 and (Sw)k+1. Starting values for the gas and oil saturations
can be obtained from the closed-form expressions for two-phase flow discussed
below. The Newton-Raphson procedure has been implemented in MATLAB file
saturations.m, which is used by res_dpdr.m to numerically integrate Eq. 7.150. Just
like the files res_oil_dpdr.m and res_gas_dpdr.m, the file res_dpdr.m can be called
by res.m to compute the difference between the pressure at the reservoir boundary
at r = re and the pressure in the wellbore, or to create an IPR (see
example_res_pres.m and example_IPR.m). Note that these files are valid for vertical
wells only. However, in theory, similar equations could be implemented for horizontal
well flow.

7.6.4 Example Gas/Oil Flow. Fig. 7.20 displays the steady-state IPR for a well with
properties given in Table 6.3 for two values of Rgo, identical to those used in Fig. 6.11.
It is assumed that Rgo is identical to Rsb because there is no free gas influx. The
water is not mobile and the flow is therefore two-phase. The initial reservoir pressure
is above the bubblepoint pressure for the low-GOR case (Rgo = 50 m3/m3) and below
it for the high-GOR case (Rgo = 150 m3/m3). The results in Figure 7.20 have been
computed with the aid of res.m and res_dpdr.m. They correspond reasonably well
with the Vogel solutions from Fig. 6.11 (which are here indicated with dotted lines).
Note, however, that the steady-state assumption is usually not applicable. Moreover,
the approximation of the oil permeability ko,ref in Eq. 6.71, used to create the low-GOR
case in Fig. 6.11, is also debatable, and therefore the discrepancy between the
results of empirical and more theoretical models may, in general, be much larger than
illustrated in Fig. 7.20.
During the initial stage of a solution gas drive, the gas saturations in the reservoir
remain relatively low because only when the saturations in the entire reservoir have
reached the critical value Sgc does the gas become mobile over larger distances.
Moreover, at reservoir pressure gas occupies a relatively small volume, and the much
higher mobility of gas than of oil results in a holdup of the flowing oil relative to the
flowing gas. This is illustrated in Fig. 7.21, generated for qo,sc = −12 × 10−3 m3/s, Rgo
= Rsb = 50 m3/m3, and pR = 9.5 MPa (i.e., just above the bubblepoint pressure of 9.0
MPa). A jump in gas saturation occurs at r = 280 m (i.e., at the point where the
pressure drops below the bubblepoint), and the magnitude of the jump is
approximately equal to the critical gas saturation Sgc = 0.15. In a more realistic,
nonsteady-state situation the jump will gradually move outward until it reaches the
outer boundary, whereafter the gas becomes mobile throughout the entire reservoir,
resulting in a significant increase in the producing GOR.

Fig. 7.20—IPR of an oil well with parameters given in Table 6.3 for two different values of the solution GOR
as indicated in the legend. The corresponding bubblepoint pressures are 9.00 MPa (1,306 psi) and 22.7
MPa (3,288 psi), respectively (arrow indicates point at which the low-GOR curve drops below bubblepoint
pressure, and dotted lines indicate Vogel solutions of Fig. 6.11).
Fig. 7.21—Gas, oil, and water saturations near an oil well with parameters given in Table 6.3. The figure
corresponds to a flow rate qo,sc = −12 × 10−3 m3/s and a GOR of 50 m3/m3. The bubblepoint pressure is 9.0
MPa and the reservoir pressure at the boundary is 9.5 MPa.

7.6.5 Closed-Form Approximations for Primary Recovery. If we consider two-

phase (oil/gas) flow and neglect the Forchheimer effect, Eq. 7.151 can be rewritten
as

The right side of this expression is a function only of pressure, while the left side is a
function only of the oil and gas saturations. With the aid of the Corey expressions for
the relative permeabilities, as discussed in Appendix F, this result can also be written
as

where
which is the scaled gas saturation. The parameters , , ng, nog, Sgc, Sorg, and Swi
are explained in Appendix F and are assumed to be known from laboratory
experiments. Using the additional assumption that the Corey exponents ng and nog
have an identical magnitude n, Eq. 7.156 can be rewritten as

from which we can solve for Sg as

Steady-state gas/oil flow, as described by Eqs. 7.150 and 7.159, may occur in a
near-well region when the reservoir is initially above bubblepoint pressure but
gradually loses pressure, either locally around the well because of drawdown or over
the entire reservoir because of depletion.
An expression similar to Eq. 7.159 can be obtained for steady-state oil/water flow:

It may be of use to describe a gradual increase of water saturation in case of the

influx of bottom water or aquifer water into a well. Eqs. 7.159 and 7.160 have been
implemented in saturations_gas_oil.m and saturations_oil_water.m.

7.6.6 Water and Gas Fronts in Secondary Recovery. During secondary recovery,
water or gas is injected in dedicated injection wells, and the saturation distribution in
the reservoir is essentially nonstationary and typically displays abrupt changes. In a
single homogeneous layer, such a front may be sharp, and the corresponding water
or gas breakthrough in the producers will occur suddenly with corresponding changes
in the near-well mobility. If the reservoir is strongly layered or otherwise
heterogeneous, the front arrival will be more smeared and the associated mobility
change less abrupt.
During secondary recovery, it may still be acceptable to use the (near-)steady-
state expression for the radial pressure gradient (Eq. 7.150), but with slowly time-
varying parameters Sg and Sw. Expressions for the saturation change in a circular
reservoir during water- or gasflooding can be obtained by starting from conservation
equations for the different phases similar to Eqs. 6.4 through 6.7 after modifying them
to include multiphase flow (i.e., relative permeability) effects. They can be
manipulated to produce a differential equation for Sg or Sw as a function of time and
the radial coordinate r, known as the (radial) Buckley-Leverett equation, which allows
for an analytical solution. Alternatively, the equations can be solved numerically with
the aid of reservoir simulation software, in which case it is also possible to model
more realistic heterogeneous, noncircular reservoir geometries. However, both
approaches are outside the scope of this book.

7.6.7 Coning and Cusping. A typical multiphase feature of near-wellbore flow is the
occurrence of disproportionately high amounts of gas at pressures below the
bubblepoint pressure, resulting in producing GORs Rgo far in excess of the initial GOR
of the reservoir fluid Rsb . This effect was not captured in the steady-state analysis
described above because it was assumed that oil and gas flow as a homogeneous
mixture with mass flow rates that remain constant along the radius r. In reality, often
a more segregated flow situation occurs in which gas from the gas cap is drawn to
the perforations through coning or cusping (see Fig. 7.22). Because of the extremely
high mobility of gas compared to that of oil, even a small contact area between the
gas and the wellbore may result in a very large gas influx. A similar effect in the form
of water coning may occur for oil/ water flow, especially when the oil viscosity is much
higher than the water viscosity. In horizontal wells the cones will be of an elongated
shape with a ridge parallel to the well. If pressure drop along a horizontal well plays
an important role, coning will first occur at the heel of the well where the drawdown is
highest.

Fig. 7.22—Gas coning (left) in a horizontal reservoir, and gas cusping (right) in a dipping reservoir (dashed
lines indicate initial GOC).

The combined multiphase flow effects of relative permeabilities (leading to the

formation of saturation fronts), gravity (leading to coning), and capillary pressures
(leading to more-diffuse saturation distributions) are complex and require advanced
numerical simulation of the near-well reservoir flow to be described accurately, which
is usually beyond the level of engineering performed in production operations.
Moreover, all these effects depend strongly on formation heterogeneities such as
high-permeability lenses, crossbedding, layered stratigraphy, faults, fractures, and
cemented regions, the position and size of which are usually poorly known, if known
at all. The capacity to theoretically predict the inflow performance of a well under
multiphase flow conditions is therefore quite limited.

7.7 Related Topics Not Considered in This Chapter

Chapters 6 and 7 together present an overview of the most important (semi)analytical
techniques to compute (semi)steady-state well inflow performance and to generate
inflow performance curves for nodal analysis. We did not treat methods for transient
reservoir flow, for which we refer to the textbooks mentioned in Section 6.10 and
more specialized texts on well testing [e.g., Kamal (2009)]. Transient flow is also
important for the development of very-low-permeability reservoirs (e.g. for shale gas
production), where it may take a long time before a (semi)steady state is reached.
Another aspect of importance for such reservoirs is the use of hydraulic fractures,
usually in the form of multiple vertical fractures in horizontal wells, which also require
a non-steady-state approach to capture their inflow performance. Moreover, also for
(semi) steady-state flow there are analytical approaches available that we did not
cover (see, e.g., Chin 2002). Overall, however, the methods discussed in our text
should give you sufficient background to read and understand the many conference
and journal papers available in this field.

7.8 Questions

7.1 In Section 7.2.2 we listed various combinations of radial boundary conditions

to solve the Laplace equation in polar coordinates. Would it also be possible to
use the following conditions? Explain your answer.

7.2 What is the difference between the skin factors Sh and Sv in Eq. 7.68?
7.3 Why is the theoretical prediction of well inflow performance under multiphase
conditions possible only to a limited extent?
7.4 Derive an expression for the steady-state PI of a vertical well located at a
distance D/2 from a vertical sealing fault in a circular homogeneous isotropic
reservoir with a large drainage radius. How do you interpret your result if D
approaches zero?
7.5 Starting from the answer to Question 7.4, compute a numerical value for the
PI of a well that is located 50 m from the fault. Use the data in Table 6.1
(values for “other figures”) and compare the result to the PI for a well in the
 same reservoir without a fault. Does the difference make sense?
7.6 Derive Eq. 7.55 starting from Eq. 7.54. (Because this question is somewhat
atypical and more a test of algebraic than petroleum engineering skills, you
may want to skip it if you are less interested in the former.) Make use of the
following identities:

Hint: You may want to first bring the equation in the following intermediate
form:

Note the different summation limits in the infinite products. After using identities
(Eqs. 7.163 to 7.165), you will end up with the pressure distributions as a
function of coordinates y and z (i.e., Eq. 7.55) and with several additional
irrelevant constants. Why are they irrelevant?

Note: for the origin of the expressions in this question, see Eqs. 1.435 and
1.438 in Gradshteyn and Ryzhik (1980). The derivation of the infinite product
terms requires analysis beyond the scope of our book (see Section 7.5 of
Whittaker and Watson 1915).
7.7 Consider a horizontal well in a box-shaped horizontal reservoir like the one
depicted in Fig. 7.6.
(a) If we disregard the effect of streamline convergence and if we choose
Pref = PR, the well has a specific PI of 0.12 B/D-psi-ft. For a reservoir
pressure of 5,500 psi, a bottomhole pressure of 5,490 psi, and a well
length of 1,500 ft, compute the drawdown and the production rate of the
 well.
(b) The specific PI including the effect of streamline convergence is 0.10
B/D-psi-ft. How large is the pressure drop resulting from streamline
convergence?
(c) If we take the pressure drop along the well into account and the pressure
at the heel of the well (i.e., at x = 0) remains the same as before, how
does that change the specific inflow along the horizontal well? Give a
qualitative answer only, and sketch the inflow profile—i.e., a plot of q’ (x)
vs. x.
7.8 Consider Eq. 7.81 for semisteady-state flow toward a horizontal well. Starting
from this equation, derive an expression for the average reservoir pressure
and the corresponding PI (i.e., Eq. 7.84).
7.9 Repeat Question 7.4 for a situation with anisotropic permeabilities. Express
your results in terms of the equivalent isotropic permeability and the
anisotropy Ixy index as defined in Eqs. 7.37 and 7.45.
7.10 (a) Compute the aspect ratio Rasp = ae / be and the PI for a partially
penetrating horizontal well in an elliptic reservoir with short axis be = 500
m, kh = 5kv = 2.0 × 10−12 m2, and all other parameters as given in Table
7.1. Assume steady-state conditions.
(b) Compute the aspect ratio Rasp = L / w and the PI for a fully penetrating
horizontal well in a box-shaped reservoir with the same permeabilities as
defined under a) and all other parameters as given in Table 7.1. Assume
steady-state conditions. Explain the difference with the results obtained
under Questions 7.10(a).
(c) The surface area of an ellipse is defined as

Repeat Question 7.10b but choose w such that the horizontal surface area of
the box becomes equal to that of the ellipse. What are the corresponding
aspect ratio and PI for this new surface area?
7.11 Consider a single vertical well draining a near-circular area in an
undersaturated reservoir. After opening up the well, the pressure drops below
the bubblepoint in the near-wellbore region. Compute the resulting saturations
for parameters given in Table 7.3, while neglecting the Forchheimer term.

Question 7.12 requires the use of MATLAB.

7.12 Reproduce Fig. 7.17 with the aid of file example_point_sources.m. Increase
the formation-damage skin factor to a high value, say 5. Explain the result.
7.9 MATLAB Assignment: Horizontal Well Inflow and Pressure Drop
7.9.1 Objectives

• Become familiar with the Dikken (1990) method to describe combined

horizontal well inflow and pressure drop
• Learn to generate an IPR for a long horizontal well
• Learn to solve a two-point boundary value problem in MATLAB with the aid of the
shooting method

Property Value Units

Bg 0.07 –
Bo 1.20 –
0.70 –
0.95 –
ng 4 –
nog 4 –
qg,sc 3.90 m 3/s
qg,sc 65 x 10−3 m 3/s
rs 0 m 3/m 3
Rs 55 m 3/m 3
Sgc 0.10 –
Sorg 0.05 –
Swi 0.20 –
μg 13.5 x 10−6 Pa · s
μo 5.8 x 10−3 Pa · s
ρg,sc 0.93 kg/m 3
ρo,sc 820 kg/m 3
Table 7.3—Input data for multiphase reservoir flow for Question 7.11.

• Obtain a feeling for the order of magnitude of horizontal well pressure drop as
a function of reservoir and fluid parameters

7.9.2 Assignment. For this assignment, consider the same horizontal well
configuration used in Question 7.7 and displayed in Fig. 7.6.
Theory. If the pressure drop along the wellbore is taken into account, the wellbore
flow and reservoir flow equations need to be considered together. This was first done
by Dikken (1990), who observed that the specific reservoir inflow should equal the
decrease in wellbore flow rate:
where x is the coordinate along the well (with the origin at the heel and positive values
in the direction of the toe), q is the wellbore flow rate (expressed at in-situ conditions
and taken as positive for flow from the heel to the toe), q’ is the specific inflow from
the well into the reservoir (expressed at in-situ conditions and taken as positive for
injection), J’ is the specific PI (which is positive by definition), p is the wellbore
pressure, and pR is the pressure at 0 < x < L, y = ±w/2, z = 0.
A key assumption is that all reservoir flow is perpendicular to the well; i.e., that
there is no flow in the reservoir in the along-well direction. If we restrict the analysis
to incompressible single-phase oil flow and neglect the effects of acceleration and
gravity, the combined equations for wellbore pressure and flow rate are obtained by
combining Eqs. 3.28 and 7.168:

where we refrained from using subscripts o (for oil) to simplify the notation and where
all flow rates are expressed at in-situ (local) conditions. Eqs. 7.169 and 7.70 form a
system of two coupled first-order differential equations. The boundary conditions can
be specified as

where is the prescribed FBHP at the heel of the well and L is the wellbore length.
Eq. 7.172 implies that we neglect semiradial inflow at the toe of the well (i.e., that the
well is fully penetrating). Unfortunately, the two boundary conditions are specified at
different ends of the well, which means that we cannot simply numerically integrate
the equations starting from one end of the well to the other. A convenient way to solve
such a two-point boundary value problem is to use a shooting method, which starts
from the known value for p(0) = at the heel and a guessed value for q(0) = at
that same location. After integration from the heel to the toe, the resulting value of
q(L) at the toe is most likely not equal to zero—i.e., not meeting the boundary
condition (Eq. 7.172). An iterative procedure can then be applied to change the value
of q(0) until the value of q(L) is (almost) zero.
The expression for J’ (x) in Eq. 7.170 depends on the reservoir geometry and
properties. For this assignment use the (constant) expression for J as given in Eq.
7.60, divided by the well length:
(Note that q’ is expressed at in-situ conditions). A first guess of the total well flow
rate qtot then follows as

which is equal to the result for a horizontal well without pressure drop.

Tasks
• Implement Eqs. 7.169 through 7.172 in a MATLAB file:
○ Create an auxiliary file hor_dfdx.m suitable for use with standard MATLAB
numerical integration routines, which gives as output dfdx = [dpdx;dqdx]
— i.e., the derivatives dp/dx and dq/dx—as function of input values f =
[p,q] and other, stationary, parameters:
function dfdx = hor_dfdx(~,f,~,d,e,h,k,mu,p_R,rho,w)
Here the variables p and q should be scalar variables of wellbore pressure
and wellbore flow rate at a certain point along the well.
○ Create a second auxiliary file hor.m that performs a single integration from
x = 0 to x = L for given values of p(0) = (known) and q(0) =
(guessed). If necessary, consult Appendix D for information on numerical
integration in MATLAB, or see pipe.m and oil_dpds.m for an example. The
first output element of hor.m should be the flow rate at the toe to enable the
direct use of hor.m with fzero:
function [q_toe,p,q,x] = …
hor(d,e,h,k,L,mu,p_R,p_wf,q_tot,rho,w)
The output variables p, q, and x are now row vectors of wellbore pressure,
wellbore flow rate, and coordinates at n points along the well as determined
by the numerical integration routine called within hor.m. The flow rate at the
toe is therefore equal to the last element of q.
○ Call hor.m from your main file to iteratively compute the correct value of
q(0). Apply the shooting method and use the standard MATLAB routine fzero
to perform the iteration. If necessary, consult Appendix D for information on
fzero, or review Questions 3.12 and 6.14, and their solutions, for examples
of its use.
q_tot = fzero(@(q_tot) …
hor(d,e,h,k,L,mu,p_R,p_wf,q_tot,rho,w),q_tot_tilde);
• Create plots of the wellbore pressure p, the wellbore flow rate q, and the
specific well inflow rate qs, as a function of distance from the heel x.
 • Compute and plot the IPR of the well with parameters given in Table 7.4. Here
the IPR is a graph of the FBHP at the heel, , vs. the total well flow rate qtot.
Note that in all previous examples, determining the IPR implied computing the
FBHP by varying the well flow rate, whereas here you need to compute the
flow rate by varying the FBHP. Choose a pressure range 32.0 MPa < < 33.5
MPa. Plot your results in both SI and field units.
• Perform a sensitivity study into the effect of the reservoir, fluid, and wellbore
parameters on the wellbore pressure drop. Express your results in terms of a
dimensionless wellbore pressure drop, defined as the pressure drop between
the toe and heel, divided by the drawdown at the heel:

Property Symbol Value Units

Reservoir height h 25 m
Reservoir length L 5000 m
Reservoir width w 500 m
Well radius rw 0.10 m
Well roughness e 30 × 10−6 m
FBHP (at the heel) 33.0 MPa
Reservoir pressure pR 33.5 MPa
Oil density ρ 850 kg/m
Oil viscosity μ 20 × 10−3 Pa · s
Permeability k 2.0 × 10−12 m2
Skin Sh 0 –
Table 7.4—Parameter values for horizontal well MATLAB assignment.

7.9.3 Deliverables

• MATLAB program listing

• Plots
• Sensitivity study
Chapter 8

Well Performance

8.1 What Is Covered in This Chapter?

In this final chapter we use nodal analysis to describe the combined performance of
the various system elements discussed in previous chapters. We address simple
production systems in the form of oil and gas wells producing from one or more
reservoirs into surface facilities, usually through a choke and flowline. We
demonstrate the use of nodal analysis to quantify the effects of changing dimensions
or properties of system elements, and to optimize both short- and long-term well
performance, including the use of artificial lift methods.

8.2 Analyzing Well Performance

Optimizing a well design or interpreting measurements from a producing well requires
tools to analyze the well performance—i.e., its flow behavior under a given set of
reservoir and operating conditions. For example, in designing a completion for a new
well, we need to assess the effect of the tubing size on well productivity and predict
the productivity change as the reservoir pressure declines. Similarly, in a producing
well, we need to decide when it makes economic sense to carry out operations to
increase the productivity—e.g., whether the skin is significantly affecting production
and the well needs stimulating—or when we should change out the tubing. For further
reading on well performance in general, we refer to the classic paper of Gilbert
(1954) and the textbooks of Nind (1964, 1981), Golan and Whitson (1991),
Economides et al. (1998, 2013), and Guo et al. (2007).

8.2.1 Performance of a Well Operating at Given Tubinghead Pressure. Consider

a vertical well that is operated at a fixed flowing tubinghead pressure (FTHP) ptf. The
flow system between the reservoir and the tubing head can be broken down into

• Inflow into the well

• Flow up the tubing to the tubing head

The total pressure drop between the reservoir and the wellhead is made up of the
drawdown associated with the inflow from the reservoir and the vertical flow pressure
drop. In oil wells the vertical pressure drop typically makes up 70–90% of the total
pressure drop between the reservoir and the wellhead; in a gas well this can be much
less because gravity effects are not so important and the pressure drop is dominated
by friction. As we have seen in Chapter 6, the inflow into the well is affected by
 • Reservoir pressure
• Reservoir geometry
• Reservoir properties
• Skin, including “skin” resulting from the completion
• Properties of the reservoir fluids

As discussed in Chapter 6, for a given reservoir pressure all these effects can be
brought into a single relationship between oil flow rate and FBHP, known as the inflow
performance relationship (IPR) of the well. Similarly, the flow up the tubing is affected
by

• FTHP
• Tubing size and other completion parameters
• Flow regime in which the well operates
• Fluid properties

As discussed in Chapters 3 and 4, for a given FTHP, all these effects can also be
brought into a relationship—the tubing intake curve—between oil flow rate and flowing
bottomhole pressure (FBHP).
As an example, consider Fig. 8.1, which depicts the IPR of an oil well for fluid
properties listed in Table 4.1. The reservoir properties are as shown in Table 6.3,
except for the absolute permeability and the reservoir conditions, which have now
been chosen as k = 7.0 × 10−13 m2 (3,546 mD), pR = 25 MPa (3,626 psi), and TR =
120°C (240°F). In this case, because the FBHP remains above the bubblepoint
pressure over the entire flow rate range, the IPR is a straight line. Fig. 8.2 depicts
the tubing intake curve corresponding to Table 4.1 (Fig. 4.8). Combining these two
curves gives us Fig. 8.3. This is an example of combining the downstream and
upstream pressure drop curves (i.e., the tubing intake curve and the IPR) as was
discussed in Section 1.6, where we covered nodal analysis. The operating point of
the well corresponds to a FBHP of 23.6 MPa (3,423 psi) and an oil flow rate of –3.7
× 10−3 m3/s (–2,011 STB/D).
If we lower the FTHP, the FBHP will drop also and the tubing intake curve will shift
down. In addition, the curve may somewhat change in shape because the boundaries
between the various flow regimes in the tubing may also move. However, the overall
effect will be a shift of the operating point to the right, corresponding to an increased
flow rate. Conversely, if we increase the FTHP, the production rate will drop; and if
we increase it too much, the well will no longer flow at all (see Fig. 8.4).
A reduction in flow rate also occurs if the FTHP remains constant but the reservoir
pressure drops, a situation that frequently happens when an oil field is being
depleted. An example is given in Fig. 8.5, where a 10% drop in reservoir pressure
results in just a 9% decrease in FBHP but a 54% decrease in oil flow rate. Moreover,
the curves now have two intersection points. As will be demonstrated in the next
section, one of these points is unstable. The intersection to the right is a stable
operating point and corresponds to the actual FBHP and flow rate, whereas the
intersection to the left is an unstable and therefore physically unrealistic operating
point. If the reservoir pressure keeps dropping further, a point will be reached at
which the well stops flowing. It may sometimes be possible to bring the well back to
production by installing a new tubing with a lower pressure drop. It is interesting that
this may be either a larger- or smaller-diameter tubing, depending on factors such as
the water cut and the gas/oil ratio (GOR). Alternatively, it may be necessary to install
a form of artificial lift, such as gas lift, an electrical submersible pump (ESP), or a
beam pump.

Fig. 8.1—Inflow performance curve of an oil well with reservoir pressure pR = 25 MPa (3,626 psi).
Fig. 8.2—Tubing intake curve of an oil well (identical to Fig. 4.8).

Fig. 8.3—Combined plot of inflow performance and tubing intake curves, determining the FBHP (pwf = 23.9
MPa (3,466 psi)) and production rate (qo,sc = –3.7×10−3 m3/s (–2,011 B/D)).

Fig. 8.4—Effect of a too-large increase in FBHP (resulting from a too large increase in FTHP): the tubing
intake curve and the IPR no longer intersect. As a result, the well will no longer flow. The minimum rate at
which the well can flow is therefore approximately −0.5 × 10−3 m3/s (−272 B/D).
Fig. 8.5—Effect of a drop in reservoir pressure. The reservoir pressure has decreased with 10% from 25
MPa (3,626 psi) to 22.5 MPa (3,263 psi), and as a result the FBHP has decreased with 9% from 23.9 MPa
(3,466 psi) to 21.8 MPa (3,162 psi) and the oil production with 54% from −3.7 × 10−3 m3/s (−2,011 B/D) to −1.7
× 10−3 m3/s (−924 B/D). The picture also depicts the presence of two operating points (a stable and an
unstable one) for the case of low reservoir pressure.

8.2.2 Intermezzo: Stability of an Operating Point. Because the relationship

between pressure drop and flow rate is nonlinear for most elements of the production
system, p-q graphs are usually curved. In particular, for wells that produce a mixture
of gas and liquids (as do most oil wells), this may lead to a situation in which the
graphs intersect at two points (e.g., as depicted in Fig. 8.5). At a first glance we
could therefore conclude that such a well could produce at two different flow rates. In
reality, however, wells will always produce at only one of these flow rates. The
reason is that only one of the operating points is stable, whereas the other one is
unstable.
To understand why an operating point can be unstable, we need to consider the
dynamics of the system. Nodal analysis, however, usually considers only the steady-
state relationship between pressure drop and flow rate. For example, the curves in
Fig. 8.5, which were determined from a steady-state analysis, can be represented
schematically as
where fus and fds are nonlinear functions representing the upstream and downstream
curves (i.e. the IPR and the tubing intake curve), respectively. The flow rate q refers
to the oil flow rate at standard conditions qo,sc, but in this section we drop the
subscript “o,sc” to avoid confusion with the subscript “0” (zero) used to indicate an
operating point. In an operating point (p0, q0), we find that

We are interested in the effect of small disturbances on the flow in the

neighborhood of an operating point:

Because we consider only small disturbances, we can linearize fus and fds. In other
words, we can take the Taylor expansions for fus and fds around q0, defined as

where primes indicate differentiation with respect to q. If we substitute the

expansions (Eqs. 8.7 and 8.8) into Eqs. 8.3 through 8.6, subtract the results, and
maintain only the constant and linear terms, we obtain linear relations for and :

Small fluctuations in flow rate imply that the flow accelerates and decelerates with
small amounts. These accelerations cause pressure fluctuations, which we can
incorporate in Eqs. 8.9 and 8.10 by adding inertia terms. As before, we assume that
fus and fds represent the pressure drops over the upstream and downstream parts of
our system, respectively. Including inertia, we can write

where ius ≥ 0 and ids ≥ 0 represent inertia of the fluid upstream and downstream of the
analysis point. The two differential equations (Eqs. 8.11 and 8.12) are dynamic
relationships between the pressure and the flow rate in the analysis point as
governed by the upstream and the downstream parts of the system. The two
acceleration terms have different signs because an increase in pressure in the
analysis node causes a deceleration of the flow in the upstream part of the system
and an acceleration of the flow in the downstream part. In an operating point the
pressures resulting from the upstream and the downstream parts have to be equal,
and therefore we can write

which can be rewritten as

Eq. 8.14 is a linear first-order differential equation for . To obtain a complete

solution, we need one initial condition. If we assume that a flow rate disturbance with
magnitude takes place at time t = 0, the initial condition becomes

Solving Eq. 8.14 results in

where C is an integration constant that can be determined with the aid of the initial
condition, expressed in Eq. 8.15. That results in C = , and therefore

Eq. 8.17 represents an exponentially growing or decreasing magnitude of

disturbance with time, depending on the sign of /(ius + ids). Because ius ≥ 0
and ids ≥ 0, it is the sign of that determines the stability of the flow in an
operating point. The terms and can be interpreted as the slopes of the curves at
the operating point. Note that a curve that decreases with increasing q has a negative
slope. Referring back to Fig. 8.5, we find that the operating point to the left is
unstable, and the one to the right stable. Stable wellbore flow can therefore occur
only at the pressure and flow rate corresponding to the operating point to the right.
(Note that in a full dynamic analysis, we should also take into account the
compressibility of the gas/liquid mixture. In that case we would arrive at a second-
order differential equation, which could display oscillatory solutions. However, the
basic results of the stability analysis would not change.)
As discussed below, in practice the criteria for stability of an operating point are
chosen slightly differently. However, it is safe to say that in case of two operating
points, it will always be the one at the highest flow rate that is stable.

8.2.3 Performance of a Well Operating Through a Surface Choke. In Section

8.2.1, assuming that the FTHP and the reservoir pressure were constant, we
calculated the resulting flow rate from the pressure drops over the tubing and the
reservoir. Therefore we could analyze the well productivity using nodal analysis with
just two elements between three nodes (see Fig. 8.6, left). However, usually the
FTHP itself will also vary with flow rate. To analyze this effect we need also to
consider the flow path downstream of the wellhead, up to a point where we can
assume that the pressure remains more or less constant. That point will normally be
the first separator, which is usually operated with automatic pressure control. The
manifold pressure, which is almost the same as the separator pressure, can then be
taken as the fixed downstream pressure for a wellbore pressure drop analysis. Recall
that we have also assumed constant pressure at the other end of the system, (i.e., in
the reservoir), or at least a pressure that varies only very slowly over time. In
conclusion, our problem is to determine the flow rate by considering the pressure
drop over the following elements:

• Reservoir (including near-wellbore region and completion)

• Tubing up to the tubing head
• Choke
• Flowline

Assume constant pressures in the reservoir and the manifold, and a constant fluid
composition [gas/oil ratio (GOR), water cut].
Here we restrict ourselves to the analysis of a three-component system consisting
of a choke, a tubing, and the near wellbore. In that case a suitable fixed downstream
pressure is the flowline pressure pfl (i.e., the pressure just downstream of the
wellhead choke); see Fig. 8.6, right. For this type of analysis it is common practice
to choose the analysis node at the top of the tubing and to establish a single
performance curve for the choke and a combined performance curve for the tubing
and the near wellbore. The latter is obtained by combining the elements for near-
wellbore pressure drop and tubing pressure drop in series. For a given reservoir
pressure, this gives the tubing performance curve—as you will recall from Chapter 3,
this is the relationship between the oil flow rate and the FTHP. The tubing
performance curve gives a total picture of the deliverability of the well. Note that the
name is slightly misleading because the tubing performance curve is dependent not
just on the ability of the tubing to transport the fluids but also on the performance of
the near-well reservoir and the completion. As the reservoir pressure declines, the
tubing performance curve will change. So a better name would be well performance
curve, but this is not normally used. The flow through the choke is governed by a
choke performance curve as discussed in Chapter 5. Note that depending on whether
the choke flow is critical or not, we do or do not need to specify the downstream
pressure (i.e. the flowline pressure).

Fig. 8.6—Nodal analysis configurations, with overbars indicating fixed pressures. Left: Fixed FTHP and
fixed reservoir pressure; analysis node at the bottom of the well. Right: Fixed flowline pressure and fixed
reservoir pressure; analysis node at the top of the well.

We can now plot the choke performance curve and the tubing performance curve
together to determine the operating point at the tubing head. As an example consider
a gas well with fluid and tubing properties given in Table 3.1 and reservoir properties
given in Table 6.2, except for the permeability and the Forchheimer coefficient, which
have now been chosen as kg = 6.4 × 10−15 m2 and β = 9.8 × 109 m−1. Fig. 8.7 depicts
the IPR and the tubing performance curve in the same graph. Note that the reservoir
pressure and the closed-in tubinghead pressure are obtained by reading the graph for
zero flow rate. Fig. 8.8 depicts the tubing performance curve of the same well on an
expanded scale, and Fig. 8.9 shows the choke performance curves for a number of
different choke sizes. The combined performance curves have been plotted in Fig.
8.10, and it can be seen that three of the operating points result in choked flow (i.e.,
flow in the critical regime where the upstream pressure is independent from the
downstream pressure). The other operating point corresponds to sub-critical flow.
Fig. 8.7—Inflow performance curve and tubing performance curve of a gas well. The reservoir pressure is
40 MPa (5,802 psi), and the closed-in tubinghead pressure is 31.8 MPa (4,612 psi).
Fig. 8.8—Tubing performance curve of the same gas well on an expanded scale.

Fig. 8.9—Choke performance curves for a gas well (identical to Fig. 5.4 where the upstream pressure p1 is
now the FTHP).
Fig. 8.10—Combined plot of tubing and choke performance curves for a gas well, determining the FTHPs
and production rates for different wellhead chokes.

8.2.4 Summary of Analysis Methods. As described, there are two commonly used
methods for analyzing the production from oil or gas wells. In each method the
behavior of the system is reduced to two relationships between pressure and flow
rate. In the first method, the analysis is performed downhole. Using the condition at
the tubing head and the pressure drop in the tubing, the tubing intake curve is
calculated at the bottom of the well. The intersection of this curve with the IPR
determines the production rate. This was illustrated in Fig. 8.3. In the second method,
the analysis is performed at the tubing head. Using the inflow performance curve and
the pressure drop in the tubing, the tubing performance curve is calculated at the
tubing head. The intersection of this curve with the choke performance curve
determines the production rate. This was illustrated in Fig. 8.5.
As discussed in Chapter 1, the choice of the analysis nodes at either the top or the
bottom of the tubing is rather arbitrary, and before the widespread use of computers,
was determined by use of gradient curves for tubing flow. In a cascade system of
components the analysis could be performed anywhere at or between the two
system boundaries. Indeed, using a computerized analysis method it would be logical
to analyze the pressure drop over the entire system with an algorithm marching from
one boundary to the other. The program would then need to change the flow rate in
an iterative fashion until the pressure drop over the system exactly matches the
known difference in pressure between the two boundaries. The advantage of
displaying the traditional combination of two curves is that it provides a quick insight
into the flow behavior of two separate system parts in one graph: one downstream
and one upstream of the analysis node.
In addition, for chokes operating in the critical regime it is not possible to compute
the pressure drop over the choke for a given downstream pressure, and the
upstream location (i.e. typically the top of the tubing) is therefore a logical analysis
node. Similarly, for semi-empirical IPRs such as the Vogel expression it is not
possible to compute the pressure drop over the reservoir for a given upstream
pressure, and the downstream location (i.e. typically the bottom of the tubing) is
therefore a logical analysis node. For these reasons, most computer programs for
nodal analysis still have an option to display the traditional tubing intake and
performance curves.

8.3 Field Development Planning and Field Management

Well performance analysis plays a crucial role in field development planning for new
fields and the management of producing fields. For a new field, a number of critical
questions must be answered:

• What form of completion must be installed in the well, and in particular what
size tubing?
• For the chosen tubing size, what is the initial production rate of the well, and
how will this vary with time?
• How long will the well be able to produce? When is the optimal time to change
the tubing size or switch to artificial lift (pumping or gas lifting)?

For a producing field, the following must be considered:

• Is the well producing as expected?

• Is the well impaired and in need of stimulation to remove the skin?
• Can the well production be improved by changing out the tubing or installing
artificial lift? Is the cost justified?
• Can the well performance be improved by increasing the perforated interval?
• Can the well performance be improved by lowering the FTHP or changing out
the choke?
• How long will the well produce? Do plans need to be made for artificial lift?

These issues cannot be decided by a production engineer in isolation; it must make

economic sense to increase the production of a single well within the total field plan,
taking into account the capacity of the surface facilities. However, the production
engineer can greatly improve the economic return from a field by continuously
monitoring all wells and taking remedial action when production declines.
8.3.1 Improved Inflow Performance. If the well has become impaired, then it can be
stimulated (acidized) to improve the inflow performance. Similarly, inflow performance
may be improved by additional perforating. To quantify the decreased production from
impairment, or the increased production from stimulation, it is useful to analyze what
happens downhole. For example, Fig. 8.11 shows the drop in oil production caused
by near-well reservoir impairment. In a similar fashion the increase in production can
be represented, provided we have good estimates of the original and the impaired
permeability.

Fig. 8.11—Decreased production due to deteriorated inflow performance resulting from near-well
reservoir impairment (“formation damage”).

8.3.2 Changing the Tubing or Choke Size. Increasing or decreasing the tubing
diameter can improve production. Changing the choke size can also be of influence.
Analysis of choke performance is best done at the tubing head (see, e.g., Fig. 8.12,
which depicts a combination of a tubing performance curve and a set of choke
performance curves for an oil well). Unlike the tubing performance curve for the gas
well depicted in Fig. 8.8, which is monotonically decreasing for an increasing rate, this
tubing performance curve has an increasing branch at the left side. The reason is the
occurrence of slip in the well at low flow rates, which results in an increased liquid
holdup and therefore an increased density of the liquid column. This is exactly the
same effect as was observed for multiphase tubing intake curves in Section 4.4.
Changing out a tubing string is expensive and therefore generally needs to be
justified on a long-term basis (see below). Changing out a choke, however, is a low-
cost operation. As shown in Fig. 8.13, there is a maximum choke size above which
the choke will no longer operate in the critical regime. As explained in Chapter 5,
operation in the critical regime is often preferred because it decouples the upstream
flow behavior from the downstream behavior and thus shields the well from pressure
fluctuations in the production facilities. Associated with the maximum choke size is a
maximum flow rate. Sometimes it is required to bean back a well (i.e., to reduce the
flow rate by installing a smaller choke). A frequently occurring reason for flow
reduction is to prevent or delay water or gas coning or to reduce the amount of gas
or water produced once the cone has reached the well. As can be seen in Fig. 8.13,
there is also a practical minimum choke size and consequently a minimum flow rate,
below which the well should not be operated. Reducing the flow rate even further will
bring the operating point to the left of the maximum of the tubing performance curve.
Although in theory this still results in a stable situation, in practice it is better to avoid
operating to the left of the maximum. This is because the very steep drop of the
tubing performance curve at low flow rates makes it nearly parallel to the choke
performance curves such that small errors in the computation may lead to
dramatically different results. The accuracy of pressure predictions using nodal
analysis is limited (typically between 5 and 20%), and sensitive results are therefore
not very meaningful. For operating points to the right of the maximum the intersection
between the curves occurs at a much larger angle, which makes the corresponding
flow rate and pressure predictions much more robust.
Fig. 8.12—Combined plot of tubing and choke performance curves for an oil well.

Fig. 8.13—Minimum and maximum choke sizes.

8.3.3 Tubing Change-Out. Analysis of the effect of declining reservoir pressure is

best performed downhole using the tubing intake curve. Assuming that the productivity
index (PI) remains constant (no impairment) while the reservoir pressure declines,
then, as shown in Fig. 8.5, the inflow performance curve will move vertically down the
pressure axis with time, keeping the same slope. The corresponding reduction in
production can be calculated. If the reservoir pressure drops too far, another and
often narrower tubing will need to be installed, as illustrated in Fig. 8.14. By plotting
the tubing intake curve for different tubing sizes on the same figure, it will be possible
to see how much a tubing change-out will extend the life of the well and to compute
whether the cost is justified in terms of the extra oil recovered. Another reason that
the 2⅞-in. tubing in Fig. 8.14 is a better choice than the 4-in. tubing is the position of
the operating point with respect to the minimum of the tubing intake curve. Just as for
choke performance curves, it is usually advisable to stay to the right of the minimum,
a condition that is fulfilled only by the operating point of the 2⅞-in. tubing. Small errors
in either the IPR or the pressure intake curve can have a large effect on the position
of the operating point, especially if the two curves are nearly parallel, as is the case
to the left of the minimum. To the right of the minimum the curves intersect at a much
larger angle, thus reducing the sensitivity to errors in the computation.
Fig. 8.14—Example depicting a higher flow rate through a smaller tubing.

8.4 Artificial Lift

8.4.1 Gas Lift. Gas lift is one of the most common methods of artificial lift. By
injecting extra gas downhole into the tubing, the fluid column becomes lighter and the
total production is increased. We have seen in Chapter 4 that there is an optimum
GOR that will minimize the pressure drop over the tubing at a given liquid flow rate
(see Fig. 4.10). Too much gas increases the pressure drop because of friction
effects. We therefore expect that for a producing well there will be an optimum GOR
at which we can inject gas to maximize the oil production rate. This is illustrated in
Fig. 8.15, which shows a set of tubing performance curves for varying GORs. The
solution GOR of the oil is 30 m3/m3 (168 scf/STB), which implies that higher GOR
values result from lift gas injected into the well. If the well should be produced at a
minimum FTHP of 2 MPa (290 psi), indicated with the dotted line, the well cannot flow
naturally and gas lift will be essential to obtain production. Under these conditions
there is indeed an optimum GOR, of approximately 100 m3/m3 (561 scf/STB). See
Question 8.12 for an example of the use of nodal analysis of a gas lifted well.
Fig. 8.15—Effect of GOR on the production from a well.

8.4.2 Electric Submersible Pumps. An ESP is a downhole centrifugal pump

consisting of a stack (up to several hundreds) of centrifugal impellers (rotors) inside
diffusors (stators). The number of stages is directly proportional to the lift capacity of
the pump and is therefore a design parameter. The pump is driven by a variable-
speed or a fixed-speed electrical motor. Like most pumps, ESPs are most efficient if
their throughput is liquid only, although they are capable of handling small amounts of
dense gas bubbles. Preferably an ESP should therefore be set deep enough for its
suction pressure to be above the bubblepoint pressure. If that is not possible, some
form of gas separation can be applied below the pump, whereafter the gas is
transported through the annulus, in parallel with the liquid stream through the pump.
The performance of an ESP is characterized by a pump curve, which is provided
by the manufacturer and typically displays the head capacity, or total (hydraulic)
head, and the efficiency, both as a function of flow rate, for one or more stages (see
Fig. 8.16, which gives an example of a pump curve for a fixed-speed ESP with 100
stages). The head capacity, expressed in m (ft), is defined as

where Δplift is the lift capacity of the pump, pdis and psuc are the discharge and suction
pressures, gl is the liquid gradient, and g is the acceleration of gravity. (Note that the
local oil and water densities ρo and ρw and flow rates qo and qw can be computed with
the aid of Eqs. 2.24 and 2.27.) The head capacity in a pump curve is usually specified
for water at standard conditions. The corresponding lift capacity in terms of pressure
is therefore obtained by multiplying the head by the liquid gradient of the liquid being
pumped. For a centrifugal pump, such as an ESP, the head is not influenced by the
density of the liquid, although Δplift and the power consumption do increase with liquid
density.

Fig. 8.16—Example of a pump curve for a fixed-speed ESP.

As indicated in the pump curve, an ESP has a limited recommended operating

window outside which the efficiency drops considerably. For this particular ESP the
optimal flow rate is approximately –2.8 × 10−3 m3/s (–1,500 B/D). The required lift
capacity as a function of flow rate in a well requiring artificial lift can be displayed in a
lift requirement curve, which is obtained by taking the difference between the IPR
and the tubing intake curve. It should be noted that the pump curve is specified as a
function of the local liquid rate, whereas the IPR and intake curve are usually
specified as functions of the oil rate under standard conditions. To allow direct
comparison with the pump curve, the lift requirement curve should therefore also be
plotted as a function of local liquid rate. The vertical axis could display either Δplift (in
units of pressure) or head (in units of length). As an example, consider the lift
requirement curve in Fig. 8.17 (dashed line), which was obtained by computing the
difference between the IPR for the low reservoir pressure case (pR = 22.5 MPa) and
the tubing intake curve in Fig. 8.5, and displaying the result in the form of Δplift as a
function of local liquid rate -ql. (Note that Fig. 8.5 is expressed in oil rates at standard
conditions, and Fig. 8.17 in liquid rates at local conditions. It can be seen that a zero
lift requirement corresponds to an oil rate qo,sc = –1.7 × 10−3 m3/s in Fig. 8.5 and to a
liquid rate ql = –3.7 × 10−3 m3/s in Fig. 8.17.) Fig. 8.17 also displays a second lift
requirement curve (solid line), which was computed under the assumption that the
reservoir pressure decreased further, to 20 MPa, while at the same time the water
cut increased from fw,sc = 0.2 to fw,sc = 0.4. It can be seen that for the second case,
the required Δplift in Fig. 8.17 at the optimum flow rate for the pump is approximately
3.9 MPa (566 psi). The corresponding liquid gradient can be computed, from
additional data not given in this example, as 8396 Pa/m (0.37 psi/ft). The hydraulic
head at 2.8 × 10−3 m3/s (1,500 B/D) in Fig. 8.16 is approximately 140 m (460 ft), and
because the figure is valid for a pump with 100 stages, the required number of stages
can be computed as

Fig. 8.17—Lift requirement curves for a situation corresponding to Fig. 8.5 (dashed line), and for a further
decreased reservoir pressure and increased water cut (solid line).
The corresponding power requirement of the pump is given by the following equation
(see also Eq. 1.1):

where η is the pump efficiency, which according to Fig. 8.16 is approximately 60% at
–2.8 × 10−3 m3/s (–1,500 B/D). Note that the electric motor that drives the pump has
its own efficiency such that the total electric power required by the ESP is higher than
follows from Eq. 8.20.
For use in a computerized nodal analysis procedure, the manufacturer’s pump
curve from Fig. 8.16 can be converted to a pump performance curve either in the
form of a look-up table or as a (curve-fitted) equation. This is illustrated in Fig. 8.18,
which consists of a combination of the solid lift requirement curve of Fig. 8.17 and the
appropriately scaled pump curve of Fig. 8.16.

Fig. 8.18—Combined plot of rescaled ESP pump and lift requirement curves.

8.5 Related Topics Not Considered in This Chapter

Not considered in this chapter are several other forms of artificial lift, surface
facilities, network topologies with loops, dynamic effects, and temperature effects.
Each topic is briefly explained below.
• Artificial lift. Apart from gas lift and ESPs, which were briefly mentioned earlier,
there are many other artificial lift techniques such as lifting the oil with beam
pumps, hydraulic jet pumps, or progressing cavity pumps. The inclusion of a
pump in the nodal analysis procedure is straightforward once the relevant pump
performance curve is available. Usually such curves are provided by
manufacturers in graphical or tabulated form, in which case they need to be
converted into computer-legible tables to be used in conjunction with nodal
analysis software. Similar relationships are required to model the performance
of compressors for gas lift or gas export. For further information, consult Brown
(1984).
• Surface facilities, such as separators or gas treatment equipment. From a
nodal analysis point of view, separators often form a convenient system
boundary because they are operated at a constant pressure. The detailed
analysis of the flow in separator trains requires the use of compositional
analysis of the well stream and flash calculations to optimize the operating
pressure such that the yield of liquid hydrocarbons is maximized. Such an
analysis is outside the scope of this book.
• Network topologies with loops. These topologies occur in pipeline networks or
complex production systems with, e.g., parallel flowlines. The analysis of
networks requires the derivation of a system of equations in analogy to the
analysis of electrical networks with the aid of Kirchhoff equations. However,
unlike in basic electrical networks, the system elements in the nodal analysis of
oil and gas production networks are usually nonlinear. Therefore their solution
requires an iterative procedure, typically involving multiple solutions of sets of
linearized system equations within a Newton-Raphson procedure. The details of
such a general nonlinear network analysis procedure are outside the scope of
this book. Note that we did treat the iterative solution of systems with simple
branches, in the form of a well producing commingled from two reservoir
layers, in Questions 3.12 and 6.14, and from three layers (requiring
multivariable root finding) in the MATLAB assignment in Section 6.12.
• Dynamic effects. The assumption of steady-state flow which was used in the
derivation of the various elements equations is usually justified for production
engineering calculation, but not always. For example, during pipeline startup or
well testing, pressure transients (i.e., the dynamic response to a change in
operating conditions), are essential. Also self-excited oscillations, such as low-
frequency slugging in offshore pipeline riser systems or heading (i.e., low-
frequency oscillating flow) in low-rate production wells, are dynamic effects that
are outside the scope of this book. For an introduction to dynamic well flow
modeling, see, e.g., Hassan and Kabir (2002) and Shoham (2006). Moreover,
transient reservoir flow (which occurs at much longer time scales than transient
pipe flow) is of importance for well testing and for the production of low-
permeability reservoirs (see also the discussion in Section 7.7).
• Temperature effects. We did take into account the effect of temperature on
 fluid properties, but we did not attempt to compute the temperature distribution
in a production system. Instead, we assumed the temperatures to be known.
Computing the temperatures would require specification of an energy balance in
the formulation of the element equations, in addition to the mass and
momentum balances that we used. It would also require specification of the
thermodynamic properties of the fluids and heat conduction properties of the
various system elements. This would then result in coupled systems of ordinary
differential equations for the system elements in terms of both pressure and
temperature, which could be solved with a method similar to that we used for
pressure only. For a description of such a combined pressure/temperature
approach, see, e.g., Brill and Mukherjee (1999), Hassan and Kabir (2002), and
Shoham (2006).

8.6 Questions

8.1 Nodal analysis concerns the relationship between pressure drop and flow rate
in the various elements of a production system. Why, then, are we plotting
pressure in a single node vs. flow rate (and not pressure drop over an element
vs. flow rate) in a typical nodal analysis plot? What is the underlying
assumption to make this work?
8.2 Why is the relationship between reservoir pressure decline and production
decline usually disproportional? Explain your answer with a picture.
8.3 What causes the occasional appearance of two intersections between
upstream and downstream performance curves? Which of the corresponding
operating points is the physically realistic one? Why is nodal analysis not of
any help in indicating which of the points is physically realistic?
8.4 Considering the intersection of an IPR and a tubing intake curve, what is the
practical rule to avoid unstable flow? Why is this different from the theoretical
stability result?
8.5 Consider Fig. 8.14. Why is the smaller tubing size the better choice? Give two
reasons.
8.6 Consider a vertical oil well producing from the oil zone of a reservoir above
bubblepoint pressure. The wellhead is close to the separator such that the
flowline pressure drop is insignificant. Figs. 8.19 and 8.20 depict two pressure
analysis plots for the well.
(a) What are the corresponding analysis points?
(b) What are the values of reservoir pressure and the separator pressure?
(c) Is the choke operating in the critical or the noncritical regime? Why?
(d) Why are the flow rates in the operating points in Figs. 8.19 and 8.20
identical?
8.7 Consider an oil production system consisting of a slightly inclined (1°) flowline
 1500 m in length, a wellhead choke, a 3000-m-deep vertical well, and a near-
well reservoir represented as a circular drainage area of 500-m radius (see
Fig. 8.21). The pressure in the production system at a certain flow rate is
given in Fig. 8.22.
(a) What are the magnitudes of the reservoir pressure, the bottomhole
pressure, the tubinghead pressure, and the manifold pressure?
(b) Why is the pressure in the flowline decreasing toward the separator?
Why is it decreasing nonlinearly?
(c) Is the choke operating in the critical regime? (Hint: Use a rule of thumb.)
Can you be sure?

Fig. 8.19—Pressure analysis plot 1 for question 8.6.

Fig. 8.20—Pressure analysis plot 2 for question 8.6.

Fig. 8.21—System configuration for question 8.7.

Fig. 8.22—Pressure in the production system at a constant flow rate as a function of distance from the
manifold.

8.8 Consider the tubing performance curve in Fig. 8.23.

(a) Why is the name “tubing performance curve” somewhat misleading?
(b) Explain why the left side of the curve is ascending and the right side
descending.
(c) For a flowline pressure of 1.2 MPa, what are the maximum and the
practical minimum flow rates?
(d) What is the maximum flow rate if we want to maintain choke flow in the
critical regime? (Hint: Use a rule of thumb.)
(e) Answer Question 8.8c again for a flowline pressure of 0.8 MPa. Make a
sketch of the corresponding choke performance curve.
8.9 Consider the nodal analysis plot in Fig. 8.18, which displays an ESP pump
performance curve together with the lift requirement curve. The reservoir
pressure used to generate the lift requirement curve was 20 MPa. Assuming
that the liquid gradient remains unchanged, how much more may the reservoir
pressure drop before the ESP runs out of its recommended operating range?
Sketch the corresponding nodal analysis plot.

Questions 8.10 through 8.12 require the use of MATLAB. You may want to complete the
MATLAB assignment in Section 8.7 first to obtain experience in combining function files
for well performance analysis.
 8.10 Fig. 8.24 displays the system configuration of a horizontal flowline connected
to a deviated well. It is producing from a volatile oil reservoir with a near-
circular drainage area. The relevant parameters have been listed in Table 8.1.
The data have been implemented in MATLAB file system_02.m. In that file, the
tubing head has been chosen as the analysis node.

Fig. 8.23—Tubing performance curve for question 8.8.

Fig. 8.24—System configuration for question 8.10.

Variable Symbol Unit Value

Production data
Water cut fw,sc – 0
Producing GOR Rgo m 3/m 3 450
Manifold pressure Pmf Pa 3 MPa
Reservoir pressure PR Pa 45 MPa
Fluid data
Oil model Volatile oil table 01
Solution GOR at bubblepoint pressure Rsb m 3/m 3 450
Gas density at standard conditions ρg,sc kg/m 3 0.80
Oil density at standard conditions ρo,sc kg/m 3 800
Water density at standard conditions ρw,sc kg/m 3 1080
Multiphase flow correlation Mukherjee and Brill (1985b)
Reservoir data
Reservoir height hR m 30
Permeability k m2 1 × 10 −14
Forchheimer coefficient β 1/m 0
Skin S – 0
Reservoir drainage radius re m 400
Wellbore radius rw m 0.1778
Flow regime Steady-state
Relative permeability data
Endpoint rel. perm. gas – 0.7
Endpoint rel. perm. oil – 0.9
Endpoint rel. perm. water – 0.5
Corey gas ng – 4
Corey oil/gas nog – 4
Corey oil/water now – 4
Corey water nw – 4
Critical gas sat. Sgc – 0.15
Residual oil sat. Sor – 0.10
Immobile wat. sat. Swi – 0.20
Well data
AHD until kick-off point Lint m 3000
Total AHD (from tubing head) Ltot m 4000
Wellbore inclination, top section α top degree 0
Wellbore inclination, bottom section αb ot degree 50
Tubing inside diameter dt m 0.076 00
Tubing roughness et m 30 × 10 −6
Choke data
Choke model Extended Gilbert (1954)
Choke size dch 1/64 in. 28
Flowline data
Flowline length Lfl m 2000
Flowline inclination αfl degree 90
Flowline inside diameter dt m 0.100 53
Flowline roughness et m 30 × 10 −6
Temperature data
Reservoir temperature TR °C 95
FBHT Twf °C 95
Intermediate temperature Tint °C 90
FTHT Ttf °C 85
Flowline entry temperature Tfl °C 80
Manifold temperature Tmf °C 55
Table 8.1—Data for question 8.10.

(a) Run the file, inspect the plot, and determine the oil flow rate. Note that
depending on your computer system specifications, running the file may
take up to a few minutes.
(b) Modify the file and repeat the analysis for different analysis nodes.
Choose the intermediate (kick-off) point, the well bottom, and the
reservoir as the analysis nodes. Generate and plot the corresponding
performance curves, and verify that they all result in the same oil flow
rate.
(c) Could you also have chosen the flowline entry and the manifold as
analysis nodes? Explain your answer. If you are not certain, give it a try.
(d) For the flow rate determined under 8.10a, do the following:
• Plot a traverse of the total pressure in the system as a function of
distance from the manifold.
• Plot a traverse of the pressure resulting from gravity.
• Plot a traverse of the pressure without the gravity contribution—i.e.,
the pressure resulting from friction and (very little) acceleration.
(e) Answer the following questions:
• What are the main causes for the pressure drop over the well?
• The flowline diameter is larger than the tubing diameter.
Nevertheless, the friction loss over the flowline is no smaller than that
over the tubing. Why?
 8.11 (a) Modify MATLAB file system_02.m such that the analysis node is at the
bottom of the well; i.e., plot the IPR vs. the tubing intake curve. Next,
change the production data and fluid properties according to Table 8.2,
and remove the choke from the system. This should result in the
performance plots shown in Fig. 8.25. Does the well flow naturally?
(b) Create a lift requirement curve (at the bottom of the well) in the form of
hydraulic head H as a function of local liquid rate ql.
(c) Using the ESP pump curve in Fig. 8.16, compute the required number of
stages.

Variable Symbol Unit Value

New production data
Water cut fw,sc – 0.60
Producing GOR Rgo m 3/m 3 20
Manifold pressure pmf Pa 0.5 × 106
Reservoir pressure pR Pa 37 × 106
New fluid data
Oil model Standing (1952) black oil correlations
Solution GOR at bubblepoint pressure Rsb m 3/m 3 20
Gas density at standard conditions ρg,sc kg/m 3 0.98
Oil density at standard conditions ρo,sc kg/m 3 865
New reservoir data
Permeability k m2 1.8 × 10 −13
Table 8.2—New data for question 8.11.
Fig. 8.25—Initial performance plots for question 8.11.

8.12 (a) Consider a gas-lifted offshore oil production well with a subsea
wellhead connected to a production facility with a flowline, a gas lift line,
and risers (see Figs. 8.26 and 8.27). The oil flows to the separator
without passing through a choke to minimize the FBHP. The reservoir
pressure , the oil manifold pressure , and the gas compressor
pressure are fixed, and the lift-gas rate is controlled with a choke at
the production platform just downstream of the compressor. A simplified
version of this system has been programmed in MATLAB file
system_03_part_1.m, which models the oil production cascade with lift
gas entering directly at the gas lift valve without considering the lift gas
cascade. Run the file to produce the upstream and downstream
performance curves for an analysis node in the well at the depth of the
gas-lift valve. Repeat the analysis for different lift gas rates and
determine the optimal rate (i.e., the lift gas rate that maximizes the oil
production rate) through trial and error. Plot a set of performance curves
in one figure for three lift-gas rates: below, at and above the optimal rate.
If the maximum amount of available lift gas is restricted to qg,sc,lift = 0.10
m3/s (305 × 103 scf/D), what is the corresponding wellbore pressure at
the depth of the gas lift valve?
(c) Next, inspect and run MATLAB file system_03_part_2.m, which models the
 flow through the gas lift line. It plots the required compressor discharge
pressure as a function of choke diameter for a lift-gas rate qg,sc,lift = 0.10
m3/s (305 × 103 scf/D) and the corresponding wellbore pressure at the
gas lift valve, p4, found under 8.12a. For an available compressor
discharge pressure p7 = 18 MPa, specify the required choke diameter as
a multiple of 1/64 in.

Fig. 8.26—Gas-lifted offshore production well with subsea flowline, gas-lift line, and risers.
Fig. 8.27—System configuration for question 8.12.

8.7 MATLAB Assignment: Well Performance

8.7.1 Objectives

• Learn to use nodal analysis to determine well performance

• Obtain insight into the effect of tubing size on well performance at different
stages of the reservoir life
• Obtain insight into the effect of choke size on well performance

8.7.2 Assignment, Part 1. Consider a multiphase well using the parameters given in
Table 8.3. Assume that the produced gas is only associated gas.

Task 1
• Inspect the MATLAB files example_intake_curve.m, pipe.m and Muk_Brill
dpds.m.
• Copy example_intake_curve.m under a new name and use the new file to
create a pressure intake curve for the well from Table 8.3 for 0 < qo,sc < 3 ×
10−3 m3/s. Choose parameter fluid equal to 4 (Mukherjee and Brill correlation)
and oil equal to 1 (Standing correlations).
 • Expand the script file and create a plot with five intake curves for the tubing
diameters given in Table 8.4. All other parameters remain the same.
• Compute the bubblepoint pressure at reservoir temperature. Hint: Use
pres_bub_Standing.m.

Variable Symbol Unit Value

Wellbore inclination α rad 0
Tubing inside diameter d m 0.062
Tubing roughness e m 30 × 10−6
Water cut fw,sc – 0.0
Flowline pressure pfl Pa 0.80 × 106
Flowing tubinghead pressure ptf Pa 2.00 × 106
Producing GOR Rgo m 3/m 3 50
Gas density at standard conditions ρg,sc kg/m 3 0.95
Oil density at standard conditions ρo,sc kg/m 3 850
Water density at standard conditions ρw,sc kg/m 3 1050
Bottomhole temperature Twf °C 120
Tubinghead temperature Ttf °C 30
Total true vertical well depth Ztot m 3000
Table 8.3—Well data.

OD (in.) OD (mm) Nominal Wt (lbm/ft) ID (mm)

2.375 60.3 4.70 50.67
2.875 73.0 6.50 62.00
3.5 88.9 9.30 76.00
4.0 101.6 11.00 88.29
4.5 114.3 12.75 100.53
Table 8.4—Tubing data for 8.7.2 Assignment, Task 1.

• Inspect the files example_IPR.m and res_oil_simp.m.

• Further expand the script file and plot the IPR for steady-state flow and the
reservoir parameters given in Table 8.5. Note: Use the effective permeability
for oil.

Question 1
1(a) Is it justified to use the IPR for single-phase oil flow? That is, is it justified
to use res_oil_simp.m, or should you have used res.m?
1(b) What is the PI of the well?
1(c) What is the maximum flow rate of the well at the given FTHP?
 1(d) What is the corresponding tubing size?
1(e) How high can the oil rate become if the FTHP is reduced to 10 bar?
1(f) Which factors could determine the lower limit for the FTHP?

8.7.3 Assignment, Part 2. Consider the same multiphase well with the parameters
given in Table 8.3—this time, however, with an increased water cut.

Variable Symbol Unit Value

Forchheimer coefficient ß 1/m 82 × 106
Permeability k m2 4 × 10−14
Endpoint relative permeability gas – 0.7
Endpoint relative permeability oil – 0.9
Endpoint relative permeability water – 0.5
Corey exponent for gas ng – 3
Corey exponent for oil in gas/oil flow nog – 3
Corey exponent for oil in gas/water flow now – 3
Corey exponent for water nw – 3
Critical gas saturation Sgc – 0.05
Residual oil saturation Sor – 0.10
Immobile water saturation Swi – 0.15

Average net pay thickness h m 20

Radius external drainage boundary re m 800
Wellbore radius rw m 0.177 8
Reservoir pressure pr Pa 26 × 106
Skin S – 0
Table 8.5—Reservoir data.

Task 2
• Inspect the files example_IPR.m, res.m and res_dpdr.m.
• Make plots for fw,sc = 0.1 and fw,sc = 0.3. Use an FTHP of 10 bar and include the
intake curves for all five tubing sizes.
• Select a tubing for the case in which fw,sc = 0.1 and plot the corresponding
traverse.

Question 2
2(a) What are the highest possible flow rates and the corresponding tubing
sizes for fw,sc = 0.1 and fw,sc = 0.3?
2(b) Why do the curves in the traverse change shape near the surface?
8.7.4 Assignment, Part 3. Consider the same multiphase well with the parameters
given in Table 8.3.

Task 3
• Write a MATLAB script file to plot the tubing performance. Make use of the
functions res_oil_simp.m and pipe.m in series. Plot the curve for fwsc = 0 and a
2.875-in. tubing. Check if the flow rate at an FTHP of 20 bar corresponds to the
flow rate found in Part 1 of the exercise.
• Expand the script file and create a plot with the tubing performance curve and
five choke performance curves for choke diameters of 32/64-in., 48/64-in.,
64/64-in., 80/64-in., and 96/64-in. Use function choke_multi_phase_simp.m, and
choose the Gilbert choke model.

Question 3
3(a) Which choke diameter results in an FTHP closest to 10 bar?
3(b) What is (approximately) the maximum flowline pressure that will still result
in critical flow through that choke?

8.7.5 Deliverables (for All Parts)

• MATLAB program listing

• Answers to the questions
• Plots
Appendix A

SI Units and Field Units

A-1 Conversion Factors

To obtain SI units, multiply a quantity given in field units with the conversion factor
specified in Table A-1.

A-2 SI Prefixes
Table A-2 shows important SI prefixes. Note that out of the seven strict base SI units
(m, kg, A, K, mol, and cd), there is one (kg) that has a prefix included in its name.
However, the prefixes for mass are based on the gram (g). Therefore, e.g., one-
millionth of a kilogram is not a micro kilogram (μkg) but a milligram (mg). This is an
unfortunate imperfection in the SI system.

A-3 Standard Conditions

• SI units: 100 kPa and 15°C. Note that sometimes a standard pressure of
101.325 kPa is used (see SPE 1982); the difference is negligible for normal
production engineering purposes.
• Field units: 14.7 psia and 60°F.
• Density of air at standard conditions: ρair,sc = 1.23 kg/m3 (76.3 × 10–3 lbm/ft3).
• Density of water at standard conditions: ρwater,sc = 999 kg/m3 (62.4 lbm/ft3 or
8.34 lbm/gal).

A-4 Pitfalls in Unit Conversions

A-4.1 Force, Mass, and Acceleration of Gravity. The relationship between force
and mass is given by Newton’s law as “force equals mass times acceleration.” This
can be expressed in SI units as

where F is force, expressed in N; m is mass, expressed in kg; x is distance,

expressed in m; and t is time, expressed in seconds. This implies that N = kg · m/s2.
As a result, a mass of 1 kg experiences an attractive force because of the Earth’s
gravitational field, which has a magnitude g = 9.80665 m/s2, of
Physical Quantity Dimension SI Units 1) Field Units Conversion Factor From Field Units to SI Units 2)
Area [L2] m2 ft2 9.290 304 × 10–2 (exact)
m2 in.2 6.451 6 × 10–4 (exact)
m2 acre 4.046 856 × 103
Compressibility [L m –1 t2] 1/Pa 1/psi 1.450 377 × 10–4
Density [L–3 m] kg/m 3 °API 141.5 × 103/(131.5 + °API) (exact)
kg/m 3 lbm/ft3 1.601 846 × 101
kg/m 3 lbm/gal 1.198 264 × 102
Energy [L m 2 t–2] J cal 4.184 (exact)
Flow rate [L3 t–1] m 3/s B/D 1.840 131 × 10–6
m 3/s scf/D 3.277 413 × 10–7
m 3/s gal/min 6.309 020 × 10–5
m 3/d B/D 1.589 873 × 10–1
Force [L m t–2] N lbf 4.448 222
Gas/oil ratio (GOR) m 3/m 3 scf/STB 1.781 076 × 10–1
Length [L] m ft 3.048 × 10–1 (exact)
[L] m in. 2.54 × 10–2 (exact)
Mass [m] kg lbm 4.535 924 × 10–1
Permeability [L2] m2 mD 9.869 233 × 10–16
Power 3) [L m 2 t–3] W hp 7.456 999 × 102
Pressure 4) [L–1 m t–2] Pa psi 6.894 757 × 103
Pressure gradient [L–2 m t–2] Pa/m psi/ft 2.262 059 × 104
Productivity index (PI) [L4 m –1 t] m 3/s · Pa B/D-psi 2.668 884 × 10–10
m 3/d · Pa B/D-psi 2.305 916 × 10–5
Specific PI [L3 m –1 t] m 2/s · Pa B/D-psi-ft 8.756 182 × 10–10
m 2/d · Pa B/D-psi-ft 7.565 341 × 10–5
Surface tension [m t–2] N/m Dyne/cm 1 × 10–3 (exact)
Temperature 5) [T] K °R 5/ (exact)
9

°C °F (°F − 32) / 1.8 (exact)

Velocity [L t–1] m/s ft/s 3.048 × 10–1 (exact)
Viscosity (dynamic) –1 –1
[L m t ] Pa · s cp 1.0 × 10–3 (exact)
Volume [L3] m3 ft3 2.831 685 × 10–2
m3 bbl 1.589 873 × 10–1
m3 gal 3.785 412 × 10–3
1) The expression “SI units” is used loosely to indicate both “strict” SI units and “allowable” units. The strict units
can be subdivided in the seven base SI units (m, kg, s, A, K, mol, and cd) and the derived SI such as N, Pa, °C,
or J. The allowable SI units are those defined in SPE (1982) and include d (day) and a (year).
2) Conversion factors have been taken from SPE (1982).

3) One hp = 550 ft-lbf/s.

4) Pressure in field units can be expressed in psig (gauge pressure) or psia (absolute pressure), where 0 psig =

14.7 psia. Pressure in SI units is often expressed in bars, which is an allowable SI unit, where 1 bar = 105 Pa.
5) Zero K (Kelvin) is absolute zero in Celsius units. Therefore, the temperature expressed in K equals the
temperature expressed in °C + 273.15.
Zero °R (Rankine) is absolute zero in Fahrenheit units. Therefore, the temperature expressed in °R equals the
temperature expressed in °F + 459.67.
Table A-1—Conversion factors to convert field units to SI units.

Symbol Name Magnitude

n nano 109
μ micro 10–6
m mili 10–3
c centi 10–2
d deci 10–1
da deca 101
h hector 102
k kilo 103
M mega 106
G giga 109
Table A-2—SI prefixes.

Field units, however, have been defined purposely such that a mass of 1 lbm
experiences an attractive force from the Earth’s gravitational field, which has a
magnitude g = 32.174 ft/s2, of exactly 1 lbf:

This simple result in field units for a mass experiencing the acceleration of gravity
leads, however, to a more complicated expression for a mass experiencing an
arbitrary acceleration. In that case we should write

where F is force, lbf; m is mass, lbm; x is distance, ft; t is time, s; and gc is a

dimensionless constant with magnitude 32.174. This implies that 1 lbf = 1/gc lbm·ft/s2.
Note that the standard acceleration of gravity is specified as 9.80665 m/s2 (32.174
ft/s2). In reality, the acceleration of gravity will show slight variations with geographical
location and altitude.
A-4.2 Amount of Substance, Molar Mass, and Molecular Weight. In SI units we
express the amount of substance in mol, defined as “the amount of substance of a
system that contains as many elementary entities as there are in 12 g of C12.” The
molar mass, is specified in kg mol–1. However, for historical reasons one mostly uses
the molecular weight, which is specified in g mol–1. We will indicate the molar mass
with the symbol M and the molecular weight with Mw. An amount of n mol of
substance with a molar mass M expressed in kg mol–1 has a mass of n M kg. An
amount of n mol of substance with a molecular weight Mw expressed in g mol–1 has a
mass of n Mw g.
In field units, the amount of substance is expressed in lbm mol, which is “the
amount of substance of a system that contains as many elementary entities as there
are in 12 lbm of C12.” The molecular weight is expressed, accordingly, in lbm (lbm
mol) –1. It has the same magnitude as when expressed in SI units, and we also
indicate it with the symbol Mw. An amount of n lbm mol of substance with a molecular
weight Mw expressed in lbm (lbm mol)–1 has a mass of n Mw lbm. For gases, the
molar mass M is related to the specific gravity γg and the density under standard
conditions ρg,sc according to

A-4.3 Dimensional Constants. The results of experimental data are often presented
as correlations with the aid of curve fitting. If the input data and the output data have
different physical dimensions, the coefficients of the fitted curve also have physical
dimensions. This implies that if you use these expressions in a different unit system,
the “constant” coefficients have a different value. To convert such a correlation from
field units to SI units, start from the expression in its original form (i.e., as expressed
in field units) and

1. Insert the variables in SI units divided by their corresponding field-to-SI

conversion factors as given in Table A-1.
2. Solve for the desired left side while combining all numerical factors.

Usually Step 1 is straightforward. Thus, e.g., to convert a pressure variable p

expressed in psia to one expressed in Pa, we simply insert
Sometimes Step 1 is slightly more complex. In that case, e.g., to convert a
temperature variable T expressed in °F to °C, we need to insert

As an example of converting a correlation with dimensional constants, consider the

Standing correlation for bubblepoint pressure expressed in field units:

where pb is expressed in psi, Rp is expressed in scf/STB, γg is dimensionless, T is

expressed in °F, and γAPI in °API. Following Step 1 as outlined above results in

Next, solving for pb and combining all numerical factors leads to

where we used the equality 7160.83 = (1.23/1.781076 × 10–1)0.83 exp(0.00091 × 32 +

0.0125 × 131.5). The result in SI units is identical to Eq. B-1 in Appendix B.
Appendix B

Fluid Properties and Correlations

B-1 Fluid Properties

Properties in Table B-1 (in SI units) have been computed from the values in Table B-
2 with the aid of the conversion factors presented in Appendix A, which were, in turn,
taken from SPE (1982). Properties in Table B-2 (in field units) have been taken from
GPSA (1998). Some typical fluid gradients have been listed in Table B-3.

Compound Molar Mass M (10–3 Critical Pres. pc Critical Temp. Gas Dens. ρg,sc1) Liquid Dens. ρl,sc1)
kg/mol) (106 Pa) Tc,abs (K) (kg/m 3) (kg/m 3)
N2 (nitrogen) 28.01 3.40 126.2 1.18 8092)
CO2 (carbon 44.01 7.37 304.1 1.86 8173)
dioxide)
H2S (hydrogen 34.08 8.96 373.4 1.44 8013)
sulphide)
H2O (water) 18.02 22.1 647.1 0.77 999
C1H4 (methane) 16.04 4.60 190.6 0.68 3004)
C2H6 (ethane) 30.07 4.88 305.4 1.27 3563)
C3H8 (propane) 44.10 4.24 369.8 1.86 5073)
C4H10 (iso- 58.12 3.64 407.8 2.45 5623)
butane)
C4H10 (n-butane) 58.12 3.78 425.1 2.45 5843)
C5H12 (iso- 72.15 3.38 460.4 3.05 624
pentane)
C5H12 (n- 72.15 3.37 469.7 3.05 631
pentane)
C6H14 (n-hexane) 86.18 3.03 506.4 3.64 663
C7H16 (n- 100.20 2.74 539.2 4.23 687
heptane)
C8H18 (n-octane) 114.23 2.49 568.4 4.82 706
C9H20 (n-nonane) 128.26 2.28 594.7 5.42 721
C10H22 (n- 142.29 2.10 617.7 6.00 734
decane)
1) At standard conditions: 100 kPa and 15°C = 288 K.

2) Density at the normal boiling point, i.e., at boiling temperature (78 K) and 100 kPa. The temperature at standard
conditions (288 K) is above the critical temperature (126 K).
3) Density at saturation pressure (bubblepoint pressure) and 288 K.
4) Estimated value. The temperature at standard conditions (288 K) is above the critical temperature (191 K).

Table B-1—Reservoir fluid properties in SI units.

Compound Molar Mass M Critical Pres. Critical Temp. Gas Specific Liquid Specific
(lbm/lbm · mole) pc (psia) Tc (°F) Gravity1,2) γ g (−) Gravity1,3) γ l (−)

N2 (nitrogen) 28.01 492.8 −232.49 0.967 2 0.809 404)

CO2 (carbon 44.01 1069.5 87.73 1.519 6 0.818 015)
dioxide)
H2S (hydrogen 34.08 1300. 212.40 1.176 7 0.801 435)
sulphide)
H2O (water) 18.02 3200.1 705.11 0.622 0 1.000 00
C1H4 (methane) 16.04 667.0 −116.66 0.553 9 (0.3)6)
C2H6 (ethane) 30.07 707.8 90.07 1.038 2 0.356 195)
C3H8 (propane) 44.10 615.0 205.92 1.522 6 0.506 985)
C4H10 (iso- 58.12 527.9 274.41 2.006 8 0.562 865)
butane)
C4H10 (n-butane) 58.12 548.8 305.51 2.006 8 0.584 025)
C5H12 (iso- 72.15 490.4 368.96 2.491 2 0.624 41
pentane)
C5H12 (n- 72.15 488.1 385.7 2.491 2 0.631 08
pentane)
C6H14 (n- 86.18 439.5 451.8 2.975 5 0.664 04
hexane)
C7H16 (n- 100.20 397.4 510.9 3.459 8 0.688 05
heptane)
C8H18 (n-octane) 114.23 361.1 563.5 3.944 1 0.706 78
C9H20 (n- 128.26 330.7 610.8 4.428 4 0.721 93
nonane)
C10H22 (n- 142.29 304.6 652.2 4.912 7 0.734 17
decane)
1) At standard conditions: 14.7 psia and 60°F = 520°R.

2) With respect to air, which has a density at standard conditions of 0.0768 lbm/ft3.

3) With respect to fresh water, which has a density at standard conditions of 62.4 lbm/ft3.

4) Density at the normal boiling point, i.e., at boiling temperature (−320°F) and 14.7 psia. The temperature at
standard conditions (60°F) is above the critical temperature (−232°F).
5) Density at saturation pressure (bubblepoint pressure) and 60°F.

6) Estimated value. The temperature at standard conditions (60°F) is above the critical temperature (−117°F).

Table B-2—Reservoir fluid properties in field units.

 Fluid Density (kg/m 3) Gradient (kPa/m) Gradient (psi/ft)
Gas 100–300 1.0–2.9 0.04–0.13
Oil 800–900 7.8–8.8 0.35–0.39
Water 1000–1100 9.8–10.8 0.43–0.48
Table B-3—Typical reservoir fluid gradients.

B-2 Oil Correlations

B-2.1 Black Oil Correlations. Black oil correlations are based on laboratory tests,
most of which were performed on crudes with a gas/oil ratio (GOR) less than 350
m3/m3 (near 2,000 scf/STB). In practice, black oil correlations are often used above
this limit, but with caution. Note that the numerical values in several of these
correlations are not dimensionless. The most widely known black oil correlations are
the Standing correlations, originally issued in Standing (1947) and Standing (1952).
The expressions below have been taken from Appendix II of the 1977 SPE reissue of
Standing (1952). They have been derived on the basis of data from 22 Californian
crudes under conditions listed in Table B-4. (Note that all correlations in this appendix
have been implemented in several MATLAB routines, found in Appendix H.)

B-2.2 Bubblepoint Pressure pb. Standing’s correlation to relate the bubblepoint

pressure pb to the bubblepoint gas/oil ratio Rsb of a black oil is

If the producing GOR Rgo of a well is used to compute pb with this correlation, the
results are valid only if Rgo = Rsb (i.e., if the reservoir pressure is at or above the
bubblepoint pressure).

B-2.3 Solution Gas/Oil Ratio Rs. Once we know pb , and therefore also Rsb , we can
estimate Rs, the solution gas/oil ratio at a pressure p above or below pb . If p > pb , the
oil is undersaturated, all gas is in solution, and Rs = Rsb . If p ≤ pb , the oil is saturated
with gas, there is free gas, and the pressure p must be the bubblepoint pressure of
the mixture of oil and still-dissolved gas. Hence Rs is given by the inverse of the above
Standing correlation, but with pressure p instead of pb :

B-2.4 Oil Formation Volume Factor Bo. As the pressure changes, the volume it
occupies changes as a result of two effects: compressibility effects and, much more
importantly, changes in the amount of dissolved gas. If p ≤ pb , the oil is saturated and
the mixture of oil and still-dissolved gas is at its bubblepoint. We can use Standing’s
correlation for the oil formation volume factor at the bubblepoint pressure:

Property SI Units Field Units

Bubblepoint pressure, pb 0.9–48.3 × 106 Pa 130–7,000 psia
Temperature, T 37–125°C 100–258°F
Solution GOR, Rs 3.5–254 m 3/m 3 20–1,425 scf/STB
Oil density, ρo,sc, and gravity, γ API 725–956 kg/m 3 16.5–63.8 °API
Gas density, ρg,sc, and gravity, γ g 0.73–1.17 kg/m 3 0.59–0.95 (air = 1.00)
Table B-4—Conditions used to derive Standing correlations.

If p > pb , the oil is undersaturated and all the gas is dissolved. As pressure changes,
all changes result from changes in density of the mixture; no extra gas is dissolved or
comes out of solution. Hence

where Bob and ρob are the oil formation volume factor and the density of the oil at the
bubblepoint pressure pb respectively, and ρo is the density at pressure p. As was
shown in Section 2.7.2, this can also be written as

where co is the isothermal compressibility of the undersaturated oil. A correlation for

co is given in Section B-2.6.

B-2.5 Densities. Using the principle of conservation of mass, we can derive that

The oil density at bubblepoint conditions, ρob , can be obtained from Eq. B-6 through
substitution of Bo = Bob . At other pressures, above pb , the density is given by
(see Eq. 2.18 in Section 2.5.3). Substitution of the Eq. B-7 relationship in Eq. B-4
recovers the Eq. B-5 relationship for Bo quoted above.

B-2.6 Compressibility. Oil compressibility is best measured in the laboratory if

accurate values are required for the density or Bo. Correlations do exist for the
compressibility. One of these is given by Vazquez and Beggs (1980) and has the form

where ρg,100 is the gas density measured at a pressure of 689 kPa (100 psig = 114.7
psia). This pressure was chosen to reflect a typical separator pressure because
usually the gas density is determined from a sample taken from a separator (see
Vazquez and Beggs 1980). The relationship between ρg,100 and ρg,sep measured at
any other separator pressure psep and temperature Tsep is given by

B-2.7 Viscosity. A commonly used empirical correlation for dead oil is the one from
Beggs and Robinson (1975). It can be expressed as

A correlation for saturated oil viscosity is also given by Beggs and Robinson (1975)
as

Vazquez and Beggs (1980) determined the following empirical correlation for under-
saturated oil:

These expressions illustrate that the oil viscosity increases with increasing oil density
and with increasing pressure (for undersaturated oil), and it decreases with increasing
temperature and increasing gas saturation (for saturated oil) (see Figs. B-1 and B-
2). The physical explanation is, simply said, that the heavier hydrocarbon components
(i.e., the ones with high carbon numbers) consist of long molecules that have a
tendency to get entangled. At increasing temperatures these long molecules tend to
stretch and untangle, whereas at increasing pressures they are pushed together.
Moreover, the addition of short gas molecules results in a molecular mixture of lower
viscosity. Table B-5 gives an overview of the conditions under which these
correlations were derived (see Beggs and Robinson 1975 and Brill and Mukherjee
1999).

B-2.8. Example B-1: Oil Formation Volume Factor (p < pb). Consider a well that
produces oil and gas at the following rates: qo,sc = 1000 m3/d, and qg,sc = 200 000
m3/d. The production history shows no indication of gas-cap gas production. The
density of the oil and gas at standard conditions are given by ρo,sc = 800 kg/m3 and
ρg,sc = 0.98 kg/m3, and the reservoir is at pressure and temperature given by pR =
20.00 MPa and TR = 150°C.

Question
What is the oil formation volume factor Bo at reservoir pressure and temperature?

Answer
The bubblepoint GOR is given by the producing GOR as
Fig. B-1—Dead oil viscosity as a function of oil density for different temperatures. Graph based on the
Beggs and Robinson (1975) correlation.
Fig. B-2—Oil viscosity as a function of pressure for two different solution gas/oil ratios and four different
temperatures: ρg,sc = 0.95 kg/m3 (γg = 0.77), ρo,sc = 850 kg/m3 (γAPI = 35°). Graph based on the Beggs and
Robinson (1975) and Vazquez and Beggs (1980) correlations.

Property Eq. B-11 Eq. B-11 Eq. B-12 Eq. B-12

SI Units Field Units SI Units Field Units

Temperature, T 21–146°C 70–295°F

Pressure, pb 0.1–36.3 MPa 15–5,265 psia 1.0–65.6 MPa 141–9,515 psia
Solution GOR, Rs 3.6–369 m 3/m 3 20–2,070 scf/STB 16–392 m 3/m 3 90–2,199 scf/STB
Density, ρo,sc, γ API 959–747 kg/m 3 16–58°API 966–739 kg/m 3 15–60°API
Density, ρg,sc, γ g 0.63–1.66 kg/m 3 0.51–1.35
Viscosity, μ0 0.12–148 mPa . s 0.12–148 cp

Table B-5—Conditions used to derive viscosity correlations.

With the aid of Eq. B-1 we find that the bubblepoint pressure equals
where T has been taken equal to the reservoir temperature. Because the reservoir
pressure is below the bubblepoint pressure, we need to compute the solution GOR
Rs at reservoir pressure with the aid of Eq. B-2:

The oil formation volume factor now follows from Eq. B-3 as

Answer With MATLAB

» R_sb = 200000/1000
R_sb = 200
» p_b = pres_bub_Standing(R_sb,0.98,800,150)
p_b = 2.6105e+007
» R_s = gas_oil_rat_Standing(20e6,0.98,800,150)
R_s = 145.4194
» B_o = oil_form_vol_fact_Standing(R_s,0.98,800,150)
B_o = 1.5656

Alternatively, the black oil properties can be computed directly as

» [B_g,B_o,R_s] = black_oil_Standing
(20e6,200,0.98,800,150)
B_g = 0.0068
B_o = 1.5656
R_s = 145.4194

B-2.9 Example B-2: Oil Formation Volume Factor (p > pb). Consider the same
situation as in Example B-1 in Section B-2.8, except for the reservoir pressure, which
is now given by pR = 40.00 MPa.

Question
What is the oil formation volume factor Bo at this higher reservoir pressure?

Answer
Because the reservoir pressure is now above the bubblepoint pressure, the solution
GOR is equal to the bubblepoint GOR:
 Rs = Rsb 200 = m3/m3.

The corresponding oil formation volume factor Bob is obtained from Eq. B-3 as

We can now compute the oil formation volume factor from Eq. B-5. This requires that
we first determine the compressibility co from Eq. B-8, which, in turn, requires
computation of ρg100 from ρg,sep with the aid of Eq. B-9. Because ρg,sep in our example
is equal to ρg,sc, we can enter standard conditions in Eq. B-9. This leads to

Answer With MATLAB. Note that this answer is continued from the previous example.
» B_ob = oil_form_vol_fact_Standing(R_sb, 0.98, 800, 150)
B_ob = 1.7515
» rho_g_100 = rho_g_Vazquez_and_Beggs(100e3, 0.98, 800, 15)
rho_g_100 = 0.8407
» c_o = compres_Vazquez_and_Beggs(40e6,R_
sb,rho_g_100,800,150)
c_o = 2.2732e-009
» B_o = oil_form_vol_fact_undersat(B_ob,c_o,40e6,p_b)
B_o = 1.6970

Alternatively, the black oil properties can be computed directly as

» [B_g,B_o,R_s] = black_oil_Standing
(40e6,200,0.98,800,150)
B_g = 0
B_o = 1.6970
R_s = 200

B-2.10 Example B-3: Oil Viscosity. Consider the same situation as in Example B-2
in Section B-2.9 above.
Question
What is the oil viscosity μo at reservoir pressure and temperature?

Answer
With the aid of Eq. B-10 the dead oil viscosity follows as

The oil viscosity at bubblepoint can then be computed with the aid of Eq. B-11 as

and the viscosity at reservoir pressure with Eq. B-12 as

Answer With MATLAB

» mu_od = oil_visc_dead_B_and_R(800,150)
mu_od = 4.7851e-04
» mu_ob = oil_visc_sat_B_and_R (mu_od, 200)
mu_ob = 1.9252e-04
» mu_o = oil_visc_undersat_V_and_B(mu_ob,40e6,26.1e6)
mu_o = 2.3350e-04

Alternatively, the oil viscosity can be computed directly as

» mu_o = oil_viscosity(40e6,200,0.98,800,150)
mu_o = 2.3348e-04

B-3 Gas Correlations

B-3.1 Pseudoproperties. The concepts of critical pressure and temperature are

exactly defined for single components. However, for mixtures the concepts are
approximations, as indicated by the use of the terms pseudocritical pressure and
temperature. We can use the Sutton (1985) correlations to estimate the
pseudocritical properties as function of the gas density:

Note that the pseudocritical temperature is expressed in K. If a compositional

description of the mixture is available, the pseudocritical properties can be
determined more accurately with the aid of mixing rules. (For further information,
consult the references mentioned in Section 2.2.) The dimensionless pseudoreduced
pressure ppr and pseudoreduced temperature Tpr are defined as

B-3.2 Density. For single-phase gas flow the gas density follows directly from the
nonideal gas law as

In the black oil model, where it is assumed that the gas composition does not change
with pressure and temperature, the same expression can be used for the gas density
in the two-phase region.

B-3.3 Viscosity. A well-known correlation for gas viscosity was presented in

graphical form by Carr et al. (1954). A numerical approximation of this correlation
was given by Dempsey (1965) and can be represented in two steps as

where

with dimensionless coefficients as given in Table B-6. The variable is the

viscosity at atmospheric pressure for which the correlation can be expressed as
with dimensional coefficients that are also given in Table B-6. The relationships
between the molar mass M, the molecular weight Mw, the gas density ρg,sc, and the
gas specific gravity γg are given in Eqs. A-5 and A-6. Correlations B-18 through B-20
are valid for the following ranges of parameter values:

Coefficient Value Coefficient Value

a0 –2.462 118 20 × 10–00 a13 –1.864 088 48 × 10–01
a1 2.970 547 14 × 10–00 a14 2.033 678 81 × 10–02
a2 –2.862 640 54 × 10–01 a15 –6.095 792 63 × 10–04
a3 8.054 205 22 × 10–03
a4 2.808 609 49 × 10–00 b0 1.112 319 13 × 10–05
a5 –3.498 033 05 × 10–00 b1 3.019 078 87 × 10–08
a6 3.603 730 20 × 10–01 b2 6.848 080 07 × 10–12
a7 –1.044 324 13 × 10–02 b3 –1.094 850 50 × 10–04
a8 –7.933 856 84 × 10–01 b4 –1.152 569 51 × 10–07
a9 1.396 433 06 × 10–00 b5 –2.913 973 49 × 10–10
a10 –1.491 449 25 × 10–01 b6 4.577 351 89 × 10–04
a11 4.410 155 12 × 10–03 b7 3.832 261 02 × 10–07
a12 8.393 871 78 × 10–02 b8 1.288 652 49 × 10–09
Table B-6—Coefficients for the Dempsey (1965) approximation of the Carr et al. (1954) gas viscosity
correlation.

The Dempsey (1965) approximations have been programmed in MATLAB routines

gas_visc_atm_Dempsey.m and gas_visc_ratio_Dempsey.m, with which Figs. B-3 and
B-4 were produced. The graphs are slightly different from the original graphs in Carr
et al. (1954), but are accurate enough for most production engineering calculations. In
case of the presence of nonhydrocarbon components in the gas, correction factors
are needed for the computation of the viscosity at atmospheric pressure (for further
information see the original publication of Carr et al. 1954).
Fig. B-4 illustrates that the viscosity of hydrocarbon gases, unlike the viscosity of
hydrocarbon liquids, increases with increasing temperature. This is because the
increased molecular motion of the small gas molecules at higher temperatures
reduces the freedom for molecules to move around. In hydrocarbon liquids this effect
is counteracted by the increased freedom of the larger molecules resulting from
stretching and disentanglement at higher temperatures.
B-3.4 Z Factor. Standing-Katz Correlation. The widely accepted correlation for the
gas compressibility factor (Z factor) for a nonideal gas or gas mixture was presented
by Standing and Katz (1942) in graphical form. Various numerical approximations
have been developed over time, and an overview is presented in Takacs (1976).
Below we reproduce the approximation by Dranchuk and Abu-Kasem (1975), which is
given in the form of an implicit function in terms of Z:

Fig. B-3—Gas viscosity at atmospheric pressure. Graph based on the Dempsey (1965) approximation of
the Carr et al. (1954) correlation.
Fig. B-4—Gas viscosity ratio. Graph based on the Dempsey (1965) approximation of the Carr et al. (1954)
correlation.

where

Here, a1 through a11 are coefficients as specified in Table B-7, and c is a function of
the dimensionless pseudoreduced pressure and temperature:

Numerical Implementation. Because Eq. B-22 is implicit in Z, it needs to be

solved iteratively—e.g., with a Newton Raphson algorithm (see Appendix D), which
we can write as

Here k is the iteration counter, and f'(Z) is the derivative of f(Z) with respect to Z
given by
 Coefficient Value
a1 0.326 5
a2 –1.070 0
a3 –0.533 9
a4 0.015 69
a5 –0.051 65
a6 0.547 5
a7 –0.736 1
a8 0.184 4
a9 0.105 6
a10 0.613 4
a11 0.721 0
Table B-7—Coefficients for Dranchuk and Abu-Kasem (1975) approximation.

A convenient starting value for Z is provided by the explicit approximation of Papay

quoted in Takacs (1976) as

Eq. B-25 can then be used to obtain improved approximations to the desired
accuracy. Alternatively, we can solve Eq. B-22 with the aid of the standard MATLAB
function fzero (see Appendix D). The Dranchuk and Abu-Kasem approximation to the
Standing and Katz correlation has been programmed in MATLAB functions
Z_factor_DAK.m and Z_factor_DAK_direct.m. (Note that these functions take
different arguments: Z_factor_DAK.m requires ppr and Tpr as input, whereas
Z_factor_DAK_direct.m requires, p, ρg,sc, and Tabs .) The range of validity for the
approximation is

which is sufficient for most applications. Figs. B-5.a and B-5.b have been generated
with the MATLAB routine, and they closely mimic the original graphical Z-factor chart as
presented in Standing and Katz (1942).

B-3.5 Example B-4: Gas Properties. Consider the same situation as in Example B-1
in Section B-2.8.
Question
What are the gas formation volume factor Bg and the gas viscosity μg at reservoir
pressure and temperature?

Answer
With the aid of the Sutton correlations (Eqs. B-13 and B-14) we find the
pseudocritical properties of the fluid as

The pseudoreduced pressure and temperature at the reservoir follow as

With the aid of the Standing-Katz chart in Fig. B-5, we find for the compressibility
factor, Z = 0.93, and with the aid of Eq. 2.34 for the gas formation volume factor,

The molar mass of the gas follows from Eq. A-5 as

M = 23.55 × 10–3 × 0.98 = 23.08 × 10–3 kg/mol.

Figs. B-5a and B-5b—Compressibility factor Z as function of pseudoreduced pressure ppr and
pseudoreduced temperature Tpr. The graph is based on the Dranchuk and Abu Kasem (1975)
approximation of the Standing and Katz (1942) correlation.
With the aid of Figs. B-3 and B-4 we now find for the gas viscosity at atmospheric
pressure, Pa·s, and for the viscosity ratio, f = 1.4. The viscosity at
reservoir pressure then follows with Eq. B-18 as

μg = 1.4 × 1.3 × 10–5 = 1.8 × 10–5 Pa·s.

Answers With MATLAB. The gas formation volume factor is computed by

» p_pc = pres_pseu_crit_Sutton(0.98)
p_pc = 4.4829e+006
» T_pc_abs = temp_pseu_crit_Sutton(0.98)
T_pc_abs = 222.6191
» p_R_pr = 20e6 / p_pc
p_R_pr = 4.4614
» T_R_pr = (150 + 273.15) / T_pc_abs
T_R_pr = 1.9008
» Z = Z_factor_DAK(p_R_pr,T_R_pr)
Z = 0.9284
» B_g = gas_form_vol_fact(20e6,150+273.15,Z)
B_g = 0.0068

Alternatively, the gas formation volume factor can be computed directly as

» B_g = gas_form_vol_fact_direct(20e6,0.98,150+273.15)
B_g = 0.0068

Gas viscosity is found by

» M = from_kg_per_m3_to_molar_mass(0.98)
M = 0.0231
» mu_g_p_sc = gas_visc_atm_Dempsey(M,150)
mu_g_p_sc = 1.3516e-05
» f = gas_visc_ratio_Dempsey(p_R_pr,T_R_pr)
f = 1.3732
» mu_g = f * mu_g_p_sc
mu_g = 1.8561e-05

Gas viscosity can also be computed directly as

» mu_g = gas_viscosity(20e6,0.98,150)
mu_g = 1.8561e-05

B-4 Water Properties

B-4.1 Water Formation Volume Factor. Fig. B-6 displays the water formation
volume factor for a range of pressures and temperatures, based on a correlation by
McCain (1990). It is clear that the corresponding volumetric changes in water are
much smaller than those in oil (refer to Fig. 2.6). In the examples and questions in this
book we there-fore simply use a water formation volume factor equal to unity.

B-4.2 Water Density. The water density mainly depends on the amount of dissolved
solids it contains. For most formation waters this is primarily NaCl, which may reach
such high concentrations that formation water is often referred to as brine. Fresh
water has a density of 999 kg/m3 (62.4 lbm/ft3) at standard conditions, whereas
NaCl-saturated brine has a density of 1200 kg/m3 (75 lbm/ft3).

B-4.3 Water Viscosity. Water viscosity depends on temperature, salinity, and

pressure, but to a much lesser extent than oil viscosity. In the examples and questions
in this book we use a constant value corresponding to a temperature of approximately
80°C (176 °F): μw = 0.35 × 10–3 Pa·s (0.35 cp).

Fig. B-6—Water formation volume factor.

Appendix C

Wellbore Surveying

C-1 Coordinates
The position of a well is usually described in rectangular coordinates, which are
referred to as northing, easting, and true vertical depth (TVD) (see Fig. C-1). These
coordinates can be obtained through evaluating a survey file that contains triplets of
measured survey data (survey points): along-hole depth (AHD), inclination (the angle
with the vertical), and azimuth (the angle with north). The AHD is sometimes referred
to as measured depth (MD). The measurements can be made with a variety of
special surveying instruments.

C-2 Survey Evaluation

Consider a set of measured survey data consisting of triplets i of along-hole depth si,
inclination αi, and azimuth βi. Various algorithms exist to reconstruct an estimate of the
northing, easting, and TVD corresponding to each triplet. Here, we discuss the most
commonly used algorithm known as the minimum curvature method or circular arc
method. The algorithm was first described by Taylor and Mason (1972) and almost at
the same time by Zaremba (1973), using goniometric functions. Here, we present an
alternative derivation using vector notation.
The survey file evaluation starts at surface and proceeds downward according to a
marching algorithm that can be derived as follows. We define orthonormal base
vectors (ex, ey, ez) along x, y, and z axes that correspond to north, east, and TVD as
depicted in Fig. C-2:

We define the position of survey points Si with the aid of position vectors xi, and the
orientation of the borehole centerline at the survey points with the aid of unit direction
vectors ei. To simplify the figure, but without loss of generality, we have chosen the
position of Si in Fig. C-2 to be on the z-axis. With the aid of the figure, it can be
verified that direction vector ei lies in a vertical plane A that is at an angle βi with
respect to the x-axis. We can now write the components of direction vector ei as a
linear combination of a horizontal direction vector eh, which lies in plane A, and the
vertical direction vector ez :
Fig. C-1—Deviated well with survey points and a kick-off point (i.e., where the well trajectory changes from
vertical to deviated).

Fig. C-2—Well trajectory in an (x, y, z) coordinate system with unit direction vectors e x, e y, and e z. Direction
vector e i is tangential to the wellbore axis in survey point Si, lies in a vertical plane A, and is at an angle αi
(the wellbore inclination) with the vertical. Plane A is at an angle βi (the wellbore azimuth) with the x-axis.

Direction vector eh can be expressed in terms of ex and ey according to

and combination of Eqs. C-1 to C-3 gives us the components of ei as

It can be verified that ei is a unit vector, given that the squares of its components add
up to unity.

Fig. C-3—Quantities involved in survey evaluation with the minimum curvature method.

Next, we use Fig. C-3 and a well-known property of the inner product of two
vectors to derive that ei-1 · ei = cos γi and therefore that

The angle γi is known as the dogleg between the two survey points Si−1 and Si.
Furthermore, the unit direction vector along the chord Ci from Si−1 to Si can be written
as

If we define the increment in AHD as

we can write for the radius of curvature Ri:

where γi should be expressed in radians. With the aid of Fig. C-3 we can now derive
that

and therefore, using the results of Eqs. C-8 and C-9, we find that
Table C-1 — Minimum curvature algorithm.

The unknown position vector xi of survey point Si can then be obtained explicitly from
the known vector xi−1 and the measured survey data as

The minimum curvature algorithm as represented by Eqs. C-1 through C-11 is

summarized in Table C-1.
The parameter γi/Δsi is a measure of the local curvature of the borehole and is
known as the dogleg severity (DLS). Usually the DLS is expressed in degrees/30 m
or degrees/100 ft. High dogleg severities are normally not desirable because they
give rise to high contact forces between the drillstring and the borehole wall and
thereby cause high drillstring drag and torque. Furthermore, high dogleg severities
may also cause problems in running the casing in the hole, and in the cased section of
the borehole high contact forces may lead to severe casing wear. A typical maximum
allowed dogleg severity is on the order of 3°/30 m.

C-3 Numerical Implementation

A MATLAB file, evaluate_survey.m, is available to perform a minimum curvature
calculation for a given survey file with AHD, inclination, and azimuth values. The
survey file should be available in the form of a text file (ASCII file) without comments
(see, e.g., the text file deviated_well_1.txt). For wellbore flow analysis a survey
evaluation is not necessary because it suffices to supply the AHD and inclination
values (see example_traverse_deviated.m). However, to interpret the pressure
profile along the well, it is usually helpful to visualize the well path. To do so,
projections of the well in three planes and a 3D view can be created with the aid of
plot_survey.m. For an example of how to use this function, see example_survey.m.
The corresponding plots have been depicted in Fig. C-4.

Fig. C-4—Projections and 3D view.

Appendix D

Numerical Methods

D-1 Root Finding

Consider a function f(x) = 0 that has at least one root (zero value) in the interval D =
(−∞, ∞), as shown in Fig. D-1. We assume that the function is implicit; i.e., we
assume that it is not possible to obtain an explicit expression for the value of x that
makes f(x) equal to zero. Therefore we need an iterative procedure to approximate
the root, for which various choices are available. Note that root finding is closely
related to optimization because finding an optimum of a function (i.e., a minimum or a
maximum) corresponds to finding a root of the first derivative of that function. (The
requirement that the first derivative of a function is zero in an optimum is a necessary
condition. Sufficient conditions also involve higher derivatives.)

D-1.1 Bisection. A very simple approach is the bisection method. It requires two
starting values x0 and x1 with corresponding function values f(x0) and f(x1) of opposite
sign, such that the solution is bracketed by the starting values (see Fig. D-1a). The
algorithm proceeds by splitting the interval I1 = [x0, x1] into two equal-length intervals
of which the one that brackets the solution is selected as the next interval I2. The
splitting and selection procedure continues until the difference between the function
values is reduced to below a predefined tolerance. The method is guaranteed to
converge as long as f(x) is continuous between the starting values. However, this
method is slow. The error ek , which is the maximum possible absolute difference
between the true root and the most recent estimate xk , k = 2,3,…, is halved every
iteration step, which implies that the method converges linearly.

Fig. D-1a—Two steps of the bisection method; Fig. D-1b—Two steps of the regula falsi method.
D-1.2 Regula Falsi. The regula falsi method also uses two starting values that
bracket the solution. It reduces the size of the bracketing interval Ik every iteration
step k = 2,3,… by replacing one of the endpoints of the interval by the point xi
corresponding to the intersection of the x-axis with the line between the points [xk-1,
f(xk-1)] and [xk , f(xk )], as shown in Fig. D-1b. It can be shown that the method
converges faster than linearly, but not as fast as the Newton-Raphson method
described in the next section.

D-1.3 Newton-Raphson Iteration. The Newton-Raphson method requires that f(x)

has a continuous first and second derivative, which implies that the first derivative is
smooth and uniquely defined for each value of x D. Moreover, it is required that the
first derivative can actually be computed, either analytically or numerically (i.e., with a
finite difference approximation), as opposed to the derivative-free bisection and
regula falsi methods. We start the iteration with a first guess x0. To improve the
estimate, we compute the slope of f(x)—i.e., the derivative f'(x)—in x = x0, which can
also be expressed as

where α and x1 have been defined in Fig. D-2a. This expression can be rewritten as

which gives us a new estimate x1 for the root. This procedure, called Newton-
Raphson iteration, can be generalized by writing

Eq. D-3 can then be applied until the difference (xk+1 – xk ) has been reduced to below
a specified value. The derivative f‘(x) should preferably be computed analytically, but
it may also be obtained with the aid of numerical perturbation:

The choice of the perturbation size Δx is not always trivial. In theory, the smaller the
perturbation, the closer the difference Δf/Δx approaches the differential df/dx, but
roundoff errors in a computer implementation may ruin the estimate if the perturbation
chosen is too small.
Convergence of the Newton-Raphson process is usually fast and can be shown to
be quadratic if the initial estimate is close enough to the root. However, the process
may get in trouble in several situations. It obviously fails for values of f‘(x) equal to
zero (see Fig. D-2b). A more subtle breakdown occurs when the root is located
closely to a change in sign of the first derivative, in which case the process may end
up in an endless loop (see Fig. D-2c). Another type of problem may occur if f(x) has
multiple roots in D, in which case the process may show convergence to a root that
was not intended to be found. Various controls on the iteration process may be
introduced to counteract these problems— e.g., allowing a maximum change in x in
each iteration step, or restarting the process with a reduced change in x when the
iteration fails to converge in a predefined number of steps. Prior inspection of the
nature of the function f(x) before using the Newton-Raphson process may help to
reduce the chance of convergence problems.

D-1.4 Multivariable Root Finding. The Newton-Raphson method can easily be

extended to multiple variables. Consider a system of n nonlinear algebraic equations
in terms of n unknowns:

Using matrix-vector notation, the system can be written in compact form as

Fig. D-2a—Principle of Newton Raphson iteration; Fig. D-2b—Convergence failure caused by a zero value
of the derivative in x0; Fig. D-2c—Convergence failure caused by an endless loop.

The multivariable equivalent to the single-variable Newton-Raphson update (Eq. D-3)

becomes

where

Eq. D-8 is known as the Jacobian matrix, or Jacobian for short. Just as in the single-
variable case, the derivatives should preferably be computed analytically but may also
be obtained through numerical perturbation.
 In practice, one normally does not actually compute the inverse of the Jacobian but
uses the computationally more attractive alternative

which involves solving a system of n linear equations to find the unknown error vector
(xk+1 − xk). In MATLAB we can use the backslash operator \ to perform this
computation. Using the variable J to indicate the Jacobian, we can then write
x(k+1) = x(k) - J\f

The iterations are repeated until some convergence criterion is met; this is usually
expressed in terms of a norm of the error vector; e.g.,

where ε is small. Alternatively, convergence may be defined in terms of the norm of f.

Just like in the 1D case, there are various ways in which the iteration procedure may
fail. In addition, the Jacobian may be ill-conditioned, and further controls on the
iteration process may therefore be required to achieve a robust multivariate Newton
Raphson procedure.

D-1.5 Standard MATLAB Functions. Single-Variable Root Finding. Often a good

alternative to programming a root-finding algorithm is to simply use a standard MATLAB
function. To find the root of a single-variable function, one can use the MATLAB function
fzero, which makes use of bisection and other derivative-free optimization methods
(see MATLAB 2017). For example, to find the root of y = cos(x) nearest to an initial
guess x = 1.5, the syntax is
>> fzero(@cos,1.5)

which produces the result

ans = 1.5708

The symbol @ in front of the function cos serves to specify a function handle—i.e., a
reference to a function that may be passed as an argument to another function (in this
case the function handle @cos is passed to the function fzero). Instead of making an
initial guess, it is also possible to specify an interval bracketing the solution. Thus,
e.g., to find the root of cos(x) inside the interval [1, 2], we write:
>> fzero(@cos,[1 2])
ans = 1.5708

Alternatively one can write

>> fzero(@(x) cos(x),[1 2])

which produces the same result. In this latter case cos(x) is defined as an
anonymous function with a single argument x as specified by the statement @(x)
preceding the function. Anonymous functions can be used to create simple functions
in MATLAB without writing an m-file. The general syntax is fname = @(arglist) expr,
where arglist is a comma-separated list of arguments and expr is the function
definition. For example, the statement test = @(x,y) x^2+y^2 defines the
anonymous function test with arguments x and y, which may be called inside another
function. The statement sqrt(test(3,4)) therefore results in the answer 5. Note that
the symbol @ now has a different role. In the same way we may call fzero with an
ordinary function—i.e., a function defined in an m-file— which may then also contain
additional parameters. If, e.g., we want to find the zeros of y = cos(x) – a for
different values of parameter a, we can define y in an m-file as
function y = example(x,a)
y = cos(x) - a

and then call fzero after first specifying a value for a:

>> a = 0.1;
>> fzero(@(x) example(x,a),[1 2])
ans = 1.4706

Multivariable Root Finding. To find the roots of a vector-valued multivariable

function (i.e., a vector of functions that each depend on multiple variables), one can
use the MATLAB function fsolve, which is part of the MATLAB Optimization Toolbox. It
performs multivariable optimization (root finding) using a variety of optimization
algorithms that are hidden to the user. In this sense it is a multidimensional version of
fzero (see MATLAB 2017). The syntax is also partly an extension of the syntax for
fzero. For example, to find the root of the vector-valued function [y1 y2] = [cos(x1) – a
sin(x2) + b] for different values of parameters a and b nearest to an initial guess [x1
x2] = [1.5 0.5], we can define y in an m-file as
function y = example(x,a,b)
y(1) = cos(x(1)) - a;
y(2) = sin(x(2)) + b;

and then call fsolve after first specifying values for a and b:
>> a = 0.1;
>> b = 0.2;
>> x start = [0,0]; % starting values for x(1) and x(2)
>> fsolve(@(x) example(x,a,b), x start)
ans = 1.4706 -0.20136

Note that in this simple example each of the function vector elements y1 and y2
depends on just one variable—i.e., x1 and x2, respectively. However, it is possible to
specify n function vector elements that can each be a function of up to n variables.
Consult the online MATLAB help function for more information on the use of fsolve.

D-2 Differential Equations

D-2.1 Initial Value Problems. All steady-state pressure drop equations for pipeline
or wellbore flow, such as Eqs. 3.24 or 3.33, can be expressed as

where g is a known nonlinear function of s and p. Eq. D-11 is a first-order ordinary

differential equation that needs one boundary condition specifying a certain value for
the pressure at a certain point along the pipeline or the well:

To obtain the pressure p at any point along the pipeline, we can integrate Eq. D-11
starting from the boundary condition (Eq. D-12):

This kind of equation is often used to describe problems that depend on time (instead
of on distance s, as in our case), in which case the boundary condition is usually
specified at the start of the time interval. Therefore the boundary condition is often
referred to as an initial condition, and the problem as an initial valueproblem.

D-2.2 Numerical Integration. Generally, the integral in Eq. D-13 cannot be solved
analytically but needs to be evaluated through numerical integration. The simplest
approach is to discretize Eq. D-11 by replacing the differential dp/ds by a difference
Δp/Δs and to rewrite the result as

This gives us an algorithm to compute an approximate new value pk+1 at sk+1 from a
known value pk at sk :

where Δs = sk+1 – sk and g(sk,pk) are a shorthand notation to indicate the evaluation of
function g(s,p) at s = sk . More formally, the same result is obtained by using a Taylor
expansion for p at sk ,
and maintaining only the first-order term. This also illustrates that the truncation error
in Eq. D-15, known as an explicit first-order Euler scheme, is of the order of (Δs)2.
Although conceptually quite simple, the first-order Euler scheme is not very
efficient for use in wellbore flow calculations, and so alternative algorithms, with a
much smaller error for the same step size, should be applied. A popular class of
integration algorithms are the Runge-Kutta routines, which use multiple evaluations of
the function g(s,p) on the interval Δs, instead of only a single evaluation at the
beginning of the interval as in the first-order Euler scheme. Especially in combination
with an automated strategy to adapt the step size in order to achieve a predefined
accuracy, they are very powerful. Many other schemes have been developed that
may have a superior computational performance or accuracy depending on the nature
of the function g(s,p) (see, e.g., Press et al. 2007).
For our purpose the fourth-order Runge-Kutta scheme with variable step size that
is readily available in MATLAB provides a fit-for-purpose solution. A simple check on the
accuracy of the numerical integration of wellbore or pipeline pressure drop
computations can be made by repeating the integration in the opposite direction. For
example, after computing the flowing tubinghead pressure (FTHP) through bottom-up
integration starting from a known FBHP, the FBHP is recalculated through top-down
integration starting from the FTHP. The difference between the two flowing bottom-
hole pressure (FBHP) values forms a good indication of the accuracy of the numerical
integration. Typically a difference of less than 1% of the total pressure drop would be
acceptable, although often a much better performance can be achieved. Note,
however, that although a small difference is a necessary condition, it is not a sufficient
one.

D-2.3 Numerical Implementation. The fourth-order Runge-Kutta routine with variable

step size in MATLAB is named ode45. To compute a pressure p = pout, at s = sout starting
from a known value at , the following commands can be used:
» interval = [s_in,s_out];
» boundcon = p_in;
» options = [];
» [s,p] = ode45(‘g_dpds’,interval,boundcon,options,p1,p2);
» n = length(p);
» p out = p(n)

The variable options is a vector with parameters to control the numerical integration;
the default is an empty vector: options = []. Useful parameters to control are the
maximum step size MaxStep and the relative tolerance RelTol, which can be set with
options = odeset(’MaxStep’,value1,’RelTol’,value2). Default values are MaxStep
= 0.1*interval and RelTol = 1.e-3. The variable ’g_dpds’ (between quotes) that
forms the first element of the argument list is the name of the user-defined MATLAB
function (m-file) that defines the function g(s,p). It is called many times by ode45
during the Runge-Kutta integration. The function most likely requires various
parameters, which can be passed by means of the argument list after options; here
we used parameters p1 and p2 as examples. The function would typically look
something like this:
function dpds = g_dpds(s,p,flag,p1,p2)
%
% User-defined function to compute the derivative dp/ds.
%
% dpds = pressure gradient, Pa/m
% flag = dummy variable, -
% p = pressure, Pa
% p1 = parameter, -
% p2 = parameter, -
% s = along-hole coordinate, m
%
dpds = …;

The three dots on the last line should be replaced by the appropriate function
definition in terms of s, p, p1, and p2. The first two elements of the argument list in
the function header contain the independent and dependent variables s and p. The
dummy variable flag is not used but has to be present as the third element of the
argument list. Thereafter follow the parameters. Note that the arguments in the calling
sequence of ode45 are not identical to those in the header of g_dpds. The output of
ode45 consists of two column vectors with values of the independent and the
dependent variables (here s and p) for regularly spaced values of s. The last element
of vector p is the required output pressure p_out.
It is possible to integrate a system of first-order differential equations, rather than
a single equation, in the same fashion. This requires only that the dependent variable
and the function definition be defined as vectors instead of scalars. We make use of
this feature to compute the individual contributions of gravity, friction, and acceleration
losses to the total pressure drop in a wellbore. For example, the function gas_dpds
has the following header:
function dpds = gas_dpds(s,p,flag,…)

where the dots indicate parameters. Now, dpds is a four-element vector defined as
dpds = [dpds_tot;dpds_grav;dpds_fric;dpds_acc]

while the variable p in the argument list is also a vector, defined as

p = [p_tot;p_grav;p_fric;p_acc]
Appendix E

Multiphase Flow Correlations

E-1 Hagedorn and Brown

E-1.1 Pseudoholdup Correlations. Hagedorn and Brown (1965) developed a set of

correlations to compute the pressure drop of gas/liquid flow in vertical wells. They
performed a large number of experiments using mixtures of air and water, and air and
crude in a test well approximately 450 m (1,500 ft) deep and with tubing sizes up to
38 mm (1.5 in.). They measured pressures at various points along the well but did not
measure liquid holdup. Instead, they computed pseudoholdup values by
backcalculating the theoretical holdup values that best corresponded with the
measured pressures. The pseudoholdup values were presented with the aid of
graphical relationships in the form of three charts for various auxiliary parameters
defined in terms of the dimensionless groups (Eqs. 4.37 through 4.40). Here we
present equations obtained by curve fitting the original charts published by Hagedorn
and Brown (1965). The first chart (see Fig. E-1) defines a relationship between the
liquid viscosity number,

Fig. E-1—Chart 1 for Hagedorn and Brown correlation, corresponding to Eq. E-3.
and a corrected liquid viscosity number,

according to

where the coefficients a0 to a3 have been listed in Table E-1. The second chart (Fig.
E-2) defines a relationship between

and the correction factor

according to

with coefficients listed in Table E-1. The third chart (Fig. E-3) gives a relationship
between

where psc is the pressure at standard conditions, and the required liquid holdup
(divided by the correction factor ψ)

according to
with coefficients listed in Table E-1. Note the various lower and upper limits for the x
values in these expressions. The correlations (fitted from the charts) should not be
used outside their limits because they may display erratic behavior. If it is required to
use values outside the limits, the necessary values should be obtained by graphical
extrapolation in the charts, and not by numerical extrapolation using the correlations.
Occasionally the holdup values Hl computed with Eqs. E-10 and E-11 are below the
liquid volume fraction λl, which is physically impossible. In that case it is customary to
use the volume fraction instead. Eqs. E-1 through E-12 have been programmed in
MATLAB file Hag_Brown_hold_up.m, which forms a subroutine in Hag_Brown_dpds.m but
may also be used on a standalone basis to recompute the holdups in a post-
processing step—e.g., after computing a traverse with Hag_Brown_dpds.m.

Coefficient Value
a0 1.912 2 × 10–3
a1 5.634 8 × 10–2
a2 –1.218 3 × 10–1
a3 9.095 0 × 10–2
b0 1.808 9 × 10–5
b1 –1.276 9 × 10–3
b2 1.002 9 × 10–2
b3 3.691 7 × 10–1
c0 1.958 5 × 10–5
c1 –1.276 9 × 10–3
c2 2.705 8 × 10–2
c3 –4.894 0 × 10–2
c4 1.00 0 × 100
d0 4.326 0 × 10–11
d1 –3.790 2 × 10–5
e0 –3.752 0 × 10–10
e1 –1.932 3 × 10–4
e2 1.000 0 × 100
f0 6.489 8 × 10–5
f1 1.147 0 × 100
f2 –1.853 0 × 10+1
g0 4.531 9 × 10–4
g1 1.000 0 × 1010
Table E-1—Coefficients for Eqs. E-3, E-6, E-10, and E-11.
Fig. E-2—Chart 2 for Hagedorn and Brown correlation, corresponding to Eq. E-6.

Fig. E-3—Chart 3 for Hagedorn and Brown correlation, corresponding to Eqs. E-10 and E-11.
E-1.2 Pressure Drop Analysis. Once the holdup values have been computed, the
mixture velocities can be computed using Eqs. 4.32, 4.33, and 4.35 as presented in
the body of the text. The mixture friction factor fms can then be obtained with the aid
of Eqs. 3.12 and 3.13 or the corresponding Moody friction factor chart (Fig. 3.3) by
replacing the single-phase properties μ, ρ, and ν by the mixture properties μms, ρms,
and νms. The pressure drop can then be computed with the aid of the expression as
proposed in the original publication of Hagedorn and Brown (1965):

where Δvm is the velocity increment over a numerical integration interval Δs. The
acceleration term in Eq. E-16 is valid only for positive acceleration (i.e. increasing
absolute value of the velocity) in a production well. To capture the general case of
positive and negative acceleration in injection and production wells, the term should
be rewritten as –ρms|Δvms|Δvms/2Δs. (Note, however, that this is of only theoretical
interest because acceleration hardly ever plays a significant role except at the top of
high-rate gas production wells, where very high velocities may occur.)
For the friction factor, Hagedorn and Brown (1965) suggested using the single-
phase Moody diagram with a multiphase Reynolds number defined as

and a mixture viscosity defined as

Note that this is an unusual mixture viscosity definition. Most multiphase flow
correlations use Eq. 4.35.
Alternatively, the friction loss can be computed according to Eq. 4.28, while the
acceleration loss term is often replaced by multiplication of the head and friction
terms with an acceleration factor, similar to what was done in the expressions for
single-phase oil flow (Eq. 3.23) and single-phase gas flow (Eq. 3.33). For a vertical
well—i.e., for sin(θ) = –1—the resulting pressure drop expression then becomes

where Ek,D is a dimensionless mixture kinetic energy term defined by Beggs and Brill
(1973) as
Note that the acceleration term in Eq. E-16 is valid only for positive accelerations (see
also the comments following Eq. E-13). Eq. E-16, with Reynolds number and
viscosity definitions shown by Eqs. E-14 and E-15, has been implemented in MATLAB
file Hag Brown dpds.m.

E-1.3 Example E-1: Hagedorn and Brown Correlation. Consider the oil well with
the properties given in Table 4.1. The local flow rates and densities just below the
tubing head are qg = –0.0391 m3/s, qo = –0.00406 m3/s, qw = –0.001 m3/s, ρg = 4.58
kg/m3, ρo = 841 kg/m3, and ρw = 1.05 × 103 kg/m3. The viscosities at ptf and Tif are μg
= 8.69 × 10–6 Pa·s, μo = 0.0122 Pa·s, and μw = 0.350 × 10–3 Pa·s.

Question 1 What is the magnitude of the liquid holdup just below the tubing head?
Answer The local oil and water fractions are

The local liquid properties are therefore

The surface area of the tubing is

such that the superficial and mixture velocities follow as

The dimensionless groups then become

With the aid of Fig. E-1 (or Eqs. E-1 through E-3), we find that

x1 = Nμ = 0.0774,

which corresponds to

Nμc = y1 = 0.00559.

Also, with the aid of Fig. E-2 (or Eqs. E-4 through E-6), we find that

which corresponds to

ψ = y2 = 1.00.

Finally, with the aid of Fig. E-3 (or Eqs. E-8, E-9 and E-11), it follows that

which results in a liquid holdup of

Hl = y3ψ = 0.369.

It remains only to be checked that the computed holdup is realistic (i.e., that it is not
less than the liquid volume fraction λl):
 Question 2 What is the corresponding magnitude of the pressure gradient? Use
Eq. E-16 with Reynolds number and viscosity definitions shown by Eqs. E-14 and E-
15.
Answer The necessary mixture properties can be computed as

such that the mixture Reynolds number becomes

With a dimensionless roughness,

and with the aid of the Moody chart (Fig. 3.3), we find a friction factor

fms = 0.0168,

such that the friction loss becomes

The head loss follows as

and with a dimensionless kinetic energy,

we find an acceleration loss

The total pressure gradient is therefore

E-2 Mukherjee and Brill

E-2.1 Flow Regimes. Mukherjee and Brill (1983; 1985a,b) performed a large number
of tests in a 1.5-in.-diameter flow loop with a U-shaped inclined section that could be
raised from horizontal to vertical. In this way they studied the behavior of flow in
upward and downward directions over all possible inclinations. They used air as the
gas phase and kerosene or lube oil as the liquid phase. Both legs of the U contained
a 32-ft transparent section to observe flow regimes and a capacitance measurement
device to measure holdups. Through curve fitting of a large number of measurements,
they developed a set of numerical expressions for the flow pattern boundaries. These
are expressed in terms of the dimensionless numbers (Eqs. 4.37 through 4.40) and a
pipeline inclination angle θΜΒ that can be related to our definition for the wellbore
inclination α according to

where d is the flow direction defined as

The boundary between annular mist flow and bubble or slug flow is expressed as

where

The boundary between bubble flow and slug flow depends on the flow direction. The
upflow boundary is given by

where
and the downflow and horizontal boundary is given by

where

Finally, the transition between slug flow and stratified flow is given by

where

The flow regime for a given set of parameters can now be determined by following
the flow chart in Fig. E-4. The flow chart logic and Eqs. E-18 through E-27 have been
programmed in MATLAB file Muk_Brill_flow_reg.m, which forms a subroutine in
Muk_Brill_dpds.m but may also be used on a standalone basis to recompute the flow
regimes in a post-processing step.

E-2.2 Holdup Correlation. Once the flow regime has been determined, the liquid
holdup can be computed. Mukherjee and Brill (1983) determined a single holdup
correlation for all flow regimes given by

where the coefficients C1 through C6 depend on the flow regime and are given in
Table E-2. Eq. E-28 has been programmed in MATLAB file Muk_Brill_hold_up.m,
which forms a subroutine in Muk_Brill_dpds.m but may also be used on a standalone
basis to recompute the holdups in a post-processing step.

E-2.3 Pressure Drop Analysis. The pressure drop equation used by Mukherjee and
Brill is flow regime dependent. For slug and bubble flow it can be written as
Fig. E-4—Flow chart to determine flow regime: 1, bubble flow; 2, slug flow; 3, annular mist flow; 4, stratified
flow. Adapted from Mukherjee and Brill (1985a).

Coefficient Uphill or Horizontal Flow Downhill Stratified Flow Other Downhill Flow
C1 –0.380 113 –1.330 282 –0.516 644
C2 0.129 875 4.808 139 0.789 805
C3 –0.119 788 4.171 584 0.551 627
C4 2.343 227 56.262 268 15.519 214
C5 0.475 686 0.079 951 0.371 771
C6 0.288 657 0.504 887 0.393 952
Table E-2—Coefficients for Eq. E-28, (after Mukherjee and Brill, 1983).

where the dimensionless mixture kinetic energy term Ek,D is the one defined in Eq. E-
17. The friction coefficient fms is determined from the Moody diagram (i.e., from Eqs.
3.12 and 3.13) with

For annular flow the same pressure drop Eq. E-29 is used, but with ρms in the friction
term replaced by ρmn, and with a friction factor

where fmn is the no-slip Moody friction factor and fr is a friction factor ratio that
depends on the holdup ratio
according to the values in Table E-3, which should be linearly interpolated as
required.
For stratified flow the friction factor is computed with the aid of a mechanistic
model that explicitly accounts for the velocities in the liquid and gas layers, and for the
shear forces between the layers and between the fluids and the wall. Here we do not
reproduce the corresponding equations, which are quite elaborate, but refer the
reader to Mukherjee and Brill (1985b) or Brill and Mukherjee (1999) for further
details. Eqs. E-29 through E-32 and the equations for stratified flow have been
programmed in MATLAB file Muk_Brill_dpds.m.

E-2.4 Example E-2: Mukherjee and Brill Correlation. Consider the same well as in
Example E-1.
Question 1 What are the flow regime and the magnitude of the liquid holdup just
below the tubing head?
Answer The first steps are identical to those of Example E-1, up to and including
the computation of the dimensionless groups Nlv, Ngv, Nd, and Nμ. Because vms < 0
(upward flow) and α = 0 (vertical well), it follows from Eq. E-19 that d = 1 and from
Eq. E-18 that θMB = π/2. To determine the flow regime, we need to follow the flow
chart in Fig. E-4. The first step requires computation of the boundary between annular
mist flow and bubble or slug flow according to Eq. E-20:

Hr fr
0.01 1.00
0.20 0.98
0.30 1.20
0.40 1.25
0.50 1.30
0.70 1.25
1.00 1.00
10.00 1.00
Table E-3—Coefficients for Eq. E-31.

A1 = 1.401 − 2.694 × 0.0774 + 0.521 × (14.8)0.329 = 2.46, Ngv,sm = 102.46 = 288.

Because θMB > 0 and Nlv < Nlv,b s (114 < 288) the flow regime is not annular flow, and
we proceed to the next step, which requires the boundary between bubble and slug
flow for upward flow (see Eq. E-22):
Because θΜΒ > 0 and Nlv < Nlvb s (14.8 < 316), it follows that the flow regime is slug
flow. The liquid holdup is given by Eq. E-28 with coefficients from the first column in
Table E-2 as

Note the difference between the holdup values computed by the two methods:
approximately 0.37 for Hagedorn and Brown (H&B) vs. 0.21 for Mukherjee and Brill
(M&B). This gives an indication of the lack of accuracy of these types of empirical
correlations. Fig. E-5 illustrates the difference and also displays the liquid volume
fractions for both methods (see also Fig. E-4). Note that for depths below about 1150
m, the liquid holdup becomes equal to unity because the pressures become higher
than the bubblepoint pressure. The holdup values for both correlations are larger than
or equal to the liquid volume fractions—as they should be. Fig. E-5 also depicts the
flow regimes for the M&B correlation, and it can be seen that the flow regime
changes from single-phase liquid flow (0) for depths below 1150 m to bubble flow (1)
between 1150 and 550 m and slug flow (2) above 550 m. An example of how to
create a graph of liquid holdup vs. depth is given in MATLAB file example_hold_up.m.

Fig. E-5—Liquid holdups, liquid volume fractions, and flow regimes for a vertical multiphase well with
parameters given in Table 4.1 for comparison between the Mukherjee and Brill (M&B) and the Hagedorn
and Brown (H&B) methods. The flow regimes (indicted with a stepped solid line) correspond to the M&B
method, with 0 indicating single-phase liquid flow, 1 bubble flow, and 2 slug flow (only the top 1800 m has
been displayed).

Question 2 What is the corresponding magnitude of the pressure gradient?

Answer The necessary mixture properties can be computed as

The mixture Reynolds number becomes

Using a dimensionless roughness,

and with the aid of the Moody chart (Fig. 3.3), we find a friction factor

fms = 0.0208,

such that the friction loss becomes

The head loss follows as

and the dimensionless kinetic energy is given by

such that the acceleration loss follows as

and the total pressure gradient is given by

Note that this value is also considerably different from the one previously
computed: 14.3 × 103 Pa/m for H&B vs. 9.18 × 103 Pa/m for M&B. However, this is a
local difference, and at greater depths the discrepancy is much smaller. Overall, the
methods produce results that are near-identical (for this example)—i.e., well within
the typical accuracy that may be expected for production engineering calculations.

E-3 Drift Flux Models

E-3.1 Profile Parameter. Another class of relatively simple multiphase wellbore flow
models is the family of drift flux models, where drift refers to the difference between
the gas velocity and the mixture velocity (see, e.g., Wallis 1969 and Shi et al. 2005a,
b). Shi et al. (2005a, b) performed a series of experiments in a flow loop containing a
0.15-m-diameter transparent pipe section that could be raised from horizontal to
vertical. They accurately measured holdups and used their experimental results to
tune the free parameters in a drift flux algorithm that is described in broad lines
below. Their aim was to develop a fast algorithm for use in a reservoir simulator. Shi
et al. (2005b) considered two-phase gas/ liquid and two-phase oil/water flow, and in
a follow-up paper they proposed a drift flux formulation for three-phase gas/oil/water
flow (Shi et al. 2005a). Here we restrict the analysis to the two-phase gas/liquid
situation, in which water and oil flow without slip.
In the drift flux approach, as first proposed by Zuber and Findlay (1965), the
difference in phase velocities, and therefore the liquid holdup, is explained as resulting
from two physical mechanisms: (1) gas has a tendency to slip through the liquid
because of a difference in densities, and (2) gas bubbles have a tendency to
concentrate at the center of the wellbore, where the fluid velocities are highest (see
Fig. E-6). The joint effect of these mechanism can be expressed as

where vg is the gas velocity; vd is the drift velocity or slip velocity, defined as the
difference between the gas and mixture velocities; C0 is the profile or distribution
parameter; and vms is the mixture velocity.
Various studies have reported values for C0; typical values range from 1.0 to 1.5
depending on factors such as the flow regime, pipe diameter, and gas fraction (for a
detailed discussion, see Shi et al. 2005b). In general, the profile effect reduces for
increasing gas fractions and increasing mixture velocities. Typical values of C0 for the
bubble flow and annular flow regimes are 1.2 and 1.0, respectively. Shi et al. (2005b)
found that for the large pipe diameter used in their experiment the profile effect was
also negligible at lower flow rates,—i.e., C0 should be taken as 1.0—but they suggest
that for smaller diameters C0 = 1.2 is a better choice. They proposed the following
relationship:

where C0,bub is the value of C0 in the bubble flow regime and where the parameter γ
has a value between zero and unity. The parameter γ should approach its upper limit
at high values of Hg or vms to ensure that C0 reduces to 1.0. Therefore, the value of γ
is obtained from

Fig. E-6—Velocity and concentration profiles in upward pipe flow. The gas bubbles tend to be
concentrated at the center of the pipe, where the velocity is highest.

where β is also a parameter with a value larger than zero defined as

and is the value of β at which C0 starts to drop below C0,bub (see Fig. E-7).
The variable vg,fld is the flooding velocity—i.e., the gas velocity that is just sufficient
to support a thin annulus of liquid such that it does not flow back down along the
wellbore (Wallis and Makkenchery 1974):
where the critical velocity vc is defined by the following equation (refer to Eq. 4.37;
note that in line with the sign convention in our text, we define upward velocities as
negative. In the original paper by Shi et al. (2005ab), upward velocities are positive):

and where is the critical Kutateladze number, which is a function of a modified

pipe diameter number (cf. Eq. 4.39), is

Fig. E-7—Profile parameter as function of β for C0,bub = 1.2 and .

according to a graphical correlation displayed in Fig. E-8. Curve fitting of the graph
results in

The ratio vms/vg,fld is typically smaller than unity for most flow regimes and may even
be negative in case of countercurrent flow of gas and liquids. It approaches unity at
the transition to annular flow, and will exceed unity for fully developed annular flow.
Note that in the latter case the value of γ will remain bounded to unity.

E-3.2 Drift Velocity. The drift velocity also depends on the flow regime. Following Shi
et al. (2005b), we consider a range of values between two extremes. For bubble
flow, corresponding to very low gas fractions, we use an expression for the gas
velocity from Harmathy (1960), who determined the rise velocity of small bubbles
through a stationary liquid to be

At the other extreme of very high gas holdup values we have to consider annular flow,
in which case the gas velocity relative to a stationary liquid becomes equal to the
flooding velocity vfld, which was defined in Eq. E-37. Between these two endpoints the
values of vd are given as a function of Hg by a flooding curve (see Shi et al. 2005b),
defined as

Fig. E-8—Critical Kutateladze number as function of modified pipe number. After Shi et al. (2005b).

where
with a linear interpolation for values of Hg between the tuning parameters a1 and a2,
where 0 < a1 < a2 < 1. The parameter mα is a multiplier to take account of the effect
of the wellbore inclination α, defined as

where n1 and n2 need to be determined experimentally and where m0 is a nonunit

multiplier for vertical flow, which serves as an additional tuning parameter. In the
experiments of Shi et al. (2005b) a total of seven parameters for gas/liquid flow were
determined by minimizing the root-mean-squared error between a large number of
measured and modeled holdup values. The resulting optimal values for large-diameter
pipes, as used in their experiments, are displayed in the first row of values in Table
E-4. The second row of values corresponds to optimal parameters when the values
of C0,bub and are fixed at 1.2 and 0.6, respectively, which may provide better results
for smaller-diameter pipes (say, below 0.10 m; no specific diameter is mentioned in
the original publication).

E-3.3 Holdup. The gas holdup Hg is implicitly related to the profile parameter C0 and
the drift velocity vd according to

and therefore it has to be computed iteratively. A simple numerical implementation is

obtained through Picard iteration (subsequent substitution) with a damping factor 0 <
f < 1:

where k is the iteration counter and where the gas volume fraction forms a convenient
starting value: .

E-3.4 Example E-3: Drift Flux Model. Consider the same well as in Example E-1.
Question What are the flow regime and the magnitude of the liquid holdup just
below the tubing head?

C0,bub a1 a2 m0 n1 n2
1.0 1.0 0.06 0.21 1.85 0.21 0.95
 1.2 0.6 0.06 0.12 1.27 0.24 1.08
Table E-4—Parameters for the gas/liquid drift flux model of Shi et al. (2005b).

Answer The drift flux model does not explicitly determine the flow regime. The
liquid holdup can be computed iteratively as follows:

1. Follow the same steps as in Example E-1, up to and including the

computation of
vsg = –12.8 m/s, vsl = –1.66 m/s, vms = –14.5 m/s.
2. Determine the drift flux parameters. Because the pipe diameter (0.0623 m) is
below 0.10 m, we choose the parameters in the second row of Table E-4.
3. Choose λg = 0.885 as a first guess for the gas holdup.
4. With the aid of Eqs. E-38 and E-39, and either Fig. E-8 or Eq. E-40,
determine

and

which, with the aid of Eq. E-37, leads to the flooding velocity

5. Next, use Eqs. E-34 through E-36 to compute a first guess of the profile
parameter C0 as follows:

6. Compute a first guess of the drift velocity vd with Eqs. E-42 and E-43
according to
 where mα = m0 = 1.27 because the well is vertical.

7. Compute an improved estimate of the gas holdup with Eq. E-46:

where we used a damping factor, f = 0.5.

8. Repeat Steps 5 through 7, using the damped Picard iteration scheme (Eq. E-
46), until convergence, and compute Hl = 1 – Hg.

In this example, one iteration is enough to reduce the absolute error in the gas holdup
to below 0.01 and another five to bring it down to below 0.001. The corresponding
parameter values are

vg,fld = –4.93 m/s, C0 = 1, Hg = 0.865, Hl = 0.135.

Note that Hl = 0.130 is almost identical to λl = 0.115; i.e., there is almost no slip of
the gas in this model. This is different from the values for Hl as obtained with both the
Hagedorn and Brown and the Mukherjee and Brill methods in Examples E-1 and E-2,
which were 0.369 and 0.211, respectively.
Appendix F

Relative Permeabilities

F-1 Physics
The physics of the relative permeability effect is related to the interfacial tension
between the phases (which gives rise to capillary pressure), the wettability of the
rock, and the geometry of the pore network (see, e.g., Dullien 1979). For the
reservoir engineering aspects of relative permeabilities, see, e.g., Muskat (1949),
Dake (1978), or Chierici (1994). The products kg = k krg, ko = k kro, and kw = k krw are
known as the effective permeabilities to gas, oil, and water, respectively. Here, k is
the absolute permeability governed by rock properties only, expressed in m2 (md),
and 0 ≤ krg ≤ 1, 0 ≤ kro ≤ 1, and 0 ≤ krw ≤ 1 are dimensionless relative permeabilities,
which are nonlinear functions of the phase saturations and represent the reduction of
permeability to one phase as a result of the presence of the other phases. In most
descriptions of two-phase oil/water flow the rock is assumed to be water-wet; i.e.,
water is the wetting phase and oil the nonwetting phase. Wettability is defined in
terms of the contact angle, which is the angle between the surfaces of a droplet and
a solid in contact, measured at the contact line and through the liquid (see Fig. F-1).
For water-wet rock, oil production corresponds to imbibition, i.e., an increasing
wetting-phase saturation. During the migration process of oil from the source rock
into the reservoir, the reverse process took place, known as drainage (i.e., a
decrease of the wetting-phase saturation). In an actual reservoir, the rock may be
water-wet, oil-wet, or, as is often the case, mixed-wet—i.e., having a contact angle
close to 90°. In case of two-phase gas/water flow, water is always the wetting
phase, and gas production therefore also corresponds to imbibition. In case of oil and
gas flow, oil is the wetting phase, and we therefore have to consider drainage, at
least under a normal production scenario. Under three-phase (gas/oil/water)
conditions, the customary approach is to assume that water is the wetting phase, oil
the intermediate phase, and gas the nonwetting phase, and that the relative
permeabilities are dependent only on the water and gas saturations. Although this
very simplistic description is known to be incorrect in most circumstances, it provides
a workable description of multiphase permeabilities for numerical reservoir simulation,
and various models based on these assumptions have been proposed (see, e.g.,
Stone 1973, Aziz and Settari 1979, or Chierici 1994).
Fig. F-1—Relation between contact angle and wettability. Left: water droplet on a water-wet surface
(contact angle < 90°). Right: water droplet on an oil-wet surface (contact angle > 90°).

Fig. F-2 depicts a typical set of relative permeability curves for oil and water flow
during imbibition in a water-wet rock. The wetting phase (i.e., the water) preferentially
occupies the smaller pores. During the water drainage process (i.e., the oil migration
process) not all water could be displaced from these small pores, and so at the end
of the process some water was left, known as connate water or interstitial water.
The imbibition process (i.e., the oil production process) therefore starts from a
situation with an initial water saturation equal to the connate water saturation.
Because there is no flow of water until the water saturation exceeds this initial value,
it is also referred to as the immobile or critical water saturation, and we will indicate
it with the symbol Swi, where the subscript i refers to “immobile.” At the end of the
water imbibition process, a certain amount of oil remains trapped in the larger pores
from which it cannot be displaced by water. This oil is known as residual oil, and the
associated saturation Sorw as the residual oil saturation after waterflooding. At the
beginning and the end of the imbibition process, the presence of connate water and
residual oil in the pores results in relative permeabilities below the theoretical
maximum of unity. These values, and , are known as the endpoint relative
permeabilities under waterflooding conditions. The more water-wet the rock, the
lower the value of .
Fig. F-3 depicts a typical set of relative permeability curves for gas and oil flow
during oil drainage. Water is assumed to be present as immobile water, and therefore
the relative permeability values should be interpreted as relative to a reduced
permeability . At increasing gas saturations, the gas does not start flowing until
it has reached the critical gas saturation Sgc. The critical gas saturation is not very
important when oil is forced through the formation but becomes important for solution
gas drive (primary recovery), where gas bubbles need to slowly grow into
interconnected paths before the oil starts flowing. The endpoint relative permeability
is equal to unity; but because we assumed the presence of connate water, the
corresponding effective permeability to oil is, in fact, equal to . The residual liquid
saturation in Fig. F-3 consists of the sum of the residual oil saturation Sorg and the
immobile water saturation Swi. The value of Sorg will typically have a finite value in case
of primary recovery but may approach zero in case of gasflooding.
Fig. F-2—Relative permeabilities to oil and water during imbibition (i.e., increasing water saturation). In
this example, , Swi, = 0.15, Sorw = 0.20, and now = nw = 3.

Fig. F-3—Relative permeabilities to gas and oil during drainage (i.e., decreasing oil saturation). In this
example, , Sgc = 0.15, Swi = 0.15, Sorg = 0.05, and ng = nog = 3.
F-2 Corey Expressions
Relative permeabilities are strongly dependent on formation and fluid properties and
should therefore ideally be determined from laboratory experiments, in particular from
coreflooding tests. In the case of heterogeneous reservoir properties, as is nearly
always the case, there is a difference between the permeabilities at the core scale
and the upscaled permeabilities, also known as pseudopermeabilities, at the near-
well reservoir length scale. In the absence of measured data, it may be necessary to
revert to empirical relationships for the relative permeabilities. These relationships are
also used to fit measured permeability data and can then be used for upscaling and
numerical simulation. For oil/water flow under imbibition conditions (i.e., oil
production), we use the relationships

where is the normalized water saturation defined as

and where now and nw are known as the Corey exponents. They are both larger than
1, with typical values between 2 and 4. For oil/gas flow under drainage conditions
(i.e., oil production again), we use

where is the normalized gas saturation defined as

Figs. F-2 and F-3 have been produced with values ng = nog = now = nw = 3. In case of
oil/water flow, the total relative permeability to liquid flow is given by (kr)ow = krow + krw.
In case of oil/gas flow, the total relative permeability is given by (kr)go = krg + krog;
similarly, for three-phase flow, we find that (kr)gow = krg + kro + krw. Many models for
gas/oil/water relative permeabilities in three-phase flow have been published, but their
details are outside the scope of this book. In these models it is usually assumed that
the presence of gas and the presence of water have an influence on the permeability
to oil, but that the presence of gas does not influence the permeability to water and
vice versa (see, e.g., Stone 1973). The simplest model is then given by
A somewhat more sophisticated example of a three-phase relative permeability model
is the so-called modified Stone II equation, which is based on a model of flow in a
bundle of parallel tubes (Stone 1973; Aziz and Settari 1979). The relative permeability
to oil is then given by

where is the permeability at residual liquid saturation:

Eq. F-8 reduces to the two-phase expressions for the oil/gas and oil/water rel perms
in the limits Sw = Swi and Sg = Sgc, respectively.
Fig. F-4 shows an example, obtained with Eq. F-8, of the relative permeabilities
for oil in three-phase flow displayed as a ternary diagram. Note that in a ternary
diagram, the horizontal gridlines correspond to the scale at the right (gas saturation in
this case), the gridlines running from top-center to bottom-right correspond to the
scale at the left axis (water saturation), and those running from bottom-left to top-
center to the scale at the bottom axis (oil saturation). It can be seen in Fig. F-4 that
the region of “mobile saturations” is occupying only a part of the total saturation
space. The cut-off values are related to the residual or immobile saturations of the
phases. The figure also illustrates that in a large region the presence of gas and
water strongly reduces the relative permeability to oil and thus the productivity of an
oil well producing under three-phase conditions.
Fig. F-4—Relative permeability to oil under three-phase flow conditions. In this example we have used the
following parameter values: , Sgc = 0.00, Sorg = Sorw = 0.10, Swi = 0.15, and ng = nog
= now = nw = 3.
Appendix G

Answers to Questions

G-1 Answers for Chapter 1, Introduction

1.1 Performance curves of production system elements display oil or gas flow
rate on the horizontal axis vs. pressure at one end of the component on the
vertical axis. Because it is tacitly assumed that the pressure at the other end
of the component stays fixed, the performance curves therefore also
represent pressure drop vs. flow rate.
1.2 Because performance curves usually represent nonlinear relationships
between flow rate and pressure. As a result, we occasionally encounter two
intersections. However, as discussed in detail in Chapter 8, only one of these
intersection points represents a physically realistic solution.
1.3 Because (a) oil and gas are compressible, and (b) gas may dissolve in oil or
oil may vaporize into gas. All these effects result in changes in the oil and gas
phase volumes with a change in pressure, whereas the mass rates of the
(pseudo)components (both of which are present in both phases in proportions
varying with pressure) remain constant. These effects are discussed in detail
in Chapter 2.
1.4. (a) Yes; one unknown flow rate can, in theory, be determined from a single
pressure drop measurement over the tubing.
(b) Yes; see above.
(c) Yes; the unknown oil and water flow rates can be determined from a
single pressure drop measurement combined with knowledge of the
water/oil ratio.
(d) No; in this case the gas/oil ratio is unknown and therefore the three
unknown flow rates cannot be determined individually. Moreover, as
discussed in detail in Chapter 4, it is not sufficient just to know the gas/oil
ratio because the oil and gas may flow with different velocities.
1.5 Refer to Eq. 1.1 and Table 1.1. The power of the motor equals 300 × 16 =
4800 W. Because the motor is only 90% effective, the power transmitted
through the shaft equals 0.9 × 4800 = 4320 W. The shaft of the motor rotates
with 240 rpm = 240 × 2π / 60 = 25.1 rad/s. The torque is therefore 4320/25.1
= 172 N·m. The reduction gear has an efficiency of 98%, so the remaining
power available for the pump is 0.98 × 4320 = 4234 W. The pump creates a
pressure differential of 160 × 103 Pa at a flow rate of 22 × 10−3 m3/s. The
power at the liquid end of the pump is therefore 160 × 103 × 22 × 10−3 = 3520
W. The efficiency of the pump now follows as 3520/4234 × 100 % = 83%.
 1.6 Oil rate: 12 000 × 1.840 × 10−6 = 0.022 m3/s.
Gas rate: 1500 × 12 000 × 2.831 × 10−2/(24 × 3600) = 5.9 m3/s.
Oil density: 141.5 × 103/(131.5 + 38) = 835 kg/m3.
Gas density: 0.82 × 1.23 = 1.0 kg/m3.
» from_bpd_to_m3_per_s(12000) ans = 0.0221
» from_ft3_per_d_to_m3_per_s(1500*12000) ans = 5.8993
» from_deg_API_to_kg_per_m3(38) ans = 834.8083
» from_gas_grav_to_kg_per_m3(0.82) ans = 1.0086
1.7 Molecular weight of C1H4 = 16.043 and of CO2 = 44.010.
Total mass: 1 × 16.043 + 0.3 × 44.010 = 29.25 lbm.
Mass: 4.536 × 10−1 × 29.25 = 13.27 kg.
Temperature: 83 × 5/9 = 46.1 K.
Pressure: (30 + 14.7) × 6.895 × 103 = 308 kPa.
» from_lbm_to_kg(29.25) ans = 13.2676
» from_deg_R_to_K(83) ans = 46.1111
» from_psi_to_Pa(30 + 14.7) ans = 3.0820e+005
1.8 Pressure = density × gravity × depth + atmospheric pressure = (1.03 × 999)
× 9.807 × 2000 + 0.10 × 106 = 20.28 × 106 Pa. Pressure in psi: 20.18 ×
106/(6.895 × 103) = 2927 psia.
» from_kg_per_m3_to_Pa_per_m(from_liq_grav_to_kg_per_
m3(1.03)) * 2000 + 0.1
ans = 2.0181e+007
» from_Pa_to_psi(2.0181e7)
ans = 2.9270e+003
1.9 Step 1: Leave the formula in its original form (i.e., expressed in field units),
and insert the variables in SI units divided by their corresponding field-to-SI
conversion factors as given in Table A-1.

Step 2: Solve for Δp and combine all numerical factors:

Note that the dimensionless constant cg has disappeared from the equation.
1.10 γAPI = 30° => ρo = 876.16 kg/m3.
gw = 0.45 psi/ft => ρw = 1037.61 kg/m3.
zOWC = 4250 ft = 1295.40 m.
ho = 4250 – 4050 = 200 ft = 60.96 m.
pOWC = zOWC × ρw × g + patm = 1295.40 × 1037.61 × 9.81 + 100 000 = 13.29 ×
106 Pa.
pGOC = pOWC – ho × ρo × g = 13.29 × 106 – 60.96 × 876.16 × 9.81 = 12.77 ×
106 Pa.
G-2 Answers for Chapter 2, Properties of Reservoir Fluids

2.1 The solution GOR, denoted as Rs, is the ratio between the amount of
dissolved gas and the amount of oil in which the gas is dissolved (both
expressed as volume at standard conditions) at a given pressure and
temperature. The producing GOR, denoted as Rgo, is the ratio between the
gas and oil rates (both expressed as volume per unit time at standard
conditions) as observed at surface. When producing oil from a reservoir above
the bubblepoint pressure, Rgo will be identical to Rsb (i.e., to Rs at bubblepoint
pressure and reservoir temperature).
2.2 Ordinary condensation occurs when the pressure of a gas/liquid mixture
increases above the dewpoint line in the phase diagram, at a given
temperature (or, more generally, when the pressure/temperature combination
passes the dewpoint line from bottom-right to top-left). Retrograde
condensation occurs when the pressure drops below the dewpoint line (or,
more generally, when the pressure/temperature combination passes the
dewpoint line from top-right to bottom-left). This somewhat counterintuitive
effect is typical for gaseous hydrocarbon mixtures with large amounts of
volatile components (wet gas).
2.3 The EOS in the black oil model is defined in terms of three parameters: Bg,
Bo, and Rs, usually in the form of an empirical correlation. The composition is
specified in the form of the solution GOR at bubblepoint, Rsb , which gives the
ratio between the two pseudocomponents. The oil and gas pseudocomponent
properties are often specified with correlations in terms of a single parameter
per pseudocomponent, the density: ρo,sc and ρg,sc in SI units, or γo (or γAPI) and
γg in field units.
2.4 The oil phase contains a mixture of oil and gas pseudocomponents with
proportions depending on pressure and temperature. The pseudoil component
is defined as the oil phase at standard conditions.
2.5 Because critical pressures and temperatures (required to compute the
reduced values) are defined only for single components. For mixtures, we use
approximate pseudocritical pressures and temperatures to compute
pseudoreduced values.
2.6 With the aid of Eqs. 2.24 and 2.27 it follows that
ρg qg + ρo qo
 This equality implies that the mixture mass flow ρm qm = ρ g qg + ρo qo
expressed at downhole conditions is equal to the mixture mass flow expressed
at standard conditions.
2.7 In SI: qg,sc = 0.6555 m3/s; qo,sc = 0.0018 m3/s; and qw,sc = 0.0000 m3/s. With
the aid of Eq. 2.27, we now find

and therefore Rgl = 0.0009/0.0024 ≈ 0.4.

2.8 The reservoir is below the bubblepoint pressure, and therefore we can use
the Standing correlation shown in Eq. B-2 to calculate the solution GOR:

» R_s = gas_oil_rat_Standing(17e6,1.11,910,76)
R_s = 90.5408
2.9 The gas specific gravity at surface (i.e., at standard conditions), is

The gas density just above the gas cap (i.e., at reservoir conditions), can be
found with the aid of Eq. B-17. For that, we first need to calculate the gas
deviation factor Z as follows. With the aid of the Sutton correlations (Eqs. B-13
and B-14), we find the pseudocritical properties of the fluid as

ppc = 5218 × 103 – 734 × 103 × 1.11 – 16.4 × 103 × 1.112 = 4.38 × 106 Pa,
Tpc,ab s = 94.0 + 157.9 × 1.11 – 27.2 × 1.112 = 236 K.
 The pseudoreduced pressure and temperature at the reservoir follow as

With the aid of the Standing-Katz chart in Fig. B.5, we find for the
compressibility factor, Z = 0.76; and then with the aid of Eq. B-17,

» p_pc = pres_pseu_crit_Sutton(1.11)
p_pc = 4.3831e+006
» T_pc_abs = temp_pseu_crit_Sutton(1.11)
T_pc_abs = 235.7559
»p_R_pr = 17e6 / p_pc
p_R_pr = 3.8786
» T_R_pr = (76 + 273.15) / T_pc_abs
T_R_pr = 1.4810
» Z = Z_factor_DAK(p_R_pr,T_R_pr)
Z = 0.7637
» B_g = gas_form_vol_fact(17e6,76+273.15,Z)
B_g = 0.0054
» rho_g_R = 1.11 / B_g
rho_g_R = 203.9255
Alternatively, the MATLAB computations can be performed more directly as
» B_g = gas_form_vol_fact_
direct(17e6,1.11,76+273.15)
B_g = 0.0054
» rho_g_R = 1.11 / B_g
rho_g_R = 203.9255
2.10 Consult examples B-3 and B-4 in Sections B-2.10 and B-3.5 for the principles
of the hand calculation. The reservoir pressure is above the bubblepoint
pressure, and therefore the oil is undersaturated. We find for the viscosities μo
= 1.2 × 10−3 Pa·s, and μg = 24 × 10−6 Pa·s.
With MATLAB we find for the oil viscosity:
» mu_od = oil_visc_dead_B_and_R(910,76)
mu_od = 0.0070
» R_sb = gas_oil_rat_Standing(19.5e6,1.11,910,76)
R_sb = 106.6472
» mu_ob = oil_visc_sat_B_and_R(mu_od,R_sb)
mu_ob = 0.0011
» mu_o = oil_visc_undersat_V_and_B(mu_ob,22e6,19.5e6)
mu_o = 0.0012
Alternatively, the oil viscosity can be computed more directly as
» mu_o = oil_viscosity(22e6,R_sb,1.11,910,76)
mu_o = 0.0012
The MATLAB results for the gas viscosity are
» M = from_kg_per_m3_to_molar_mass(1.11)
M = 0.0261
» mu_g_p_sc = gas_visc_atm_Dempsey(M,76)
 mu_g_p_sc = 1.1148e-05
» p_pc = pres_pseu_crit_Sutton(1.11)
p_pc = 4.3831e+06
» T_pc_abs = temp_pseu_crit_Sutton(1.11)
T_pc_abs = 235.7559
» p_R_pr = 22e6 / p_pc
p_R_pr = 5.0193
» T_R_pr = (76 + 273.15) / T_pc
T_R_pr = 1.4810
» f = gas_visc_ratio_Dempsey(p_R_pr,T_R_pr)
f = 2.1619
» mu_g = f * mu_g_p_sc
mu_g = 2.4101e-05
Alternatively, also the gas viscosity can be computed directly as
» mu_g = gas_viscosity(22e6,1.11,76)
mu_g = 2.4101e-05
2.11 We can use the Vazquez and Beggs correlation (Eq. B-8) to calculate co.
However, we first need Eq. B-9 to convert the gas density ρg,sc to the density
at the reference separator pressure ρg,100, which was used to derive the
correlation in Eq. B-8. The values for psep and Tsep are, in our case, simply the
standard conditions psep = 100 kPa and Tsep = 15°C:

» R_sb = gas_oil_rat_Standing(19.5e6,1.11,910,76)
R_sb = 106.6472
» rho_g_100 = rho_g_Vazquez_and_
Beggs(100e3,1.11,910,15)
rho_g_100 = 1.0266
» c_o = compres_Vazquez_and_Beggs(22e6,R_
sb,rho_g_100,910,76)
c_o = 1.7072e-009
2.12 Consult examples B-1 and B-2 in Sections B-2.8 and B-2.9 for the principles
of the hand calculation. For p = 15 MPa and T = 85°C, the oil is saturated and
we find Bo = 1.47. For p = 30 MPa and T = 105°C, the oil is undersaturated
and we find Bo = 1.86.
Using matlab, the results are
» p_b = pres_bub_Standing(250,1.02,805,85)
p_b = 2.4529e+007
» R_s = gas_oil_rat_Standing(15e6,1.02,805,85)
R_s = 138.9552
» B_o = oil_form_vol_fact_Standing(R_s,1.02,805,85)
B_o = 1.4666
» p_b = pres_bub_Standing(250,1.02,805,105)
p_b = 2.6467e+007
 » B_ob = oil_form_vol_fact_Standing(250,1.02,805,105)
B_ob = 1.8790
» rho_g_100 = rho_g_Vazquez_and_
Beggs(100e3,1.02,805,15)
rho_g_100 = 0.8785
» c_o = compres_Vazquez_and_Beggs(30e6,250,
rho_g_100,805,105)
c_o = 3.0125e-009
» B_o = oil_form_vol_fact_undersat(B_ob,c_o,30e6,p_b)
B_o = 1.8591
Also in this case, the MATLAB computations can performed more directly as
» [B_g,B_o,R_s] = black_oil_
Standing(15e6,250,1.02,805,85)
B_g = 0.0067
B_o = 1.4666
R_s = 138.9552
» [B_g,B_o,R_s] = black_oil_
Standing(30e6,250,1.02,805,85)
B_g = 0
B_o = 1.8591
R_s = 250

G-3 Answers for Chapter 3, Single-Phase Flow in Wells and

Pipelines

3.1 The Reynolds number is a dimensionless number that quantifies the relative
importance of inertial forces relative to viscous forces in fluid flow. The
transition from laminar to turbulent flow is associated with a “critical” value of
the Reynolds number with a magnitude that depends on the geometry of the
flow conduit. For example, in pipe flow the transition occurs at values of NRe
between 2000 and 3000.
3.2 (a) Gravity; (b) Friction.
3.3 This concerns acceleration in space, not in time. When gas moves up the
well, it experiences a reduction in pressure leading to expansion (i.e., a
reduction in density). To maintain the same mass flow rate, the velocity has to
increase in proportion to the decrease in density. The increase in velocity is by
definition the acceleration and, according to Newton’s law, requires a force—
and therefore a pressure gradient—in the same direction. This may play a role
in gas flow at very high rates. In liquid flow the effect can be completely
neglected.
3.4 See Fig. 3.5. The gravity component decreases in relative importance closer
to the surface because of the decrease in gas density caused by the decrease
in pressure. Because the mass flow rate remains constant, a decrease in
density implies an increase in velocity. This results in an increase in the relative
importance of the friction component (which is strongly velocity dependent).
The acceleration component becomes of significance only close to the
 surface, where the rapid pressure drop causes a rapid gas expansion and
hence a rapid change in velocity (just visible in Fig. 3.5).
3.5 The mass balance (Eq. 3.1) now becomes

and the momentum balance (Eq. 3.2) becomes

Expanding these equations, dropping all terms higher than first order in ds, and
simplifying the results gives

3.6 The fluid velocity in the pipeline is given by

The oil viscosity can be found from the correlation in Eq. B-10 as

The Reynolds number and the dimensionless roughness follow from Eqs. 3.10
and 3.11 as

which allow us to read the friction factor from the Moody diagram in Fig. 3.3
as
 f(ε,NRe) = f(1.3 × 10−5,4.3 × 10−4)

The pipeline inclination, seen from the origin at the manifold, should be
negative to correspond to uphill production flow. Expressed in radians, the
pipeline inclination therefore becomes

The pressure at the outlet can now be computed with the aid of Eq. 3.31 as

3.7 The average absolute temperature along the well is Tav,abs = 273.15 + (120 +
30)/ 2 = 348 K. The molar mass follows from Eq. A-5 as

M = 23.55 × 10−3 ρg,sc = 23.55 × 10−3 × 0.95 = 22.4 × 10−3 Kg/mol.

The coefficients k1 and k2 are given by Eqs. 3.38 and 3.39 as

With the aid of Eq. 3.43 we now find that

which is reasonably close to the numerical result of 29.0 × 106 Pa.

3.8 The equivalent diameter is found from Eq. 3.48 as
 and the velocity in the annulus as

For the remaining steps, see the answer to Question 3.6. The numerical
results are

3.9 In line with our sign convention for wellbore flow (positive for injection and
negative for production), the MATLAB m-file pipe.m has been defined such that
the origin of the coordinate along the flowline is at the manifold. Therefore,
flow toward the manifold, as occurs in production wells, has a negative sign.
Furthermore, a negative value of the flowline inclination indicates a decreasing
flowline elevation, seen from the manifold. The m-file can be used for single-
phase gas flow, two-phase (oil/water) liquid flow, or multiphase flow,
depending on the value of the parameter fluid. In the case of single-phase oil
flow, the input for gas and water flow rates and densities may be assigned
arbitrary values. See script file answer_question_3_9.m for the numerical
implementation. The output is
p_mf = 1.2996e+005
3.10 See script file answer_question_3_10.m for the numerical implementation.
Using a tight tolerance (options = odeset(‘MaxStep’,10,’RelTol’, 1e–3) in
pipe.m, the pressure drop over the wellbore is 27.18 × 106 Pa. The error is
−27.19 Pa, which gives a relative error with an absolute value of only 1.00 ×
10−6. With the default tolerance, the absolute value of the relative error
 increases to 3.72 × 10−6.
3.11 See script file answer_question_3_11.m for the numerical implementation. The
exit pressure at the gas plant is pout = 8.97 MPa. The pressure drop over the
line is therefore Δp = pin – pout = 10.00 – 8.97 = 1.03 MPa. The pressure as a
function of distance from the plant is displayed in Fig. G-1.
3.12 See script file answer_question_3_12.m for numerical implementations using
the MATLAB routine fzero and using an alternative, user-written Newton-
Raphson algorithm. Running the file results in
>> answer_question_3_12
q_g_sc = 80.9268
q_g_sc = 80.9268
iter = 7

Fig. G-1—Pressure in the gas pipeline.

G-4 Answers for Chapter 4, Multiphase Flow in Wells and Pipelines

4.1 In a production well, the liquid holdup is typically larger than the liquid fraction
of the fluid flowing through the well because of the slip between the liquid and
gas phases.
4.2 Superficial phase velocities (i.e., superficial gas or liquid velocities) are equal
to the respective phase volume flow rate divided by the total pipe cross-
sectional area (see Eqs. 4.9 and 4.10). The in-situ phase velocities (i.e., the
local or true phase velocities) are equal to the respective phase volume flow
 rate divided by the fraction of the pipe’s cross-sectional area that is actually
occupied by that phase (see Eqs. 4.7 and 4.8). Superficial velocities are
therefore always lower than the corresponding in-situ velocities. Superficial
velocities are analogous to Darcy velocities in porous media flow. To compute
the in-situ velocities from the superficial velocities, it is necessary to first
determine the phase holdups—i.e., the relative fractions of the pipe’s cross-
sectional area that are occupied by the respective phase (see Eqs. 4.14
through 4.17).
4.3 Usually the interaction between gas and liquid in two-phase flow is too
complex to be described in detail, and the exchange of momentum between
the gas and the liquid phase is therefore typically not taken into account
explicitly. (Exceptions are stratified or annular flow, which have relatively
simple geometries, and sophisticated mechanistic models, which are outside
the scope of this book.) The usual solution is to express the mass and
momentum balances as mixture equations in terms of a mixture velocity and a
mixture density. Two additional closure equations are then specified for the
relationships between the mixture velocity and the phase velocities, and the
mixture density and the phase densities, respectively. The closure equations
are usually empirical or semiempirical.
4.4 Because they have been derived from (a) a limited number of experiments on
(b) scaled (i.e., not full-size) setups, (c) using a limited range of fluids
(sometimes not even hydrocarbons, but, e.g., water and air), (d) at pressures
and temperatures that are typically much lower than in real wells.
4.5 See Fig. 4.9. The graph has a minimum near 1 × 10−3 m3/s. To the right of the
minimum, frictional forces dominate, which increase for increasing flow rates,
just as in single-phase flow. To the left of the minimum, gravity forces
dominate, which increase for decreasing flow rates. This is because at
decreasing flow rates the liquid holdup increases as a result of an increasing
slip of the gas phase through the liquid phase.
4.6 See Fig. 4.3.

4.7 The local phase rates can be obtained from Eq. 2.31 as
 4.8 The pipe’s cross-sectional area is given by A = ¼ π d2 = 0.0117 m2. Thus we
find

4.8 (a) and (b) The FTHP is approximately equal to 17 MPa. At the bottom of
the hole, the contribution of acceleration to the pressure drop is zero. The
contributions of friction and gravity are approximately 28 – 17 = 11 MPa
and 38 – 17 = 21 MPa, respectively. The total pressure drop is therefore
11 + 21 = 32 MPa, and the FBHP follows as 17 + 32 = 49 ≈ 50 MPa, as
shown in Fig. G-2.

Fig. G-2—Pressure drop components.

(c) Note the curvature of the friction curve (slightly concave) and the gravity
curve (slightly convex) near the surface, as shown in Fig. G-3.
(d) Brill and Mukherjee, and Shi et al., because they can cope with deviated
 wells.
4.9 Read the depth corresponding to a FTHP of 5 MPa from the curve for a GOR
of 50 m3/m3. This is approximately 1200 m. Go down 3000 m (the well depth)
to 4200 m and read the FBHP. This is approximately 28 MPa.
4.10 Type help_pipe or open the file in the editor to inspect the input requirements.
Use of the Mukherjee and Brill correlation (fluid = 4) and the Standing black
oil correlations (oil = 1) gives
>> p_wf = pipe(0,0.1005,30e-6,4,1,5e6,[-50*0.01,- 0.01,0],
[0.95,850,1000],0,3000,60,60)
p_wf = 2.8160e+07
4.11 (a) In the input data of the example (line 32), replace
p_tf = 0.5e6; % FTHP, Pa
by a higher value, e.g.,
p_tf = 2.0e6; % FTHP, Pa
At increasing FTHPs the tubing intake curves shift upward because the
pressure in the entire well increases. Moreover, the curves become flatter
and more and more resemble an intake curve for single-phase flow (see
Fig. G-4). The reason is that because of the increased pressure, there is
less gas coming out of solution and the well starts increasingly to behave
as a single-phase well.
(b) In the input data of the copied file (line 32), replace
p_tf = 0.5e6; % FTHP, Pa
by
p_wf = 26e6; % FBHP, Pa
In the section for the multiphase case (lines 92 and 93), replace
p_wf = pipe (alpha,d,e,fluid,oil,p_tf,q_sc,
rho_sc,0,s_tot,T_tf,T_wf);
results(i,1:2)= [-q_o_sc p_wf];
by
p_tf = pipe (alpha,d,e,fluid,oil,p_wf,q_sc,
rho_sc,s_tot,0,T_wf,T_tf);
results(i,1:2)= [-q_o_sc p_tf];
Note: Here we have highlighted the changes by underlining them, but no
underlining is used in the actual MATLAB file.
In the plot commands (line 97) replace
ylabel(‘Flowing bottomhole pressure\it p_{wf},\ rm Pa’)
by
ylabel(‘Flowing tubinghead pressure\it p_{tf},\rm Pa’)
The tubing performance curve for pwf = 26 MPa displays a maximum
unlike the single-phase gas curve in Fig. 3.6, which decreases
monotonically. The explanation is identical to the one for the difference in
 tubing intake curves between single-phase and multiphase flow: the effect
of gas slipping through the oil at low flow rates causes liquid holdup and
therefore increases the head loss. At increasing FBHPs the tubing intake
curves shift upward because the pressure in the entire well increases
(see Fig. G-5). Moreover, the curves become flatter and increasingly
resemble an intake curve for single-phase flow, just as happened with the
tubing intake curves in Part (a) of this question.

Fig. G-3—Pressure drop components for oil with a much higher solution GOR.
Fig. G-4—Multiphase tubing intake curves for different FTHPs.

4.12 With the aid of Fig. E-1 (or Eqs. E-1 through E-3), we find that

x1 = Nµ = 0.131,

corresponds to

Nµc = y1 = 0.0074,

and with the aid of Fig. E-2 (or Eqs. E-4 through E-6), we find that

which corresponds to

ψ = y2 = 1.43.
Fig. G-5—Multiphase tubing performance curves for different FBHPs.

Finally, with the aid of Fig. E-3 (or Eqs. E-8, E-9 and E-11), it follows that

which results in a liquid holdup of

Hl = y3ψ = 0.65.

4.13 Fig. G-6 shows a flow chart that has been filled in.
4.14 (a) This equation describes a flooding curve, which is the drift velocity vd as
a function of gas holdup Hg in the Shi et al. drift flux model.
(b) C0 is the profile parameter, which represents the effect of gas bubbles
concentrating at the center of the wellbore, where the fluid velocities are
highest. A value of unity implies a flat concentration profile—a situation in
which the bubbles are equally distributed over the wellbore area. (Note:
This says nothing about the velocity profile.)
(c) At the transition to annular flow, the ratio vg/vg,fld = 1 and the gas velocity
vg is just sufficient to support a thin annulus of liquid such that it does not
flow back down along the wellbore. Although the mixture velocity vms is
 somewhat lower than vg, because of liquid holdup, the ratio vg/vg,fld at fully
developed annular flow will be much larger than unity, and so also the
ratio vms/vg,fld will exceed unity.

Fig. G-6—Flow chart for Mukherjee and Brill algorithm.

G-5 Answers for Chapter 5, Flow Through Restrictions

5.1 Because the main assumption in the derivation of the Bernoulli equation is the
occurrence of frictionless flow. This is a reasonable assumption for the
converging part of a choke, but not for the diverging part.
5.2 The maximum pressure drop is mainly caused by acceleration of the fluid—
i.e., by a conversion of potential energy into kinetic energy. A small amount of
the maximum pressure drop is caused by dissipation (i.e., by conversion of
potential energy into heat). The permanent pressure drop is entirely caused by
dissipation. A Venturi flow meter has a lower pressure drop than an orifice
meter. Moreover, there are nonintrusive meters that use different physical
principles—e.g., acoustic meters, which measure the upstream and
downstream sound propagation velocities—and do not cause a pressure drop
at all.
5.3 For isentropic flow we have
 from which follows that a decrease in pressure leads to a decrease in
temperature.
5.4 Because then downstream disturbances (in the surface facilities) cannot
propagate to the well. Such disturbances may destabilize the wellbore flow or,
in the worst case, kill the well.
5.5 Because the flow rates remain far below the speed of sound, which is
extremely high in nearly incompressible fluids.
5.6 Because the expression for critical flow over an orifice can be derived by
considering the converging part of the restriction only, for which the theory of
isentropic (or polytropic) flow is reasonably accurate (although the restriction
to ideal gases already introduces inaccuracies for hydrocarbon flow).
However, the expression for critical flow over a choke also involves the partial
pressure recovery over the diverging part, which cannot be described
accurately without a much more involved thermodynamic analysis. The usual
oilfield approach to determine the total pressure drop over the choke with the
aid of an overall discharge coefficient Cd is very approximate. Therefore the
resulting expression for critical flow is also not too accurate.
5.7 An underlying assumption of isentropic flow is that the fluid expansion or
compression happens so fast that there is no time for heat exchange with the
surroundings (i.e., the flow is adiabatic). This is an accurate assumption for
single-phase gas flow. However, in multiphase flow the presence of liquid
drops in the gas stream allows for some rapid cooling or heating of the gas (in
which case the liquid droplets are considered part of the surroundings of the
gas). Because we maintain the assumption of reversibility, the flow will be
somewhere between isentropic and isothermal, a condition known as
polytropic.
5.8 No, this is not correct for Cd. The statement does hold for , which plays a
role in quantifying the maximum pressure drop developed over the contracting
part of the choke. A value of implies frictionless (i.e., completely
reversible) flow, which is theoretically impossible. Any additional loss because
of friction results in an increase in the maximum pressure drop, which because
appears in the denominator implies a value of below unity. However, Cd,
which plays a role in quantifying the permanent pressure drop developed over
the total choke, can be larger than unity if the permanent pressure drop is
smaller than the frictionless maximum pressure drop.
5.9 See Eq. 5.11:
5.10 See Eq. 5.22:

5.11 (a) See Eq. 5.39:

Because (p3/p1) > (p3/p1)crit, the flow is noncritical.

(b) See Eq. 5.36:

5.12 (a) See Eq. 5.54:

 (b) See Eq. 5.53:

Because (p4/p1) < (p4/p1)crit, the flow is critical.

5.13 (a)/(b) See script file answer_question_5_13.m for the numerical
implementation. The result is
>> answer_question_5_13
upstream pressure from choke_multi_phase_simp: 2.5722e+06
upstream pressure from choke_multi_phase: 3.00237e+06 critical flow
boundary (p_4/p_1)_crit from choke_multi_phase: 0.544165
p_4/p_1 from choke_multi_phase: 0.399684
critical flow boundary (p_1/p_4)_crit from choke_multi_phase: 1.83768
p_1/p_4 from choke_multi_phase: 2.50198
The result from choke_multi_phase_simp.m is identical to the answer to
Question 5.12 (which is what you should have found from your hand
calculation). The upstream pressure as computed by choke_multi_phase.m is
approximately 17% higher. Moreover, the critical flow boundary (p4/p1)crit is
0.54 (= 1/1.84) instead of 0.6 (= 1/1.7). This illustrates the uncertainty
associated with multiphase choke models.
5.14 See script file answer_question_5_14.m for the numerical implementation.
(a) See Figs. G-7 and G-8. The difference in shape between the choke
 performance curve and the tubing intake curve is the presence of a
minimum in the latter and the absence of such a minimum in the former.
The minimum is caused by slip between the gas and liquid phases in the
tubing. See also the answer to Question 4.5.

Fig. G-7—Choke performance curve.

Fig. G-8—Tubing intake curve.

(b) See Fig. G-9. The choke operates in the critical regime for flow rates
above 1.35 × 10-3 m3/s. The critical pressure ratio shows a slight drop
with increasing flow rates because the downstream pressure is changing
with flow rate. This is different from the situation depicted in Fig. 5.7
because in that case the downstream pressure of the choke is kept fixed,
whereas here the downstream pressure of the flowline is kept fixed.

G-6 Answers for Chapter 6, Inflow Performance: The Basics

6.1 In the case of (nearly) incompressible single-phase liquid (oil or water) flow.
In practice the PI is often used when those conditions do not hold, in which
case it can be interpreted as a linearized version of the IPR. The range of
validity is then limited to operating conditions close to the linearization point
(i.e., the point at which the PI has been estimated or computed).
6.2 In steady-state reservoir flow, the flow rates and pressures remain constant.
In pseudosteady-state reservoir flow, the flow rates stay constant but the
pressures decrease linearly with time.
6.3 S is dimensionless.
6.4 Crossflow occurs when the closed-in pressure of a zone exceeds the closed-
in pressure of another zone plus the gravity head between the two zones.
 Usually the pressure differences between zones are the result of
disproportionate depletion—e.g., because of a large permeability contrast
between the two zones.

Fig. G-9—Critical and actual pressure ratios.

6.5 See Eq. 6.37:

from which it follows that

In field units this becomes

 6.6 See Fig G-10 for the spider plot. It can be concluded that the sensitivity to
changes in the permeability k is much larger than the sensitivity to changes in
the radii rw and re. For example, doubling the permeability results in doubling
the PI, but doubling the well radius leads only to an increase in the PI with a
factor 1.09 and doubling the external radius to a drop in the PI with a factor
0.92. The reason for this difference is the occurrence of the radii inside the
logarithm term (see Eq. 6.37).

Fig. G-10—Spider plot indicating PI sensitivity to variations in parameter values.

6.7 Starting from Eq. 6.35, an expression for the average reservoir pressure can
be derived as

where the first term between square brackets was already given in Eq. 6.25,
while the second term follows as
 Substitution of these results gives

Fig. G-11—Commingled gas production (dash-dotted line represents combined IPR for the well).

Solving for pR, substituting in Eq. 6.36, and rearranging the result then leads to

6.8 First guess for pR,av is

 Modifying Eq. 6.77 for single-phase gas flow and rewriting the result in terms
of flow rate qg,sc as a function of the equivalent drawdown gives

With the aid of Eq. (6.63) we find

This is accurate enough, and here is no need to recompute the flow rate.

6.9 When the FBHP has dropped to the bubblepoint pressure, the flow rate
follows as

The AOFP can now be computed with the aid of Eq. 6.74 as

When the FBHP drops further to 25 MPa, the flow rate follows with Eq. 6.73
as

6.10 After closing in the well, crossflow will occur (see Fig. G-11). The CBHP is
the BHP at which the inflow from the lower layer into the well just balances the
outflow from the well into the upper layer. This occurs at a BHP of
approximately 9.3 MPa, and the corresponding flow rates are approximately
0.3 m3/s. At an FBHP of 6 MPa the total inflow will be the sum of the inflow
from the two layers, which amounts to approximately 3.5 m3/s.
 6.11 See script file answer_question_6_11.m for the numerical implementation.
Running the file gives
>> answer_question_6_11
J = 2.4345e-10
J_fu = 0.9122
6.12 See script file answer_question_6_12.m for the numerical implementation.
Running the file results in the IPRs depicted in Fig. G-12. It is clear that
performing an acid treatment is far more effective than drilling a double-
diameter well. Moreover, the cost of the acid job will in most circumstances be
much lower than the additional drilling costs.

Fig. G-12—IPRs for increased values of skin S and wellbore radius rw .

6.13 See script file answer_question_6_13.m for the numerical implementation.

Running the file results in
>> answer_question_6_13
p_wf = 20000000
m_wf = 1.6268e+09
p_R = 40000000
m_R = 4.8051e+09
mu_g_ref = 1.1834e-05
B_g_ref = 1.3627
Modifying Eq. 6.77 for single-phase gas flow in terms of pseudopressures and
then rewriting the result in terms of flow rate qg,sc as a function of the
 drawdown mR − mwf = m(pR) − m(pwf) give

(See Eqs. 6.38 and 6.59 for the similarity to the radial flow case.) The small
difference between the answers to Questions 6.6 and 6.13 is caused by the
approximation of the average pressure in Question 6.6 and the incomplete
iteration. The answer to Question 6.13 is therefore more accurate. However, it
should be noted that inflow performance expressions such as Eq. 6.77 are
approximate anyway, and the small difference between the answers is
therefore probably not relevant for most applications.
6.14 See script file answer_question_6_14.m for the numerical implementation.
Running the file results in
>> answer_question_6_14
q_g_sc = 0.3279
p_wc = 9.3456e+06
q_g_sc = -3.4960

G-7 Answers for Chapter 7, Inflow Performance: Further Topics

7.1 No, that is not possible. To compute a solution to the Laplace equation (i.e.,
to compute a spatial pressure distribution), it is necessary to specify the
pressure in the reservoir at least at one point. Darcy’s law only gives us a
relationship between flow rate and pressure gradients, and does not allow us
to compute absolute pressures without having additional information. As an
aside, we note that the boundary conditions (Eqs. 7.161 and 7.162) can be
rewritten as

which implies that they both prescribe exactly the same flow rate. This is in line
with the assumption of steady-state flow such that no storage in the reservoir
can take place.
 7.2 See the explanation in Section 7.3.4. These skin factors both correspond to
an additional pressure drop caused by formation damage as defined in Eq.
7.69, but use different geometrical variables depending on their relevance to
either horizontal or vertical wells.
7.3 See the discussion in Sections 6.8 and 7.6. In general, the amounts of oil,
gas, and water entering a well vary over time and are strongly influenced by
near-well heterogeneities, the size and location of which are poorly known, if
they are known at all.
7.4 Refer to the image well approach for the four-well configuration in Section
7.2.4. Following a similar approach using just a single image well, the pressure
field for the isotropic permeability can be described as

where it has been assumed that the distance D is measured along the x-axis.
For (x, y) = (Dx/2 + rw,0) Eq. 7.26 gives the pressure in the producer in the
top-right quadrant:

The steady-state PI for a producer near a no-flow boundary therefore

becomes

For D approaching zero, the exact expression reduces to

This expression represents the PI of a well in a semicircular drainage area. It

has half the value of the PI of a well in a circular reservoir as given in Eq. 6.22.
7.5 If the well is located 50 m from the fault, we have D = 100 m. The PI then
 follows as

The PI of a well in a circular reservoir follows from Eq. 6.22 as

The PI for the well near the fault is considerably lower than the PI for the well
in the reservoir without a fault, which makes sense. Reducing the value of D to
zero leads to a PI with half the value of the PI of the well in the circular
reservoir (just as derived in the previous question).
7.6 Starting from Eq. 7.54,

the infinite sum of logarithmic terms can be rewritten as

With the aid of identities (Eqs. 7.164 through 7.166) it follows that
 The first three terms are irrelevant for our purpose because they represent a
constant pressure. As explained in Section 7.3.3, we still have to define a
reference pressure to fix the value of constant c0.
7.7 (a) The drawdown is, by definition,

Δp = pR – pwf = 5500 – 5490 = 10 psi.

The production rate follows as

(b) The production rate now reduces to

Furthermore, as evident from Eq. 7.64, we have

and therefore

The corresponding pressure drop, caused by streamline convergence,

follows as

(c) The effect of pressure drop along the horizontal well would be to reduce
the total well flow rate. Going from the toe to the heel of the well, the
well-bore pressure would gradually decrease. Therefore also the
drawdown and the specific inflow would decrease. This would,
schematically, result in the inflow profile shown in Fig. G-13.

7.8 Starting from Eq. 7.81, an expression for the average reservoir pressure can
be derived as
Fig. G-13—Varying specific inflow due to pressure drop in a horizontal well.

with which we find

and

7.9 Refer to the theory in Section 7.2.5. Starting from the approximate result
obtained in the answer to Question 7.4, the steady-state PI now becomes

7.10 (a) See Eq. 7.101, from which follows

 For the PI, see Eqs. 7.90 and 7.112:

(b) The aspect ratio is given by

which is nearly identical to Rasp,ell. For the PI, see Eqs. 7.90 and 7.93:

The PI for the elliptical reservoir is approximately 40% higher because the
ellipse has inflow from all around, whereas the box-shaped reservoir has
no-flow boundaries at the short sides. Moreover, the average distance
from the well to the boundaries, in the direction perpendicular to the well,
is less in the ellipse than in the box.
(c) The new width becomes
 The new aspect ratio and the new PI follow as

The PI for the elliptical reservoir is now approximately 17% higher.

7.11 See Eq. 7.159. With the aid of Eq. 2.31, it follows that

7.12 Modify example_point_sources.m by changing the following lines of code:

n_br = 0; % number of branches, -
N_bb = 80; % number of sources to represent back bone between
branches (incl. kick-off pts.), -
S_v = 5; % skin factor (as measured in a vertical well in the same
reservoir), -
This will produce the pressure contour plot depicted in Fig. G-14. The skin
results in a high pressure drop right at the wellbore. Because the FBHP and
the far-field reservoir pressure stay unchanged, the remaining pressure drop in
the reservoir is lower than it would be without skin, resulting in a “flat” pressure
profile with a large distance between the pressure contours. See script file
answer_question_7_12.m for the numerical implementation.
Fig. G-14—Modified version of Fig. 7.17 with increased formation-damage skin factor (Sv = 5).

G-8 Answers for Chapter 8, Well Performance

8.1 By plotting the pressure in a single (analysis) node vs. flow rate for both the
upstream and the downstream part of a system, we can determine the
operating point (i.e., the pressure in the analysis node and the flow rate
through the entire system). The underlying assumption is that the pressures at
the upstream and downstream boundaries of the system remain constant. See
also the discussion in Section 1.6.
8.2 See Fig. 8.5 and the corresponding text in Section 8.2.1.
8.3 The curvature of one or both of the performance curves may lead to two
intersections (see Fig. 8.5). The intersection corresponding to the highest flow
rate is always the physically realistic (stable) solution. Nodal analysis is based
on the assumption of steady-state flow. However, the analysis of the stability
of an operating point requires a dynamic (i.e., nonsteady-state) analysis to
establish whether small disturbances grow or shrink with time.
8.4 The practical rule is to select an operating point to the right of the minimum of
the tubing intake curve. Theoretically, an operating point to the left may still be
acceptable, but the near-parallel alignment of the curves to the left of the
minimum makes the position of the intersection sensitive to small errors. Also,
a full dynamic analysis reveals that for points to the left of the minimum, the
likelihood of unstable flow is higher than for points farther to the right.
 8.5 Because it (1) provides a higher flow rate and (2) is less prone to result in
unstable flow.
8.6 (a) Fig. 8.19: bottom of the tubing; Fig. 8.20: top of the tubing.
(b) pR = 20.5 MPa; ps = 1.2 MPa.
(c) The choke is operating in the critical regime because the FTHP is much
higher than 1.7 times the flowline pressure. Also a correct answer:
because the operating point is located in the linear part of the choke
performance curve.
(d) Because we consider steady-state flow. In that case the mass flow rate
is the same in the entire system, and therefore also the volume flow rate
expressed at identical reference conditions (i.e., standard conditions in
our case) is the same in the entire system.
8.7 (a) Reservoir pressure: 43 MPa
Bottomhole pressure: 39 MPa
Tubinghead pressure: 12 MPa
Manifold pressure: 3 MPa
(b) The pressure is decreasing, even though the flow toward the manifold is
downhill, because the pressure decrease caused by friction exceeds the
pressure increase caused by gravity. The effect is nonlinear because the
decreasing pressure leads to dissolution of gas, causing an increasing
gas flow rate and thus increasing friction. See Fig. G-15, which depicts
the detailed contributions of friction, gravity, and (zero) acceleration to the
total pressure drop over the flowline.
Fig. G-15—Pressure in the flowline with contributions from friction, gravity, and acceleration (relative to
the downstream choke pressure).

(c) The pressure just downstream from the choke is approximately 7 MPa.
The ratio between upstream and downstream pressures is therefore
approximately 12/7 ≈ 1.7, which is roughly equal to the rule-of-thumb
critical boundary, so it is not possible to decide whether the choke
operates in the critical regime.
8.8 (a) The tubing performance curve represents not only the tubing
performance but also the near-wellbore performance. The near-wellbore
performance is governed by reservoir and completion properties, not by
the tubing properties.
(b) Left side: Decreasing flow rates result in increasing slip and therefore
increasing holdup and increasing head losses. Right side: Increasing flow
rates result in increasing friction losses (and also some increased head
losses because of an increased density from compression).
(c) See Fig. G-16. Maximum: approximately 1.32 × 10–3 m3/s. Practical
minimum: approximately 0.22 × 10–3 m3/s.
(d) Not possible. The critical flow boundary is at a FTHP of approximately
1.7 times the downstream pressure (i.e., at 1.7 × 1.2 = 2.0 MPa).
(e) The critical flow boundary is now at approximately 1.7 × 0.8 = 1.36 MPa.
The corresponding flow rate is 1.15 × 10–3 m3/s. See Fig. G-17 for the
shape of the choke performance curve.
 8.9 The reservoir pressure may drop another 1.4 MPa (200 psi), as shown in Fig.
G-18.

Fig. G-16—Maximum and minimum flow rates.

Fig. G-17—Critical flow boundary and corresponding flow rate.

Fig. G-18—Effect of reduced reservoir pressure on lift requirement curve and resulting shift in operating
point to the boundary of the recommended ESP operating range.

8.10 (a) (b) See script file answer_question_8_10_part_1.m for the numerical
implementation and Fig. G-19 for the results. The oil flow rate through the
system is –2.22 × 10–3 m3/s (i.e.,–1206 B/D).
(c) No. The flow through a choke in the critical regime is completely
determined by the upstream pressure, and therefore it is not possible to
perform a nodal analysis for the complete range of flow rates starting
from the downstream end.
(d)/(e)/(f) See script file answer_question_8_10_part_2.m for the numerical
implementation and Figs. G-20 through G-22 for the results. The
main causes for the pressure drop over the well are
• Friction losses over the choke
• Gravity losses over the wellbore
• Friction losses in the reservoir
The friction losses over the flowline and the tubing are relatively small. The
friction loss over the flowline is no smaller than that over the tubing because
the average pressure in the flowline is much lower, causing more gas to come
out of solution, which increases the pressure gradient and thus counteracts the
effect of a larger-diameter tubing. This type of plot is quite useful for
debottlenecking a production system through identifying system elements with
undesirably high pressure losses.
8.11 (a) See script file answer_question_8_11.m for the numerical implementation.
The well does flow naturally because the IPR is just crossing the tubing
intake curve. However, the uncertainty in production engineering
calculations is considerable, so the well may just as well not flow.
Anyway, the flow rate is so low that artificial lift is required.
Figure G-19—Nodal analysis of the same system for four different analysis nodes, with all analysis nodes
upstream of the choke. Solid lines: upstream performance curves. Dotted lines: downstream performance
curves. (Note the shifted vertical scales in the top-left plot.)
Fig. G-20—Traverse for total pressure.

(b)/(c) See Fig. G-23 for the lift requirement curve. Note that the horizontal
axis displays the local liquid rate. The required head at the optimal pump rate
of 2.8 × 10–3 m3/s (1500 B/D) is 540 m. According to the pump curve, which is
valid for 100 stages, the available head at the optimal pump rate is 140 m (460
ft). The required number of stages is therefore

Fig. G-21—Traverse for pressure due to gravity.

Fig. G-22—Traverse for pressure without gravity effects.
Fig. G-23—Lift requirement curve.

8.12 (a) See script file answer_question_8_12.m for the numerical

implementation. The maximum achievable oil rate of qo,sc = 4.361 × 10–3
m3/s (2370 B/D) corresponds to an optimal lift gas rate of qg,sc,lift = 1.00
m3/s (3051 × 103 cf/D) and a pressure at the gas lift valve of p4 = 7.00
MPa (1,015 psi), as shown in Figs. G-24 and G-25. For a restricted
maximum lift gas rate of qg,sc,lift = 0.10 m3/s (305 × 103 cf/D), the pressure
at the gas lift valve is p4 = 9.26 MPa (1343 psi).
(b) See Fig. G-26. The required choke size is dch = 5/64 in.

Fig. G-24—Performance curves for different lift-gas rates.

Fig. G-25—Performance curves for different lift-gas rates (detail). The maximum oil flow rate is achieved
for an optimal lift-gas rate of qg,sc,lift = 1.00 m3/s (3,051 × 103 scf/D).
Fig. G-26—Required compressor discharge pressure as a function of choke setting. For an available
discharge pressure p7 = 18 MPa, the closest choke size expressed in units of 1/64 in. is dch = 5/64 in.
Appendix H

MATLAB Files

H-1 Guidance for Use

The MATLAB files listed here were programmed to illustrate the concepts discussed in
the book and to serve as a basis for questions and assignments. They can be
downloaded from the web address listed on the title page at time of printing. They
must not be used for actual production engineering calculations. For those, refer to
professional nodal analysis software packages; among the suppliers of frequently
used programs are Pipesim (2017), Prosper (2017), and WellFlo (2017).
The understanding and use of the MATLAB files requires basic programming skills,
but no advanced MATLAB knowledge. In particular, the use of compact vector notation
as offered by the MATLAB programming language has mostly been avoided, thus
somewhat sacrificing computational and notational efficiency for the benefit of
readability and educational value.
The files have been tested for all questions and assignments in the book and
should therefore run without errors. However, they may crash when used with other
parameter settings. In particular, the multiphase routines for choke flow and reservoir
flow are not robust for all combinations of parameters. Other reasons for poor
performance may be exceedance of the range of validity of the underlying
correlations or models, imperfections in the models themselves, or, of course,
programming errors. If you spot any bugs, please let us know.
If you run into problems while attempting the questions and assignments, please
check the following points, which have proved to be the most frequently occurring
source of errors:

• Use strict SI units, either the base unit or the derived units (see Section 1.2.5).
Do not use prefixes—except, of course, in kg, which is a base unit (see also
the remark in Section A-2)—or so-called allowable SI units. For example, do
not use MPa or m3/d, but use Pa and m/s instead. In the occasional input file
that uses field units, the relevant variables have been indicated with an addition
_FU.
• Use negative values for production flow rates of oil, gas, and water. Positive
values imply injection.
• Specify temperatures in °C, except in some files related to single-phase gas
flow that use absolute temperatures, indicated with Tabs and expressed in K.

H-2 Conversion Factors

from_bbl_per_ft3_to_m3_per_m3.m
from_bbl_to_m3.m
from_bpd_psi_ft_to_m2_per_d_Pa.m
from_bpd_psi_ft_to_m2_per_s_Pa.m
from_bpd_psi_to_m3_per_d_Pa.m
from_bpd_psi_to_m3_per_s_Pa.m
from_bpd_to_m3_per_d.m
from_bpd_to_m3_per_s.m
from_cal_to_J.m
from_cp_to_Pa_s.m
from_cSt_to_m2_per_s.m
from_deg_API_to_kg_per_m3.m
from_deg_API_to_liq_grav.m
from_deg_C_to_deg_F.m
from_deg_F_to_deg_C.m
from_deg_R_to_K.m
from_deg_to_rad.m
from_dyne_per_cm_to_N_per_m.m
from_ft2_to_m2.m
from_ft3_per_bbl_to_m3_per_m3.m
from_ft3_per_d_to_m3_per_d.m
from_ft3_per_d_to_m3_per_s.m
from_ft3_per_s_to_m3_per_s.m
from_ft3_per_s_to_m3_per_s.m
from_ft3_to_m3.m
from_ft_per_s_to_m_per_s.m
from_ft_to_m.m
from_gal_per_min_to_m3_per_s
from_gas_grav_to_kg_per_m3.m
from_gas_grav_to_mol_weight.m
from_gas_grav_to_molar_mass.m
from_hp_to_W.m
from_in2_to_m2.m
from_in_to_m.m
from_J_to_cal.m
from_K_to_deg_R.m
from_kg_per_m3_to_deg_API.m
from_kg_per_m3_to_gas_grav.m
from_kg_per_m3_to_lbm_per_ft3.m
from_kg_per_m3_to_liq_grav.m
from_kg_per_m3_to_mol_weight.m
from_kg_per_m3_to_molar_mass.m
from_kg_per_m3_to_Pa_per_m.m
from_kg_per_m3_to_ppg.m
from_kg_per_m3_to_psi_per_ft.m
from_kg_to_lbm.m
from_lbf_ft_to_N_m.m
from_lbf_to_N.m
from_lbm_per_ft3_to_kg_per_m3.m
from_lbm_per_ft3_to_Pa_per_m.m
from_lbm_per_ft3_to_psi_per_ft.m
from_lbm_to_kg.m
from_liq_grav_to_deg_API.m
from_liq_grav_to_kg_per_m3.m
from_m2_per_d_Pa_to_bpd_psi_ft.m
from_m2_per_s_Pa_to_bpd_psi_ft.m
from_m2_per_s_to_cSt.m
from_m2_to_ft2.m
from_m2_to_in2.m
from_m2_to_mD.m
from_m3_per_d_Pa_to_bpd_psi.m
from_m3_per_d_to_bpd.m
from_m3_per_d_to_ft3_per_d.m
from_m3_per_m3_to_bbl_per_ft3.m
from_m3_per_m3_to_ft3_per_bbl.m
from_m3_per_s_Pa_to_bpd_psi.m
from_m3_per_s_to_bpd.m
from_m3_per_s_to_ft3_per_d.m
from_m3_per_s_to_ft3_per_s.m
from_m3_to_bbl.m
from_m3_to_ft3.m
from_m_per_s_to_ft_per_s.m
 from_m_to_ft.m
from_m_to_in.m
from_mD_to_m2.m
from_mol_weight_to_gas_grav.m
from_molar_mass_to_gas_grav.m
from_mol_weight_to_kg_per_m3.m
from_molar_mass_to_kg_per_m3.m
from_N_m_to_lbf_ft.m
from_N_per_m_to_dyne_per_cm.m
from_N_to_lbf.m
from_Pa_per_m_to_kg_per_m3.m
from_Pa_per_m_to_lbm_per_ft3.m
from_Pa_per_m_to_psi_per_ft.m
from_Pa_s_to_cp.m
from_Pa_to_psi.m
from_per_Pa_to_per_psi.m
from_per_psi_to_per_Pa.m
from_ppg_to_kg_per_m3.m
from_psi_per_ft_to_kg_per_m3.m
from_psi_per_ft_to_lbm_per_ft3.m
from_psi_per_ft_to_Pa_per_m.m
from_psi_to_Pa.m
from_rad_to_deg.m
from_W_to_hp.m

H-3 Survey
deviated_well_01.txt
deviated_well_02.txt
evaluate_survey.m
example_survey.m
from_alpha_to_theta.m
from_theta_to_alpha.m
plot_survey.m

H-4 Fluid Properties

 black_oil_Glaso.m
black_oil_Standing.m
compres_Vazquez_and_Beggs.m
example_plot_black_oil_1.m
example_plot_black_oil_2.m
gas_form_vol_fact.m
gas_form_vol_fact_direct.m
gas_oil_rat_Glaso.m
gas_oil_rat_Standing.m
gas_visc_atm_Dempsey.m
gas_visc_ratio_Dempsey.m
gas_viscosity.m
interfacial_tensions.m
local_q_and_rho.m
oil_form_vol_fact_Glaso.m
oil_form_vol_fact_Standing.m
oil_form_vol_fact_undersat.m
oil_visc_dead_B_and_R.m
oil_visc_sat_B_and_R.m
oil_visc_undersat_V_and_B.m
oil_viscosity.m
pres_bub_Glaso.m
pres_bub_Standing.m
pres_bub_volatile_oil.m
pres_pseu_crit_Sutton.m
rho_g_Vazquez_and_Beggs.m
temp_pseu_crit_Sutton.m
volatile_oil.m
vol_oil_table_01 (binary file)
water_viscosity.m
Z_factor_DAK
Z_factor_DAK_direct
Z_factor_Papay

H-5 Pipe Flow

 Beggs_Brill_dpds.m
Duns_Ros_dim_less_nrs.m
example_flowline.m
example_hold_up.m
example_intake_curve.m
example_traverse.m
example_traverse_deviated.m
example_well.m
gas_dpds.m
Hag_Brown_hold_up.m
Hag_Brown_dpds.m
local_gas_liq_props.m
Moody_friction_factor.m
Muk_Brill_dpds.m
Muk_Brill_flow_regime.m
Muk_Brill_hold_up.m
oil_dpds.m
pipe.m
Reynolds_number.m
Zig_and_Syl_fric_fact.m

H-6 lift tables

Eclipse_lift_table.m
Imex_lift_table.m

H-7 Choke Flow

choke_gas.m
choke_multi_phase.m
choke_multi_phase_boundary.m
choke_multi_phase_simple.m
choke_oil.m
example_choke_performance_curve.m

H-8 Reservoir Flow

error_drdp.m
 example_gas_pseu_pres.m
example_IPR.m
example_PI_oil_hor
example_point_sources.m
example_res_pres.m
example_res_pres_oil_hor
Forchheimer.m
gas_dmdp.m
gas_pseu_pres.m
mobilities.m
rel_perms
res.m
res_dpdr.m
res_gas_dpdr.m
res_gas_simp
res_oil_dpdr.m
res_oil_box.m
res_oil_ellipse.m
res_oil_simp.m
res_Vogel.m.
saturations.m
saturations_gas_oil
saturations_oil_water
Stone_II.m

H-9 Systems
system_01.m
system_02.m
system_03_part_1.m
system_03_part_2.m

H-10 Questions
answer_question_3_9.m
answer_question_3_10.m
answer_question_3_11.m
 answer_question_3_12.m
answer_question_5_13.m
answer_question_5_14.m
answer_question_6_10.m
answer_question_6_11.m
answer_question_6_12.m
answer_question_6_13.m
answer_question_7_12.m
answer_question_8_10_part_1.m
answer_question_8_10_part_2.m
answer_question_8_11.m
answer_question_8_12.m

H-11 Graphical User Interface

These files are used by the graphical user interface for the MATLAB exercise in Chapter
1 and should not be edited.
nodal.fig
nodal.m
system_01_downstream_GUI.m
system_01_upstream_GUI.m

H-12 Demo
These executables can be used to run the first MATLAB assignment, Nodal Analysis of
a Simple System, in Chapter 1 if you do not have access to MATLAB software. You
should first install a run-time compiler by running MyAppInstaller_web.exe, which
requires Internet access and administrator rights on your computer. Thereafter you
can run the file nodal.exe.
MyAppInstaller_web.exe
nodal.exe
Nomenclature

Note: Some symbols occur more than once because they have a different meaning in
different parts of the text.

Symbol Description Dimension SI Units

a long axis of an ellipse L m
a parameter in drift flux model – –
A cross-sectional area L2 m2

A coefficient various various

b short axis of an ellipse L m
B formation volume factor – –
c coefficient – –
c compressibility L m –1 t2 1/Pa
c specific heat capacity L2 T–1 t–3 J/(kg·K)
C coefficient, constant various various
C0 profile parameter – –
d diameter L m
d direction – –
D distance L m
e pipe roughness L m
e unit vector – –
E energy L2 m t-2 J = N·m
E gas expansion factor – –
f fraction – –
f friction factor – –
f function various various
f vector-valued function – –

F force L m t–2 N
F force per unit length m t–2 N/m
F Forchheimer term – –
g function various various
g pressure gradient L–2 m t–2 Pa/m
gc gravitational constant – –
h height L m
H auxiliary variable (elliptic coordinates) – –
H holdup – –
H hydraulic head – m
i inertia L–4 m kg/m 4
I electric current q t–1
I index – –
I interval – –
J productivity index L5 m –1 t2 m 3/Pa·s
J* linearized IPR valid for low flow rates L5 m –1 t2 m 3/Pa·s
J´ specific productivity index L4 m –1 t2 m 2/Pa·s
k coefficient various various
k iteration counter – –
k permeability L2 m2
K auxiliary variable – –
K permeability tensor L2 m2

L length L m
m mass m kg
m multiplier – –
m number of components – –
m real-gas pseudopressure m Pa
ṁ mass flow rate m t–1 kg/s
M molar mass m n–1 kg/kmol
M torque L2 m t–2 N m
n coefficient – –
n Corey exponent – –
N number – –
p pressure L–1 m t–2 Pa
p´ derivative of pressure with respect to flow rate L–4 m t–1 Pa·s/m 3
P power L2 m t–3 W = Nm/s
P perimeter L m
q flow rate L3 t–1 m 3/s
q´ specific flow rate L2 t–1 m 2/s
Q heat flow rate L2 m t–3 J/s
r radial coordinate L –
r (oil/gas) ratio – –
R (gas/oil) ratio – –
R universal gas constant L2 m n-1 t2 T–1 J/(K·kmol)
s distance along well or pipeline L m
S saturation – –
S skin – –
t time t s
T temperature T °C
v specific volume L3 m –1 m 3/kg
v velocity L t–1 m/s
V electric potential L2 m q–1 t–2 V
V volume L3 m3
w width L m
x coordinate L m
x mass fraction – –
x unknown parameter – –
x vector of unknown parameters – –
y coordinate L m
z´ elevation (positive up) L m
Z gas expansion factor – –
α coefficient – –
α inclination – radians
β azimuth – radians
β Forchheimer’s coefficient L–1 1/m
β parameter in drift flux model – –
β ratio of areas – –
γ dogleg (angle) – radians
γ parameter in drift flux model – –
γ ratio of specific heats – –
γ reduction factor – –
γ (specific) gravity – –
ε local liquid/gas ratio – –
ε pipe roughness – –
η efficiency – –
η elliptic coordinate – radians
θ angle with respect to horizontal – radians
Θ auxiliary variable – –
λ fraction – –
λ mobility L3 m –1 t m 2/(Pa·s)
μ dynamic viscosity L–1 m t–1 Pa·s
ξ elliptic coordinate L m
Ρ density L–3 m kg/m 3
σ interfacial tension m t–2 N/m
ϕ porosity – –
φ wetting angle – –
Φ factor L6 t–2 s 2/m 6
ψ angle, radial coordinate – radians
ψ correction factor – –
ω angular velocity rad/s
 t–1

Subscripts
0 operating point
abs absolute
acc acceleration
act actual
ana analytical
ann annulus
asp aspect
atm atmospheric
av average
B at bubblepoint
bot bottom
bub bubble
c casing
c contraction
ch choke
conv converging
crit critical
d diameter
d discharge
d drift
dis discharge
ds downstream
D dimensionless
e expected
e external
eff effective
ell ellipse
eq equivalent
f flow
f friction
fg friction, gas
fl flowline
fld flooding
fo friction, oil
frac fracture
fric friction
fw friction, water
g gas
g Gravity
gc gas, critical
gl gas/liquid
go gas/oil
gow gas/oil/water
grav Gravity
gv gas velocity
gw gas/water
h horizontal
h hydraulic
hor horizontal
i counter
int intermediate
ip intake pressure
k kinetic
kin kinetic
Ku kutateladze
l liquid
lv liquid velocity
m mixture
max maximum
mf manifold
mn mixture, no-slip
ms mixture, slip
MB Mukherjee and Brill
num numerical
o oil
ob oil at bubblepoint
og oil/gas
opt optimal
org oil, residual, gas
orw oil, residual, water
ow oil/water
p pressure
p produced
pc pseudocritical
perm permanent
pr seudoreduced
rec recovery
red reduced
ref reference
rg relative, gas
ro relative, oil
rog relative, oil/gas
row relative, oil/water
rw relative, water
R reservoir
Re Reynolds
sc at standard conditions
s in solution (in oil)
s skin
s slip
sb in solution at bubblepoint pressure
sc standard conditions
sep separator
sg superficial, gas
sl superficial, liquid
st stages
suc suction
sw in solution in water
t total
t tubing
tf flowing tubinghead
tot total
us upstream
v valve
v velocity
v vertical
v volatilized
v volume
w water
w (molecular) weight
w well
w wet
wc closed-in wellbore
wf flowing wellbore
wi water, immobile
wo water/oil
ws static wellbore
μ liquid viscosity
μc corrected liquid viscosity
Superscripts
0 endpoint
g originating from the downhole gas phase
o originating from the downhole oil phase
Glossary

AHD along-hole depth

AIChE American Institute of Chemical Engineers
AIME American Institute of Mining and Metallurgical Engineers
AOFP absolute open-flowing potential
API American Petroleum Institute
ASME American Society of Mechanical Engineers
bbl barrel
BHP bottomhole pressure
BHT bottomhole temperature
B/D barrel(s) per day
BSW base sediment and water
CBHP closed-in bottomhole pressure
CIM Canadian Instiute of Mining, Metallurgy and Petroleum
CTHP closed-in tubinghead pressure
CGR condensate/gas ratio
DLS dogleg severity
EOS equation of state
ESP electrical submersible pump
E&P exploration and production
FDP field development plan(ning)
FBHP flowing bottomhole pressure
FBHT flowing bottomhole temperature
FTHP flowing tubinghead pressure
FTHT flowing tubinghead temperature
FVF formation volume factor
GLR gas/liquid ratio
GOC gas/oil contact
GOR gas/oil ratio
GUI graphical user interface
GWC gas/water contact
H&B Hagedorn and Brown
ICV inflow control valve
ID inside diameter
II injectivity index
IPR inflow performance relationship
LC level control
M&B Mukherjee and Brill
MD measured depth
ML multilateral
MWD measurement while drilling
OCR oil/condensate ratio
OD outside diameter
OGR oil/gas ratio
OWC oil/water contact
PC pressure control
PDG permanent downhole gauge
PI productivity index
ppg lbm per gallon
PVT pressure/volume/temperature
rpm revolutions per minute
scf standard cubic foot
SCSSV surface-controlled subsurface safety valve
SPE Society of Petroleum Engineers
STB stock-tank barrel
STB/D stock-tank barrels per day
THP tubinghead pressure
THT tubinghead temperature
TVD true vertical depth
VLE vapor/liquid equilibrium
WOR water/oil ratio
References

Al-Hussainy, R., Ramey, H. J., and Crawford, P. B. 1966. The Flow of Real Gases
through Porous Media. Journal of Petroleum Technology 18 (5): 624–636.
https://doi.org/10.2118/1243-A-PA.
Al-Safran, E. M. and Kelkar, M. 2009. Predictions of Two-Phase Critical-Flow
Boundary and Mass-Flow Rate Across Chokes. SPE Production and Operations
24 (2) 249–256. https://doi.org/10.2118/109243-PA.
Arnold, K. and Stewart, M. 1998. Surface Production Operations, second edition,
Vols. 1 and 2. Houston, TX: Gulf.
Aziz, K. and Settari, A. 1979. Petroleum Reservoir Simulation. London: Applied
Science.
Babu, D. K. and Odeh, A. S. 1988. Productivity of a Horizontal Well. SPE Reservoir
Engineering 4 (4): 417–421. https://doi.org/10.2118/18298-PA.
Bear, J. 1972. Dynamics of Fluids in Porous Media. Amsterdam: Elsevier. Reprint:
New York, NY: Dover, 1988.
Beggs, H. D. 1991. Production Optimization Using NODAL Analysis. Tulsa, OK: Oil
and Gas Consultants International Publications.
Beggs, H. D. and Brill, J. P. 1973. A Study of Two-Phase Flow in Inclined Pipes.
Journal of Petroleum Technology 25 (5): 607–617. https://doi.org/10.2118/4007-
PA.
Beggs, H. D. and Robinson, J. R. 1975. Estimating the Viscosity of Crude Oil
Systems. Journal of Petroleum Technology 27 (9): 1140–1141.
https://doi.org/10.2118/5434-PA.
Bellarby, J. 2009. Well Completion Design. Amsterdam: Elsevier.
Bieker, H. P., Slupphaug, O., and Johansen, T. A. 2007. Real-Time Production
Optimization of Oil and Gas Production Systems: A Technology Survey. SPE
Production and Operations 22 (4): 382–391. https://doi.org/10.2118/99446-PA.
Bird, R. B., Stewart, W. E., and Lightfoot, E. N. 2002. Transport Phenomena, second
edition. New York, NY: Wiley.
Boyce, W. E. and DiPrima, R. C. 2012. Elementary Differential Equations, tenth
edition. New York, NY: Wiley.
Brill, J. P. and Mukherjee, H. 1999. Multi-Phase Flow in Wells. SPE Monograph
Series 17. Richardson, TX: SPE.
Brown, K. E. 1984. The Technology of Artificial Lift Methods, Volume 4: Production
Optimization of Oil and Gas Wells by Nodal Systems Analysis. Tulsa, OK:
 PennWell.
Brown, K. E. and Lea, J. F. 1985. Nodal Systems Analysis of Oil and Gas Wells.
Journal of Petroleum Technology 37 (10): 1751–1763.
https://doi.org/10.2118/14714-PA.
Butler, R. M. 1994. Horizontal Wells for the Recovery of Oil, Gas and Bitumen.
Petroleum Society Monograph 2. Calgary, AB, Canada: Petroleum Society of the
CIM.
Carr, N. L., Kobayashi, R., and Burrows, D. B. 1954. Viscosity of Hydrocarbon
Gases under Pressure. Journal of Petroleum Technology 6 (10): 47–55.
https://doi.org/10.2118/297-G.
Carslaw, H. S. and Jaeger, J. C. 1959. Conduction of Heat in Solids, second edition.
Oxford: Clarendon Press.
Chierici, G. L. 1994. Principles of Petroleum Reservoir Engineering. Berlin: Springer.
Chilingarian, G. V., Robertson, J. O., and Kumar, S. 1987. Surface Operations in
Petroleum Production, Vols. 1 and 2. Amsterdam: Elsevier.
Chin, W. C. 2002. Quantitative Methods in Reservoir Engineering. Woburn, TX: Gulf
Professional.
Civan, F. 2000. Reservoir Formation Damage. Houston, TX: Gulf.
Colebrook, C. F. 1939. Turbulent Flow in Pipes with Particular Reference to the
Transition Region between the Smooth and Rough Pipe Laws. Journal of the
Institute of Civil Engineers 11: 133–156.
Cornelius, K. C. and Srinivas, K. 2004. Isentropic Compressible Flow for Non-Ideal
Gas Models for a Venturi. Journal of Fluids Engineering 126 (2): 238–244.
https://doi.org/10.1115/1.1677499.
Dake, L. P. 1978. Fundamentals of Reservoir Engineering. Amsterdam: Elsevier.
Danesh, A. 1998. PVT and Phase Behavior of Petroleum Reservoir Fluids.
Amsterdam: Elsevier.
Dempsey, J. R. 1965. Computer Routine Treats Gas Viscosity as a Variable. Oil and
Gas Journal, 16 August, 141–142.
Dietz, D. N. 1965. Determination of Average Reservoir Pressure from Build-Up
Surveys. Journal of Petroleum Technology 17 (18): 955–959.
https://doi.org/10.2118/1156-PA.
Dikken, B. J. 1990. Pressure Drop in Horizontal Wells and Its Effect on Production
Performance. Journal of Petroleum Technology 42 (11): 1426–1433.
https://doi.org/10.2118/19824-PA.
Dranchuk, P. M. and Abu-Kasem, J. H. 1975. Calculation of Z-Factors for Natural
Gasses Using Equations-of-State. Canadian Journal of Petroleum Technology 14
(3): 34–36. https://doi.org/10.2118/75-03-03.
Dullien, F. A. L. 1979. Porous Media—Fluid Transport and Pore Structure. New York,
 NY: Academic Press.
Duns, H. and Ros, N. C. J. 1963. Vertical Flow of Gas and Liquid Mixtures in Wells.
Proc., Sixth World Petroleum Congress, Sec. 2, 451–465, Paper 22-PD6.
Economides, M. J., Deimbacher, F. X., Brand, C. W. et al. 1991. Comprehensive
Simulation of Horizontal-Well Performance. SPE Formation Evaluation 6 (4): 418–
426. https://doi.org/10.2118/20717-PA.
Economides, M. J., Hill, A. D., Ehlig-Economides, C. et al. 2013. Petroleum
Production Systems, second edition. Upper Saddle River, NJ: Prentice Hall.
Economides, M. J., Watters, L. T., and Dunn-Norman, S. 1998. Petroleum Well
Construction. New York, NY: Wiley.
Evinger, H. H. and Muskat, M. 1942. Calculation of Productivity Factors for Oil-Gas-
Water Systems in the Steady State. Trans., AIME 146: 194–203.
Fetkovich, M. J. 1973. The Isochronal Testing of Oil Wells. Paper presented at the
SPE Fall Meeting, Las Vegas, Nevada, USA, 30 September–3 October. SPE-
4529-MS. https://doi.org/10.2118/4529-MS.
Firoozabadi, A. 1999. Thermodynamics of Hydrocarbon Reservoirs. New York, NY:
McGraw-Hill.
Forchheimer, P. 1901. Wasserbewegung durch Boden. Zeitschrift des Vereins
deutscher Ingenieure 45: 1782–1788.
Giger, F. M. 1987. Low-Permeability Reservoirs Development Using Horizontal Wells.
Paper presented at the Low Permeability Reservoirs Symposium, Denver,
Colorado, USA, 18–19 May. SPE-16406-MS. https://doi.org/10.2118/16406-MS.
Gilbert, W. E. 1954. Flowing and Gas-Lift Well Performance. API Drilling and
Production Practice 143: 127–157.
Glasø, O. 1980. Generalized Pressure/Volume/Temperature Correlations. Journal of
Petroleum Technology 32 (5): 785–795.
Golan, M. and Whitson, C. H. 1991. Well Performance, second edition. Englewood
Cliffs, NJ: Prentice Hall.
Govier, G. W. and Aziz, K. 2008. The Flow of Complex Mixtures in Pipes, second
edition. Richardson, TX: SPE.
GPSA. 1998. Engineering Data Book—FPS, eleventh edition. Tulsa, OK: Gas
Processors Suppliers Association.
Gradshteyn, I. S. and Ryzhik, I. M. 1980. Table of Integrals, Series and Products.
London: Academic Press.
Guo, B., Lyons, W. C., and Ghalambor, A. 2007. Petroleum Production Engineering:
A Computer-Assisted Approach. Houston, TX: Gulf Professional.
Hagedorn, A. R. and Brown, K. E. 1965. Experimental Study of Pressure Gradients
Occurring During Continuous Two-Phase Flow in Small-Diameter Vertical
Conduits. Journal of Petroleum Technology 17 (4): 475–484.
 https://doi.org/10.2118/940-PA.
Hagoort, J. 1988. Fundamentals of Gas Reservoir Engineering. Amsterdam:
Elsevier.
Harmathy, T. Z. 1960. Velocity of Large Drops and Bubbles in Media of Infinite or
Restricted Extent. AIChE Journal 6 (2): 281–288.
https://doi.org/10.1002/aic.690060222.
Hasan, A. R. and Kabir, C. S. 2002. Fluid Flow and Heat Transfer in Wellbores.
Richardson, TX: SPE.
Hawkins, M. F. 1956. A Note on the Skin Effect. Journal of Petroleum Technology 8
(12): 65–66. https://doi.org/10.2118/732-G.
Hill, A. D., Zhu, D., and Economides, M. J. 2008. Multilateral Wells. Richardson, TX:
SPE.
Jansen, J. D., Bosgra, O. H., and van den Hof, P. M. J. 2008. Model-Based Control
of Multiphase Flow in Subsurface Oil Reservoirs. Journal of Process Control 18
(9): 846–855. https://doi.org/10.1016/j.jprocont.2008.06.011.
Joshi, S. D. 1988. Augmentation of Well Productivity with Slant and Horizontal Wells.
Journal of Petroleum Technology 40 (6): 729–739. https://doi.org/10.2118/15375-
PA.
Joshi, S. D. 1991. Horizontal Well Technology. Tulsa, OK: PennWell.
Kamal, M. M. 2009. Transient Well Testing, SPE Monograph Series 23. Richardson,
TX: SPE.
Karnopp, D. C., Margolis, D. L., and Rosenberg, R. C. 2000. System Dynamics, third
edition. New York, NY: Wiley.
Lake, L. W., ed. 2007. Petroleum Engineering Handbook, Vols. 1–7. Richardson,
TX: SPE.
MATLAB. 2017. https://mathworks.com/help (accessed 17 March 2017).
McCain, W. D. 1990. The Properties of Petroleum Fluids, second edition. Tulsa, OK:
PennWell.
Moody, L. F. 1944. Friction Factors for Pipe Flow. Trans., ASME 66: 671–684.
Moran, M. J. and Shapiro, H. N. 1998. Fundamentals of Engineering
Thermodynamics, third edition. New York, NY: Wiley.
Morse, P. M. and Feshbach, H. 1953. Methods of Theoretical Physics, Vols. 1 and 2.
New York, NY: McGraw Hill.
Mukherjee, H. and Brill, J. P. 1983. Liquid Holdup Correlations for Inclined Two-Phase
Flow. Journal of Petroleum Technology 35 (5): 1003–1008.
https://doi.org/10.2118/10923-PA.
Mukherjee, H. and Brill, J. P. 1985a. Empirical Equations to Predict Flow Patterns in
Two-Phase Inclined Flow. International Journal of Multi-Phase Flow 11 (3): 299–
315. https://doi.org/10.1016/0301-9322(85)90060-6.
Mukherjee, H. and Brill, J. P. 1985b. Pressure Drop Correlations for Inclined Two-
Phase Flow. ASME Journal of Energy Resources Technology 107 (4): 549–554.
https://doi.org/10.1115/1.3231233.
Muskat, M. 1937. The Flow of Homogeneous Fluids through Porous Media. New
York, NY: McGraw-Hill.
Muskat, M. 1949. Physical Principles of Oil Production. New York, NY: McGraw-Hill.
Naus, M. M. J. J., Dolle, N., and Jansen, J. D. 2006. Optimization of Commingled
Production Using Infinitely Variable Inflow Control Valves. SPE Production and
Operations 21 (2): 293–301. https://doi.org/10.2118/90959-PA.
Nind, T. E. W. 1964. Principles of Oil Well Production. New York, NY: McGraw-Hill.
Nind, T. E. W. 1981. Principles of Oil Well Production, second edition. New York,
NY: McGraw-Hill.
Ouyang, L. B. and Aziz, K. 2001. A General Single-Phase Wellbore/Reservoir
Coupling Model for Multilateral Wells. SPE Reservoir Evaluation and Engineering
4 (4): 327–335. https://doi.org/10.2118/72467-PA.
Perkins, T. K. 1993. Critical and Sub-Critical Flow of Multi-Phase Mixtures through
Chokes. SPE Drilling and Completion 8 (4): 271–276.
https://doi.org/10.2118/20633-PA.
Perry, R. H., Green, D. W., and Maloney, J. O., eds. 1997. Perry’s Chemical
Engineers’ Handbook, seventh edition. New York, NY: McGraw-Hill.
Pipesim. 2017. http://www.software.slb.com/products/pipesim#js (accessed 17
March 2017).
Press, W. H., Teukolsky, S. A., Vetterling, W. T. et al. 2007. Numerical Recipes: The
Art of Scientific Computing, third edition. Cambridge: Cambridge University Press.
Prosper. 2017. http://www.petex.com/products/?ssi=3 (accessed 17 March 2017).
Ros, N. C. J. 1960. An Analysis of Critical Simultaneous Gas-Liquid Flow through a
Restriction and Its Application to Flow Metering. Applied Scientific Research 9
(1): 374–388. https://doi.org/10.1007/BF00382215.
Ros, N. C. J. 1961. Simultaneous Flow of Gas and Liquid as Encountered in Well
Tubing. Journal of Petroleum Technology 13 (10): 1037–1049.
https://doi.org/10.2118/18-PA.
Rossi, D. J., Gurpinar, O., Nelson, R. et al. 2000. Discussion on Integrating
Monitoring Data Into the Reservoir Management Process. Paper presented at the
SPE European Petroleum Conference, Paris, France, 24–25 October. SPE-
65150-MS. https://doi.org/10.2118/65150-MS.
Russell, D. G., Goodrich, J. H., Perry, G. E. et al. 1966. Methods of Predicting Gas
Well Performance. Journal of Petroleum Technology 18 (1): 99–109.
https://doi.org/10.2118/1242-PA.
Sachdeva, R., Schmidt, Z., Brill, J. P. et al. 1986. Two-Phase Flow through Chokes.
Paper presented at the SPE Annual Technical Conference and Exhibition, New
 Orleans, Louisiana, USA, 5–8 October. SPE-15657-MS.
https://doi.org/10.2118/15657-MS.
Schlichter, C. S. 1898. Theoretical Investigation of the Motion of Ground Waters. In
19th Annual Report of the U.S. Geological Survey, 1897–1898, Part 2: Papers
Chiefly of a Theoretic Nature. Washington, DC: USGS.
Schüller, R. B., Munaweera, S., Selmer-Olsen, S. et al. 2006. Critical and Sub-
Critical Oil/Gas/Water Mass Flow Rate Experiments and Predictions for Chokes.
SPE Production and Operations 21 (3): 372–380. https://doi.org/10.2118/88813-
PA.
Schüller, R. B., Solbakken, T., and Selmer-Olsen, S. 2003. Evaluation of Multi-Phase
Flow Rate Models for Chokes under Sub-Critical Oil/Gas/Water Flow Conditions.
SPE Production and Facilities 18 (3): 170–181. https://doi.org/10.2118/84961-
PA.
Shi, H., Holmes, J. A., Diaz, L. R. et al. 2005a: Drift-Flux Parameters for Three-
Phase Steady-State Flow in Wellbores. SPE Journal 10 (2): 130–137.
https://doi.org/10.2118/89836-PA.
Shi, H., Holmes, J. A., Durlofsky, L. J. et al. 2005b. Drift-Flux Modeling of Two-Phase
Flow in Wellbores. SPE Journal 10 (1): 24–33. https://doi.org/10.2118/84228-PA.
Shoham, O. 2006. Mechanistic Modeling of Gas-Liquid Two-Phase Flow in Pipes.
Richardson, TX: SPE.
SPE. 1982. The SI Metric System of Units and SPE Metric Standard. Richardson,
TX: SPE.
SPE. 2015. SPE Style Guide, 2015-2016 ed. Richardson, TX: SPE.
Standing, M. B. 1947. A Pressure-Volume-Temperature Correlation for Mixtures of
California Oils and Gases. API Drilling and Production Practice, 275–287.
Standing, M. B. 1952. Volumetric and Phase Behavior of Oil Field Hydrocarbon
Systems. New York, NY: Reinhold. Reprint: Dallas, TX: SPE, 1977.
Standing, M. B. 1971. Concerning the Calculation of Inflow Performance of Wells
Producing From Solution Gas Drive Reservoirs. Journal of Petroleum Technology
23 (9): 1141–1142. https://doi.org/10.2118/3332-PA.
Standing, M. B. and Katz, D. L. 1942. Density of Natural Gases. Trans., AIME 146:
140–149.
Stone, H. L. 1973. Estimation of Three-Phase Relative Permeability and Residual Oil
Data. Journal Canadian Petroleum Technology 12 (4): 53–61.
https://doi.org/10.2118/73-04-06.
Sutton, R. P. 1985. Compressibility Factors for High-Molecular Weight Reservoir
Gases. Paper presented at the SPE Annual Technical Conference and Exhibition,
Las Vegas, Nevada, USA, 22–26 September. SPE-14265-MS.
https://doi.org/10.2118/14265-MS.
Takacs, G. 1976. Comparisons Made for Computer Z-Factor Calculations. Oil and
 Gas Journal, 20 December, 64–66.
Taylor, H. L. and Mason, C. M. 1972. A Systematic Approach to Well Surveying
Calculations. SPE Journal 12 (6): 474–488. https://doi.org/10.2118/3362-PA.
Van Everdingen, A. F. 1953. The Skin Effect and Its Influence on the Productive
Capacity of a Well. Journal of Petroleum Technology 5 (6): 171–176.
https://doi.org/10.2118/203-G.
Vazquez, M. and Beggs, H. D. 1980. Correlations for Fluid Physical Property
Prediction. Journal of Petroleum Technology 32 (6): 968–970.
https://doi.org/10.2118/6719-PA.
Verruijt, A. 1970. Theory of Ground Water Flow. London: MacMillan.
Vogel, J. V. 1968. Inflow Performance Relationships for Solution-Gas Drive Wells.
Journal of Petroleum Technology 20 (1): 83–92. https://doi.org/10.2118/1476-PA.
Vreedenburgh, C.G.J. 1936. On the Steady Flow of Water Percolating Through Soils
With Homogeneous-Anisotropic Permeability. Proc. First International Conference
on Soil Mechanics and Foundation Engineering, 1 222-225, Cambridge,
Massachusetts, USA, 22-26 June.
Wallis, G. B. 1969. One-Dimensional Two-Phase Flow. New York, NY: McGraw-Hill.
Wallis, G. B. and Makkenchery, S. 1974. The Hanging Film Phenomenon in Vertical
Annular Two-Phase Flow. ASME Journal of Fluids Engineering 96 (3): 297–298.
https://doi.org/10.1115/1.3447155.
Walsh, M. P. and Lake, L. W. 2003. A Generalized Approach to Primary
Hydrocarbon Recovery. Amsterdam: Elsevier.
WellFlo. 2017. Weatherford. http://www.weatherford.com/doc/wft303551 (accessed
17 March 2017).
White, F. M. 2016. Fluid Mechanics, 8th ed. McGraw Hill. New York, NY: McGraw-
Hill.
Whitson, C. H. and Brulé, M. R. 2000. Phase Behavior. SPE Monograph Series 20.
Richardson, TX: SPE.
Whittaker, E. T. and Watson. G. N. 1915. A Course of Modern Analysis, second
edition. Cambridge: Cambridge University Press.
Wiggins, M. L., Russell, J. E., and Jennings, J. W. 1992. Analytical Inflow
Performance Relationships for Three-Phase Flow in Bounded Reservoirs. Paper
presented at the SPE Western Regional Meeting, Bakersfield, California, USA, 30
March–1 April. SPE-24055-MS. https://doi.org/10.2118/24055-MS.
Zaremba, W. A. 1973. Directional Survey by the Circular Arc Method. SPE Journal
13 (1): 5–11. https://doi.org/10.2118/3664-PA.
Zigrang, D. J. and Sylvester, N. D. 1985. A Review of Explicit Friction Factor
Equations. ASME Journal of Energy Resources Technology 107 (2): 280–283.
https://doi.org/10.1115/1.3231190.
Zuber, N. and Findlay, J. A. 1965. Average Volumetric Concentration in Two-Phase
Flow Systems. ASME Journal of Heat Transfer 87 (4): 453–468.
https://doi.org/10.1115/1.3689137.
 INDEX
A
absolute open-flowing potential (AOFP), 117, 137, 138
absolute permeability, 186, 291
acceleration loss, 48
along-hole depth (AHD), 45, 257, 259, 260
analysis node, 14–16, 19, 207, 208, 211, 212
anisotropic permeability
fully penetrating horizontal wells, 172–173
two-dimensional reservoir flow, 161–164
annular geometry flow, 56
AOFP. See absolute open-flowing potential (AOFP)
artificial lift, 10
ESPs, 217–219
forms, 220
gas lift, 216
average reservoir pressure, 171

B
barefoot completion, 8
base sediment and water (BSW), 28
base SI units, 3
beam pumps, 10
Bernoulli equation, 88, 89, 94, 107
binary-mixture model, 33
bisection method, 263–264
black oil, 29
correlations, 38
model, 35–36
box-shaped reservoir, 164, 165, 168
BSW. See base sediment and water (BSW)
bubblepoint line, 28, 29
bubblepoint pressure, 25
Buckley-Leverett equation, 192

C
choked flow, 85, 209
choke performance curves
multiphase flow, 102, 103, 105, 109, 110
single-phase gas flow, 91, 92, 100
well performance, 209–211, 213–215
condensate/gas ratio (CGR), 25
conformal mapping, 160
coning, 192–193
cricondenbar, 28
cricondentherm, 29
critical flow
definition, 85
multiphase flow, 101–102, 107–109
single-phase gas flow, 95–97
critical point, 28
cusping, 192–193

D
daily production control, 5
Darcy’s law, 122, 157
debottlenecking, 16
depletion flow, 119
derived SI units, 3
dewpoint line, 28
differential equations
initial value problem, 268
numerical implementation, 269–270
numerical integration, 268–269
diffusivity equation, 154
dogleg severity (DLS), 260
drift flux model, 72
dry gas, 29
reservoirs, 29
dual completion, 8

E
effort variables, 10–12
electric submersible pumps (ESPs), 10, 217–219
element equations
pressure drop analysis, in multiphase flow, 72
system models, 11–13
equations of state (EOS)
single-phase gas, 30–31
single-phase oil, 31–32
VLE, 29–30
equipotential lines, 157
ESPs. See electric submersible pumps (ESPs)
exploration and production (E&P), 4, 24, 31
extended black oil model. See volatile oil model

F
falloff, 116
FBHP. See flowing bottomhole pressure (FBHP)
field development planning
iterative processes, 4
well performance
inflow performance improvement, 212, 213
producing fields, 212
tubing and choke size, 213–215
tubing intake curve, 215–216
field units, 235–236
finite skin approach, 131
flash test, 27
flowing bottomhole pressure (FBHP), 53, 72, 73, 79–80, 202, 204, 205
flowing tubinghead pressure (FTHP), 9, 53, 72, 73, 75, 79, 201–205
flow performance tables. See lift tables
flow regimes
horizontal flow, 64
vertical flow, 65
flow variables, 10–12
fluid properties, 239–240
Forchheimer’s coefficient, 122, 132, 136, 209
formation damage, 128–129
friction loss, 46–48
fully penetrating horizontal wells
anisotropic permeability, 172–173
average reservoir pressure, 171
horizontal flow, 164–166
near-well radial flow, 166–170
semisteady-state flow, 171–172
skin factor, 170–171

G
gas compressibility factor, 31
gas condensate reservoirs, 29
gas correlations, 247–254
gas density at standard conditions, 24
gas deviation factor, 31
gas expansion factor, 27
gas formation volume factor, 25–26, 37
gas lift, 10, 216
gas/liquid ratio (GLR), 27
gas/liquid separator, 9
gas/oil ratio (GOR), 25, 27, 73, 216
general radial flow solution, 156
geometric skin factor, 170–171
Glasø correlations, 38
gradient curves, 72–73
gravity loss, 48
H
Hagedorn and Brown model, 71–72, 271–278
head loss, 45–46, 48
horizontal wells
inflow performance
fully penetrating, 164–174
multilateral wells, 180–186
multiphase flow, 186–193
partially penetrating, 174–180
schematics, 7
hydrocarbon fluids, categories of, 29

I
ICVs. See inflow control valves (ICVs)
ideal-gas law, 30
inclination, 45
inertia coefficient, 122
infinite-acting flow, 118–119
inflow control valves (ICVs), 146, 147
inflow performance
definition, 116
governing equations, 119–122
horizontal wells
fully penetrating, 164–174
multilateral wells, 180–186
multiphase flow, 186–193
partially penetrating, 174–180
importance, 116–117
multilayer performance, 140–141
multiphase flow, 136–140
single-phase gas flow, 131–136
single-phase oil flow, 122–128
skin factor, 128–131
two-dimensional reservoir flow
anisotropic permeability, 161–164
image wells, 159–161
Laplace equation, 153–155
linear superposition, 157–159
polar coordinates, 155–157
well operation and reservoir flow stages, 117–119
inflow performance relationship (IPR), 116, 202, 204, 209, 211, 216, 218
injectivity index (II), 116
interfacial tensions, 26

J
joint node, 14
Joule-Thomson cooling, 122
L
Laplace equation, 153–155, 160, 166, 176
lift requirement curve, 218
lift tables, 79–80
linear superposition, 157
liquid holdup, 63, 66

M
MATLAB assignment
choke flow, 113–114
Commingled production with a smart well, 145–151
drift flux, 83–84
horizontal well inflow and pressure drop, 195–199
hydrocarbon properties, 40–42
multiphase flow, 83–84
nodal analysis, 18–21
single-phase gas flow, 59–61
well performance, 229–232
MATLAB files
black oil correlations, 38
choke flow, 87, 91, 95, 100, 109, 345
conversion factors, 3, 340–342
demo, 346
fluid properties, 343–344
graphical user interface, 346
guidance for use, 339
horizontal well inflow performance, 174, 177, 182
lift tables, 80, 344
Moody friction factor, 48
multilateral wells, 184
pipe flow, 344
reservoir flow, 345
routines, 2–3
survey, 343
systems, 346
tubing performance, 53
measurement-while-drilling (MWD) tools, 46
mechanical interference, 128
mechanistic models, 70
microbial damage, 129
modified black oil model. See volatile oil model
Moody chart, 47
Moody friction factor, 46
Mukherjee and Brill correlation, 73, 278–283
Mukherjee and Brill model, 72
multilateral wells, 184, 186
multiphase flow
 choke flow
critical flow empirical models, 101–102
polytropic flow, 104–107
pressure drop, 109–110
sub-critical flow empirical models, 102–104
correlations
drift flux models, 284–290
Hagedorn and Brown, 271–278
Mukherjee and Brill, 278–284
dimensional analysis, 70–71
flow regimes, 63–65
inflow performance, 136–140
coning, 192–193
cusping, 192–193
gas/oil flow, 189–191
phase saturations, 188–189
pressure gradient, 187–188
primary recovery approximations, 191–192
relative permeabilities, 186–187
secondary recovery, 192
lift tables, 79–80
MATLAB assignment, 83–84
pressure drop analysis
element equations, 72
governing equations, 68–70
gradient curves, 72–73
holdup and friction correlations, 70–72
tubing intake curve, 73–79
multivariable root finding, 265–266

N
near-well radial flow
fully penetrating horizontal wells, 166–170
multilateral wells, 182–184
partially penetrating horizontal wells, 177
Newton-Raphson iteration method, 189, 220, 264–265
nodal analysis
MATLAB assignment, 18–21
multiphase flow (see multiphase flow)
performance curves, 16
principle, 14
procedure, 15–16
production optimization, 16–17
well performance (see well performance)
non-SI units, 3
notation conventions, 3
O
oil and gas production system
basic elements, 7–9
feedback control process, 6
functions, 7
oil correlations, 241–247
oil density at standard conditions, 24
oil formation volume factor, 25, 26, 37
oil/gas ratio (OGR), 25
oil models
black oil model, 35–36
compositional model, 32–33
volatile oil model, 33–35
openhole completion, 8
orifice equation, 89
orifice plate, 86

P
partially penetrating horizontal wells
elliptic coordinates, 174–176
horizontal flow, 176–177
near-well radial flow, 177
specific PI, 177–179
well models, 179–180
perforations, 8
permanent downhole gauges (PDGs), 9, 116
petroleum life cycle model, 4–5
phase changes, 129
phase saturations, 186, 188–189
PI. See productivity index (PI)
pipelines, 7
p-q graph, 15–16, 205
pressure buildup tests, 116
pressure drop analysis, in multiphase flow
element equations, 72
governing equations, 68–70
gradient curves, 72–73
holdup and friction correlations, 70–72
pressure falloff tests, 116
pressure recovery factor, 90
pressure/temperature phase diagram, 28–29
pressure transient analysis, 127
producing gas/oil ratio, 27
producing oil/gas ratio, 27
production engineering, 4–6
production optimization, 5, 16–17
productivity index (PI), 116, 117, 125, 128, 130, 137, 158–161, 168, 172–182, 196, 215
pump curve, 217–219

R
regula falsi method, 264
relative permeabilities
Corey expressions, 293–295
multiphase flow, 186–187
physics, 291–293
reservoir fluids
equations of state, 29–32
fluid properties
calculations, 37–38
standard conditions, 24–26
oil models, 32–36
pressure/temperature phase diagram, 28–29
production variables, 27–28
reservoir management, 5
residual mud solids, 128–129
restrictions, 85–87
retrograde condensation, 29
Reynolds number, 46–47

S
sand control equipment, 8
saturated oil reservoirs, 29
saturation-pressure paths, 188
scaled pipe roughness, 46–47
semiempirical models, 70
semisteady-state flow, 119
separator gas, 24
shrinkage factor, 26
single-phase fluid flow
equation of state, 44–45
friction loss, 46–48
head loss, 45–46
mass balance, 43–44
momentum balance, 44
pressure drop, 48
single-phase gas flow
governing equations, 51
inflow performance
analytical approximations, 132–136
numerical solution, 131–132
Matlab assignment, 59–61
restrictions
critical flow, 95–97
isentropic flow, 91–95
 pressure drop, 97–101
solutions, 51–52
single-phase oil flow
governing equations, 49–50
inflow performance
analytical solution, 123–125
average reservoir pressure, 125–126
semisteady-state flow, 126–127
steady-state flow, 122–123
restrictions
choke performance curve, 91, 92
permanent pressure drop, 89–91
reversible pressure drop, 87–89
solutions, 50
SI units, 3
conversion factor, 233–235
dimensional constants, 236–237
force and mass, 233, 235–236
molar mass, 236
prefixes, 233, 235
skin factor, 170–171
slip, 63, 65
solution gas/oil ratio, 25
solution oil/gas ratio, 25
specific heat capacities, 26
SPE Symbols Standard, 3
standard conditions, 12, 13, 19, 23–26, 35
standard MATLAB functions, 266–268
Standing correlations, 38, 237, 241
state variables, 24
steady-state flow, 119
steady-state path, 188
storage tanks, 7
streamlines, 157
strict SI units, 3
surface-controlled subsurface safety valve (SCSSV), 8
surface facilities, 7–9, 212, 220
surveillance, 6
system equations, 13–14
system models
element equations, 11–13
flow and effort variables, 10–12
system equations, 13–14
topology, 10–11

T
Taylor bubbles, 64
test separator, 9
thick skin approach, 131
transient flow, 118–119
transitional flow, 119
tubing, 8
tubing intake curves
multiphase flow, 73–79
single-phase fluid flow, 52–56
tubing performance curves
single-phase fluid flow, 52–55
well performance, 209, 210
turbulence coefficient, 122

U
undersaturated oil reservoirs, 29
unit systems, 3

V
valve coefficient, 91
vapor/liquid equilibrium (VLE), 29–30
vena contracta, 86
venturi, 87, 89
vertical wells
gradient curves, 73
inflow performance, 126, 153
pseudoholdup correlations, 271
schematics, 7
well performance, 201
volatile oil, 29
volatile oil model, 33–35
volatilized oil/gas ratio, 25
volume-averaged reservoir pressure, 125

W
water density at standard conditions, 25, 254
water formation volume factor, 26, 254, 255
water/oil ratio (WOR), 28
water properties, 254–255
wellbore surveying, 46
evaluation, 257–260
numerical implementation, 260–261
rectangular coordinates, 257
wellhead, 8, 91
well performance
artificial lift
ESPs, 217–219
 forms, 220
gas lift, 216
dynamic effects, 220
field development planning
inflow performance improvement, 212, 213
producing fields, 212
tubing and choke size, 213–215
tubing intake curve, 215–216
FTHP, 201–205
network topologies with loops, 220
operating point, stability of, 205–208
surface choke, 208–211
surface facilities, 220
temperature effects, 220–221
well trajectory, 45, 53, 81, 258
wet gas, 29, 37
WOR. See water/oil ratio (WOR)

Z
Z factor, 31, 249, 251–252