Processs Control in Compresoor

Advances in Industrial Control
Steve S. Niu
Process
Control
for Pumps and
Compressors
Series Editor
Michael J. Grimble, Industrial Control Centre, University of Strathclyde, Glasgow,
UK
Editorial Board
Graham Goodwin, School of Electrical Engineering and Computing, University of
Newcastle, Callaghan, NSW, Australia
Thomas J. Harris, Department of Chemical Engineering, Queen’s University,
Kingston, ON, Canada
Tong Heng Lee , Department of Electrical and Computer Engineering, National
University of Singapore, Singapore, Singapore
Om P. Malik, Schulich School of Engineering, University of Calgary, Calgary, AB,
Canada
Kim-Fung Man, City University Hong Kong, Kowloon, Hong Kong
Gustaf Olsson, Department of Industrial Electrical Engineering and Automation,
Lund Institute of Technology, Lund, Sweden
Asok Ray, Department of Mechanical Engineering, Pennsylvania State University,
University Park, PA, USA
Sebastian Engell, Lehrstuhl für Systemdynamik und Prozessführung, Technische
Universität Dortmund, Dortmund, Germany
Ikuo Yamamoto, Graduate School of Engineering, University of Nagasaki,
Nagasaki, Japan
Advances in Industrial Control is a series of monographs and contributed titles focusing on
the applications of advanced and novel control methods within applied settings. This series
has worldwide distribution to engineers, researchers and libraries.
The series promotes the exchange of information between academia and industry, to
which end the books all demonstrate some theoretical aspect of an advanced or new control
method and show how it can be applied either in a pilot plant or in some real industrial
situation. The books are distinguished by the combination of the type of theory used and the
type of application exemplified. Note that “industrial” here has a very broad interpretation; it
applies not merely to the processes employed in industrial plants but to systems such as
avionics and automotive brakes and drivetrain. This series complements the theoretical and
more mathematical approach of Communications and Control Engineering.
Indexed by SCOPUS and Engineering Index.
Proposals for this series, composed of a proposal form (please ask the in-house editor below),
a draft Contents, at least two sample chapters and an author cv (with a synopsis of the whole
project, if possible) can be submitted to either of the:
Series Editor
Professor Michael J. Grimble:
Department of Electronic and Electrical Engineering, Royal College Building, 204
George Street, Glasgow G1 1XW, United Kingdom;
e-mail: m.j.grimble@strath.ac.uk
or the
In-house Editor
Mr. Oliver Jackson:
Springer London, 4 Crinan Street, London, N1 9XW, United Kingdom;
e-mail: oliver.jackson@springer.com
Proposals are peer-reviewed.
Publishing Ethics
Researchers should conduct their research from research proposal to publication in line with
best practices and codes of conduct of relevant professional bodies and/or national and
international regulatory bodies. For more details on individual ethics matters please
see: https://www.springer.com/gp/authors-editors/journal-author/journal-author-helpdesk/
publishing-ethics/14214.
Steve S. Niu
Process Control for Pumps

and Compressors
Steve S. Niu
Idmation
Houston, TX, USA
ISSN 1430-9491 ISSN 2193-1577 (electronic)

ISBN 978-3-031-43121-0 ISBN 978-3-031-43122-7 (eBook)
https://doi.org/10.1007/978-3-031-43122-7
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Switzerland AG 2023
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Paper in this product is recyclable.

To my wife, Min, whose unwavering love,
patience, and support have been the driving
force behind my pursuit of knowledge and the
completion of this book.
Series Editor’s Foreword
Control engineering is viewed rather differently by researchers who produce general

theories and engineers who must design, calibrate, implement, and maintain indus-
trial control systems. Researchers often develop algorithms for control problems
with a well-defined mathematical basis; engineers have more immediate concerns
over the limitations of equipment, quality of control, safety, security, and system
downtime. The monograph series Advances in Industrial Control (AIC) attempts to
bridge this divide by encouraging the consideration of advanced control techniques
when they offer real benefits.
The rapid development of new control theory, techniques, and technology has an
impact on all areas of engineering. This series focuses on applications of advanced
control that may even stimulate the development of new industrial control paradigms.
This is desirable if the different aspects of the “control design” problem are to
be explored with the same dedication that “analysis and synthesis” problems have
received in the past. The series enables researchers to introduce new ideas motivated
by challenging problems in the applications of interest. It raises awareness of the
various benefits that advanced control can provide at the same time as acknowledging
the challenges that can arise.
This AIC monograph was produced by Steve S. Niu who previously contributed
to the series with the text Process Control - Engineering Analysis and Best Practices
(978-3-030-97066-6). This new volume is concerned with two of the most impor-
tant components in a process plant, namely, pumps and compressors. The topic is
introduced in Chap. 1, including an overview of equipment and some discussion of
practical aspects. Chapter 2, on the characteristics of pumps and compressors, intro-
duces topics needed to understand and model these components and covers problems
such as surge in compressors.
Chapter 3 considers the use of these components in the wider system of the
process plant and discusses their operating characteristics that are important for
control design. Chapter 4 is concerned with the control strategies that can be applied
vii
viii Series Editor’s Foreword
including so-called “Protective Control” that prevents a machine from violating oper-
ating limits. If the regulatory control fails to hold the plant at the setpoint, the protec-
tive controller should respond quickly to bring the operating point back within the
operating envelope. Anti-surge control is an example.
Chapter 5 discusses coordinate systems and various approaches to defining surge
indicators that are needed for online control, including the anti-surge control appli-
cation. Chapter 6, on basic control systems, describes practical solutions that are
tailored to the application.
Chapter 7, on advanced control solutions, includes load-balancing control for
compressor trains. The problem of complexity—a significant problem in real
systems—is discussed. Integrated control design and feedforward control are
covered, as is multi-machine optimization. A plant-wide control example is provided
involving an integrated oil-and-gas production facility. The final chapter, Chap. 8, on
the commissioning, startup, and monitoring of the process control solution, concerns
an area often neglected in more academic texts but very important in a real industrial
process.
This should be a very useful text for anyone working on real process plant
controls. It should be valuable to process plant engineers and managers, researchers
in process control, and to engineering students looking for a future in the chemicals,
petrochemicals, or pharmaceutical industries.
Glasgow, UK Michael J. Grimble

July 2023
Foreword
This is the most in-depth and insightful book on the subject I have ever read—on all
fronts and is most likely unique.
Despite the complexity of the subject (and it is very complex on many fronts!),
this book is an “easy read.” The language used is very grounded, practical, and
understandable, with no academic intellectualization. The descriptions are extremely
well written and easy to comprehend, and the diagrams are very well presented.
I really liked the approach of getting the context going with machine performance
characteristics, then zooming down to the basic control levels, then progressively
broadening the control and safeguarding scopes, combining the practical with theo-
retical, and covering not just the technical side of things, but also the approaches and
insights to the large number of side issues surrounding the discipline. These may
not be appreciated initially by students of this subject until they become engaged, at
which point the penny will drop (“Ah, that’s what Steve meant in Chapter x…”).
The balance between the specialist, mechanical, technical, process, operational,
off-normal situations, design, control, and safeguarding is great, as is highlighting the
difference between vendor and client objectives. The insights, descriptions, expla-
nations, and practical considerations given are first class—from concept through all
phases to operation. It is a great achievement! I would recommend this book.
Melbourne, Australia George Prattos

May 2023 ex FIEAust, NER, RPEQ, FSEng (TUV), MIICA;
Principal Process Control Engineer—StarOpt
Controls (retired)
ix
Preface
Industrial operations involve the continuous processing and movement of mate-

rials through interconnected piping and equipment. Material processing is primarily
performed by static equipment such as vessels, distillation columns, and chemical
reactors, while rotating equipment like motors, turbines, pumps, and compressors
handle material movements. Any critical equipment failure can cause partial or
complete plant shutdown, resulting in significant losses ranging from thousands to
millions of dollars. Furthermore, rotating equipment is a major energy consumer.
According to Stoffel (2015), Electric Motor-Driven Systems (EMDS) account for
43–46% of global electricity consumption, with compressors, pumps, and fans
consuming 39%, 19%, and 19%, respectively. Therefore, the safe and efficient oper-
ation of pumps and compressors is critical for process operation and has significant
economic incentives.
While static equipment is considered an integral part of the process configuration,
rotating equipment such as pumps and compressors have received much less attention
in the process control community. They are often viewed as specialized equipment
requiring proprietary technology implemented by dedicated specialists from equip-
ment manufacturers or their partners. As a result, the control of rotating equipment
is usually excluded from mainstream process control discussions and outside the job
scope of process control engineers. Rotating equipment control system is hosted in
dedicated hardware separate from the existing control system (e.g., DCS or PLC) and
rotating equipment control scheme is treated as a black-box component in process
control solution.
This approach has serious drawbacks. Relying on proprietary technology, hard-
ware, software, and services from third parties throughout the equipment control
solutions’ life cycle can be very costly. For example, a compressor control malfunc-
tion may require an overseas service trip by the compressor control vendor that could
take days or weeks to accomplish, resulting in multiple-week production shutdowns
or deferments. However, with an open-platform solution, a site instrument techni-
cian may pinpoint the problem to a simple instrument failure and fix it in a few
hours. The current trend is to migrate from equipment-centric solutions to process-
centric solutions and from proprietary to open platforms, which will positively impact
xi
xii Preface
equipment control. The recent acquisition of Compressor Control Company (CCC)

by Honeywell marks a significant shift away from proprietary technology and closed
systems toward open-platform solutions. This move is expected to drive innovation
and efficiency in the compressor control industry. Process control engineers will play
a crucial role in this migration.
Moreover, the equipment-centric control approach focuses on protecting the
equipment and pays much less attention to its integration with process control,
resulting in lower reliability and efficiency. For instance, centrifugal pumps and
compressors’ most crucial control requirements are capacity control and anti-surge
control. Capacity control is a requirement at the process level and anti-surge control
at the equipment level. Process control requirements dictate the capacity control
design, which interacts with anti-surge control for improved performance. The best
efficiency can only be achieved from a sound overall process control strategy, which
determines how the equipment should be controlled. A black-box approach for anti-
surge control renders this integration and optimization difficult, if not impossible. A
process-centric solution is thus required, starting with a holistic view of the process
flow and ensuring the overall stability and efficiency of the control strategy. A sound
overall control strategy provides favorable operating conditions for the equipment.
As a result, a process-centric solution tends to be more reliable and efficient than an
isolated equipment-centric solution and is typically also more operator-friendly.
Although a pump or compressor works by different principles than a distillation
column or chemical reactor, they are not more complex from the process control
perspective. The characteristic information required by control design common to
all processes and equipment, such as the supply-and-demand relationship, cause-
and-effect relationship, and dynamic response behavior, can be extracted using the
same skills and procedures as with any other complex process. Therefore, pump and
compressor control can be treated as a standard process control problem, designed
with standard technologies, and implemented on standard control system hardware
and software. Once implemented, the final control solution can be operated by the
same operators and maintained by the same maintenance team as other standard
process control solutions.
Despite the critical role of process control in pumps and compressors, books
providing systematic discussions on this topic are scarce. As noted in the preface of
Elliott and Bloch’s 2021 book on compressor technology, “… changing technology
made some books outdated, … the existing books on compressor technology did
not at all, or did not sufficiently, delve into the details of compression technology of
interest to us in 2020.” The same is true for books on compressor control.
This book provides a comprehensive guide to designing and operating integrated
control solutions for processes that involve pumps and compressors. Its goal is to
equip readers with the essential knowledge and skills from the machine characteristics
and overall control strategy to detailed design and best practices. This book is titled
Process Control for Pumps and Compressors rather than Pump and Compressor
Control to emphasize that pumps and compressors are a part of the overall process
flow and should not be treated in isolation when it comes to control.
Preface xiii
This book aims to equip readers with the following knowledge and skills for
designing and operating integrated control solutions for processes with pumps and
compressors:
• To provide readers with an understanding of the critical roles that pumps and
compressors play in the continuous operation of oil and gas, refining, petrochem-
ical, and chemical plants, as well as their dynamic behaviors and characteristics,
including the surge and choke phenomenon that can limit the machine’s operation.
• To equip readers with the ability to design and implement capacity control, anti-
surge control, safeguarding logic, and real-time monitoring solutions based on
operating requirements, dynamic supply/demand model, cause/effect relation-
ship, online and offline data, and close collaboration with other engineering
specialties.
• To enable readers to support the day-to-day operation of the in-house open-
platform compressor control applications, including the ability to troubleshoot
and fix common problems.
• To provide readers with a basic understanding of the principles of third-party
proprietary technologies to support decision-making between in-house and third-
party solutions for new applications and when vendor support is required to
troubleshoot existing third-party applications.
Pumps and compressors share fundamental operating principles, and therefore
their analysis and control methods are similar. This book takes a unified approach
to their treatment, highlighting their commonalities and differences to help readers
gain a deeper understanding of both.
This book is intended for process control engineers responsible for designing and
maintaining process control solutions for pumps and compressors. It can also serve
as a valuable resource for undergraduate process control courses to increase their
coverage on pump and compressor control. Front-line personnel such as process
engineers, rotating equipment engineers, and reliability engineers who seek to better
understand the perspective and requirements of process control will also find this
book useful.
To further aid readers’ comprehension of compressor characteristics and the
design of anti-surge control, an Excel1 software tool called CPACS (Compressor
Performance Analysis and Control Support) is provided as a companion to this book.
This convenient tool is suitable for both learning and practical applications, as most
control engineers have access to Excel on their desktops.
Considering the readers’ diverse backgrounds, this book’s content is arranged in
the following manner:
• Chapter 1: This chapter provides readers with an overview of pumps and compres-
sors and their critical role in the process industry, which is essential background
information for process control. The chapter also highlights the basic control
1 Excel® is a registered trademark of Microsoft Corporation.

xiv Preface
concepts and prevalent technologies, preparing the reader for the discussions in
later chapters.
• Chapter 2: The core theory and essential knowledge related to pump and
compressor characteristics, such as performance curves, system resistance,
minimum flow requirements, and surge and choke phenomenon, are presented
in this chapter. Due to the complex behavior of these machines and the vast infor-
mation available, it is necessary to delineate control-related knowledge from other
irrelevant information. This knowledge forms the basis for operation and control.
• Chapter 3: This chapter provides an overview of the operating requirements and
control philosophy, highlighting the typical ends and means of operation and
control. The requirements and principles apply to both manual operation and
automatic control, serving as the foundation for the overall control strategy in
Chap. 4.
• Chapter 4: The overall control strategy for pumps and compressors in the process
control context is discussed in this chapter. While the control strategy may vary
with the process configuration, the basic principles and general guidelines remain
the same, which are provided to assist with detailed design.
• Chapter 5: The description of the behavior of pumps and compressors in practical
applications differs from what are used in theoretical analysis and conceptual
understanding. This chapter provides detailed discussions on common approaches
to defining the surge indicators, which are essential for online control, especially
anti-surge control.
• Chapter 6: Several typical control schemes for simple applications involving only
a single pump or compressor are presented in this chapter. The intention is to
demonstrate how the concept and analysis presented in Chaps. 4 and 5 are applied
to real-world applications. These design examples also serve as building blocks
for advanced control schemes discussed in Chap. 7.
• Chapter 7: Several control solutions to address more complex applications that
involve multiple pumps and compressors are presented in this chapter. These
examples use the basic design schemes from Chap. 6 as building blocks and
follow the overall control strategy in Chap. 4 to address more complex application
scenarios. Given the infinite variations in process flow configuration, the focus is
on the thinking process and general methodology of analysis design.
• Chapter 8: This chapter presents practical considerations in implementing the
process control solution in the field, including the commissioning, monitoring,
and troubleshooting of the control solution. Performance monitoring and problem
troubleshooting are essential aspects of daily operations. Inferential properties and
exception-based monitoring are application areas that are receiving increasing
attention.
• Appendix A: Anti-surge control is a critical part of the process control solution,
and calculating the surge indicator is central to anti-surge control. While surge
indicator calculation can be performed manually based on the discussions in this
book, it can become extremely tedious and error prone as the problem becomes
complex with a large amount of data. Software tools are typically needed to
Preface xv
facilitate control design. This appendix demonstrates the work process of surge
indicator calculation, with the help of the companion CPACS software tool.
The first three chapters of this book require minimal background knowledge in
process control. They can be helpful for engineers in various fields, including process,
mechanical, instrumentation, and rotating equipment to gain a better understanding
of the perspective and requirements of process control.
Rotating equipment is a complex and crucial field requiring specialized engi-
neering expertise. As a process control practitioner or researcher, it is unnecessary to
have the same level of knowledge as a rotating equipment engineer. Therefore, it is
essential for process control practitioners to determine the appropriate level of under-
standing required for their work and focus their effort on pertinent information related
to process control. To this end, this book offers a balanced discussion between theory
and application, with a practical-oriented theme emphasizing essential knowledge
and best practices for practical applications. Diagrams and tables are used throughout
the book to facilitate the discussions and enhance comprehension. Nonetheless,
readers are expected to have sufficient background in several overlapping areas,
including process engineering, operations, instrumentation, mechanics, and funda-
mental process control theory and techniques. Advanced topics are identified with a
† or ‡ symbol in the section titles for easy reference.
The author would like to express sincere gratitude to Prof. Deyun Xiao of Tsinghua
University in Beijing and George Prattos of StarOpt Controls in Melbourne for
their invaluable contribution in providing insightful feedback and proofreading the
manuscript. Additionally, the author would like to acknowledge the contributions of
former colleagues from multiple teams who engaged in rigorous discussions in the
office and close collaboration in control rooms in designing and commissioning the
many compressor control applications. Finally, the author would like to thank Editor
Oliver Jackson for his unwavering dedication in guiding and shaping the manuscript
with his professional expertise.
Houston, USA Steve S. Niu

May 2023
Contents
1 Introduction to Pumps and Compressors . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Type of Pumps and Compressors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Rotating Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Pumps and Compressors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.3 Centrifugal and Reciprocating Machines . . . . . . . . . . . . . . . . 3
1.1.4 Fans and Blowers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 Applications of Pumps and Compressors . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.1 Liquid Pumping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.2 Gas Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Operation and Control of Pumps and Compressors . . . . . . . . . . . . . . 11
1.3.1 Operating Requirements and Control Objectives . . . . . . . . . . 11
1.3.2 Overall Control Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.3 Process Control Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.4 Monitoring and Safeguarding . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.4 Practical Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.4.1 Process-Centric Versus Equipment-Centric Solutions . . . . . . 20
1.4.2 In-House Versus Third-Party Solutions . . . . . . . . . . . . . . . . . . 21
1.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2 Characteristics of Pumps and Compressors . . . . . . . . . . . . . . . . . . . . . . . 25
2.1 Basic Properties of Fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1.1 Basic Principles of Fluid Dynamics . . . . . . . . . . . . . . . . . . . . . 26
2.1.2 Basic Principles of Thermodynamics . . . . . . . . . . . . . . . . . . . 27
2.1.3 Gas Compression Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1.4 Liquid Transport Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2 Description of Pumps and Compressors . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2.1 Three Variables: Speed, Head, and Flow . . . . . . . . . . . . . . . . . 33
2.2.2 Two Curves: Performance and Resistance Curves . . . . . . . . . 37
2.2.3 One Point: Operating Point . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.2.4 † Other Variables and Curves . . . . . . . . . . . . . . . . . . . . . . . . . . 43
xvii
xviii Contents
2.3 Behaviors of Dynamic Machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

2.3.1 Relationship Among Head, Flow, and Speed . . . . . . . . . . . . . 49
2.3.2 † Suction and Discharge Relationship . . . . . . . . . . . . . . . . . . . 51
2.3.3 ‡ Euler’s Equation and Slope of Performance Curve . . . . . . . 53
2.3.4 ‡ System Resistance and Slope of Resistance Curve . . . . . . . 58
2.4 Surge and Choke Phenomena . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.4.1 Surge and Choke Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.4.2 Surge Line and Choke Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3 Operating Requirements and Control Objectives . . . . . . . . . . . . . . . . . . 67
3.1 Operating Objectives and Requirements . . . . . . . . . . . . . . . . . . . . . . . 67
3.2 Maintaining the Operating Point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.2.1 Operating with Single Control Handle . . . . . . . . . . . . . . . . . . 69
3.2.2 Operation with Multiple Control Handles . . . . . . . . . . . . . . . . 73
3.2.3 † Capacity Turndown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.3 Protecting the Operating Envelope . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.3.1 Operation Under Abnormal Operating Conditions . . . . . . . . 77
3.3.2 ‡ Flow in System Resistance Components . . . . . . . . . . . . . . . 80
3.4 Transitioning Between Operating Modes . . . . . . . . . . . . . . . . . . . . . . . 82
3.4.1 Startup of Centrifugal Machines . . . . . . . . . . . . . . . . . . . . . . . . 82
3.4.2 Shutdown of Centrifugal Machines . . . . . . . . . . . . . . . . . . . . . 85
3.4.3 † Crippled Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
4 Overall Control Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4.1 Overall Control Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4.1.1 General Control Philosophy . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4.1.2 Overall Control Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.1.3 Layered and Integrated Design . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.1.4 Migration from Proprietary to Open-Platform
Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.2 Regulatory Control: Capacity Control . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.2.1 Capacity Control Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.2.2 Process Dynamics and Cause-and-Effect
Relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.2.3 Process Measurements and Controlled Variables . . . . . . . . . . 95
4.2.4 Final Control Elements and Manipulated Variables . . . . . . . . 96
4.2.5 Capacity Control Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.3 Protective Control: Anti-surge Control . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.3.1 Protective Control Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.3.2 Process Dynamics for Operating Envelope . . . . . . . . . . . . . . . 105
4.3.3 Measurements and Controlled Variables . . . . . . . . . . . . . . . . . 105
Contents xix
4.3.4 Recycle Valve and Manipulated Variables . . . . . . . . . . . . . . . 107

4.3.5 Protective Control Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . 109
4.4 Control Integration and Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . 111
4.4.1 Sequential Control: Mode Transition . . . . . . . . . . . . . . . . . . . . 112
4.4.2 Instrumented Safeguarding Against Failures . . . . . . . . . . . . . 112
4.4.3 Online Performance Monitoring . . . . . . . . . . . . . . . . . . . . . . . . 113
4.4.4 † Integration of Capacity and Anti-surge Control . . . . . . . . . 113
4.4.5 ‡ Load Balancing and Optimization . . . . . . . . . . . . . . . . . . . . . 116
4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5 Invariant Coordinates and Surge Indicators . . . . . . . . . . . . . . . . . . . . . . 121
5.1 Inlet Conditions and Invariant Coordinate Systems . . . . . . . . . . . . . . 121
5.1.1 API Datasheet for Compressor . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.1.2 † Impact of Inlet Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.1.3 Invariant Coordinate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.2 Equivalent Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
5.2.1 † Alternative Variables for Polytropic Head . . . . . . . . . . . . . . 128
5.2.2 † Alternative Variables for Volumetric Flow . . . . . . . . . . . . . . 129
5.2.3 † Equivalent Coordinate Systems . . . . . . . . . . . . . . . . . . . . . . . 130
5.3 Surge Reference Line and Surge Indicators . . . . . . . . . . . . . . . . . . . . . 132
5.3.1 Surge Reference Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
5.3.2 Surge Indicators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
5.4 Calculation of Anti-surge Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 142
5.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
6 Basic Control Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
6.1 Centrifugal Pumps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
6.1.1 Capacity Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
6.1.2 Minimum and Maximum Flow Control . . . . . . . . . . . . . . . . . . 157
6.1.3 A Complete Control Design . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
6.2 Centrifugal Compressors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
6.2.1 Capacity Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
6.2.2 Anti-surge Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6.2.3 Anti-surge Parameter Calculation . . . . . . . . . . . . . . . . . . . . . . . 172
6.2.4 † Flowmeter Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
6.2.5 Instrumented Safeguarding . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
6.2.6 A Complete Control Design . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
6.3 Reciprocating Pumps and Compressors . . . . . . . . . . . . . . . . . . . . . . . . 185
6.3.1 Capacity Control for Reciprocating Machines . . . . . . . . . . . . 185
6.3.2 Protective Control for Reciprocating Machines . . . . . . . . . . . 186
6.4 Practical Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
6.4.1 Speed of Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
6.4.2 † Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
xx Contents
6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
7 Advanced Control Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
7.1 Integration Between Capacity and Anti-surge Control . . . . . . . . . . . . 191
7.1.1 Capacity Control Revisited . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
7.1.2 An Unintegrated Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
7.1.3 Integrated Design of Capacity and Anti-surge Control . . . . . 195
7.1.4 † Integrated Design with Feedforward Compensation . . . . . 198
7.2 Load-Balancing Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
7.2.1 † Load Balancing Among Compressor Trains . . . . . . . . . . . . 200
7.2.2 ‡ Load Balancing Among Stages . . . . . . . . . . . . . . . . . . . . . . . 202
7.3 A Fully Integrated Control Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
7.3.1 A Complete Solution for Compressor Control . . . . . . . . . . . . 205
7.3.2 † Integration Between Control and Safeguarding . . . . . . . . . 208
7.3.3 A Real-World Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
7.4 Multi-Machine Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
7.4.1 † Pumps and Compressors in Parallel . . . . . . . . . . . . . . . . . . . 211
7.4.2 † Pumps and Compressors in Series . . . . . . . . . . . . . . . . . . . . 213
7.5 Implementation Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
7.5.1 Naming Convention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
7.5.2 Manual Versus Automatic Operation . . . . . . . . . . . . . . . . . . . . 216
7.5.3 † Controller Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
7.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
8 Commissioning, Startup, and Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . 219
8.1 Application Life Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
8.2 Commissioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
8.2.1 Factory Acceptance Test (FAT) . . . . . . . . . . . . . . . . . . . . . . . . . 221
8.2.2 Site-Acceptance Test (SAT) . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
8.3 Plant Startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
8.3.1 Incipient Surge Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
8.3.2 Pre-Startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
8.3.3 During Startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
8.3.4 Post-Startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
8.4 Performance Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
8.4.1 Performance Indicators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
8.4.2 Performance Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
8.5 Troubleshooting and Problem-Solving . . . . . . . . . . . . . . . . . . . . . . . . . 236
8.5.1 † Surge Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
8.5.2 † Troubleshooting of Surge Control Scheme . . . . . . . . . . . . . 238
8.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
Contents xxi
A Performance Analysis and Control Design with Software Tool . . . . . . . 241

Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Symbols and Abbreviations
Symbols
Letters
β Impeller backlean angle

η Efficiency
κ Ratio of specific heat
ω Angular velocity
ρ Density
τ Torque
A Area (m 2 )
C Constant coefficient
Cv Specific heat at constant volume
Cp Specific heat at constant pressure
D Pipe diameter
F Mass flow rate
H Head
M Molecular weight (kg/kmol); Mach number
N Rotational speed (rpm)
P Pressure (kPa)
R Gas constant
S Slope
T Temperature
U Impeller tip speed
W Power
Z Compressibility
d Orifice diameter (m)
h Fluid head in height (m)
n Polytropic exponent
r Radius
P Differential pressure
xxiii
xxiv Symbols and Abbreviations
Constants
g Gravitational Acceleration, g = 9.80665 m/s2

P0 Atmospheric pressure, 101.35 kPa
R Gas constant, R = 8.31446 J · K−1 · mol−1
T0 Absolute zero temperature, −273.15 ◦ C
Functions
cot Cotangent
ln Base-e logarithm (natural)
log Base-10 logarithm
Subscripts
s Suction side, e.g., suction pressure Ps

d Discharge side, e.g., discharge pressure Pd
v Volumetric, e.g., volumetric flow Fv
m Mass, e.g., mass flow Fm
e Equivalent, e.g., equivalent flow Fe
p Polytropic, e.g., polytropic head H P
0 Standard, e.g., standard flow F0
r Reference condition, e.g., reference pressure Pr
Fonts
Italics Used for emphasis or for special terminologies

Typewriter Used for tag names, DCS variables, or programming codes
Symbols and Abbreviations xxv
Abbreviations
AI Analog In
AO Analog Out
API American Petroleum Institute
ARWU Anti-reset windup
ASC Anti-surge Control
ASCV Anti-surge Control Valve
ASME American Society of Mechanical Engineers
ASP Anti-surge Parameter
AUTO Automatic Mode
BEP Best Efficiency Point
BHP Brake Horsepower
BPD Barrels Per Day
CAPEX Capital Expenditure
CAS Cascade Mode
CCL Capacity Control Line
CET Cause-and-effect Table
CGC Cracked Gas Compressor
COD Control Overview Diagram
CPM Control Performance Monitoring
DCS Distributed Control System
DD Detailed Design
DP Differential Pressure
DWD Deep water disposal
E&P Exploration and Production
EOS Equation of State
ESD Emergency Shutdown
FAT Factory Acceptance Test
FCCU Fluid-bed Catalytic Cracking Unit
FCE Final Control Element
FEED Front End Engineering Design
FGS Fire and Gas System
GT Gas Turbine
HMI Human–Machine Interface
HVAC Heating, Ventilation, and Air Conditioning
I/O Input/Output
IGV Inlet Guide Vane
ISA Instrument Society of America
ISO International Standard Organization
KPI Key Performance Index
LNG Liquefied Natural Gas
MAN Manual Mode
MCSF Minimum Continuous Stable Flow (of a pump)
xxvi Symbols and Abbreviations
MGI Miscible Gas Injection

MMSCMD Million Standard Cubic Meters per Day
MOC Management of Change
MOL Main Oil Line
NPSH Net Positive Suction Head (of a pump)
NPSHA Net Positive Suction Head Available (of a pump)
NPSHR Net Positive Suction Head Required (of a pump)
NRV Non-return Valve
OPEX Operational Expenditure
P&ID Piping and Instrument Diagram
PAS Process Automation System
PCN Process Control Narratives
PFD Process Flow Diagram
PGC Produced Gas Compressor
PI Proportional–Integral Algorithm
PID Proportional–Integral–Derivative Control
PLC Programmable Logic Control
PV Process Variable
RO Restriction Orifice
SAT Site-acceptance Test
SCL Surge Control Line
SIS Safety Instrument System
SLL Surge Limit Line
SOP Standard Operation Procedure
SRL Surge Reference Line
STL Surge Trip Line
VFD Variable Frequency Drive
VRU Vapor Recovery Unit
VSGB Variable-Speed Gearbox
List of Figures
Fig. 1.1 Type of rotating equipment by principle of work . . . . . . . . . . . . . 2

Fig. 1.2 Type of compressors by capacity . . . . . . . . . . . . . . . . . . . . . . . . . 3
Fig. 1.3 Reciprocating compressor and bicycle pump . . . . . . . . . . . . . . . . 4
Fig. 1.4 Example: human heart as a reciprocating machine . . . . . . . . . . . 5
Fig. 1.5 A simple pumping process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Fig. 1.6 Large pumps in pipeline operation . . . . . . . . . . . . . . . . . . . . . . . . 7
Fig. 1.7 A complete surface facility with pumps and compressors
(reproduced, with permission, from Niu and Xiao (2022)) . . . . . 7
Fig. 1.8 A typical gas compression process (reproduced,
with permission, from Niu and Xiao (2022)) . . . . . . . . . . . . . . . . 8
Fig. 1.9 A compressor system with two compressors in parallel . . . . . . . 10
Fig. 1.10 Multi-phase depletion compressors . . . . . . . . . . . . . . . . . . . . . . . . 10
Fig. 1.11 A typical ethylene plant with compressors . . . . . . . . . . . . . . . . . . 11
Fig. 1.12 Process automation system (PAS) (reproduced,
Fig. 1.13 Closed-loop control (feedback control) . . . . . . . . . . . . . . . . . . . . 14
Fig. 1.14 A process with noise and disturbance . . . . . . . . . . . . . . . . . . . . . . 15
Fig. 1.15 Process measurement components . . . . . . . . . . . . . . . . . . . . . . . . 16
Fig. 1.16 Valve characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Fig. 1.17 Control objectives and constraints (reproduced,
Fig. 1.18 On/off control and PID control . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Fig. 1.19 Structure of a PID control loop . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Fig. 1.20 Layers of process safety protections (reproduced,
Fig. 2.1 Conservation of mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Fig. 2.2 Bernoulli’s equation for flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Fig. 2.3 Gas density calculation in a flowmeter datasheet . . . . . . . . . . . . . 30
Fig. 2.4 Compression cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Fig. 2.5 Three basic variables for pump performance description . . . . . . 33
xxvii
xxviii List of Figures
Fig. 2.6 Three basic variables for compressor performance

description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Fig. 2.7 Impellers and diffusers in a centrifugal machine . . . . . . . . . . . . . 34
Fig. 2.8 Three basic variables: head, flow, and speed . . . . . . . . . . . . . . . . 35
Fig. 2.9 Head and pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Fig. 2.10 Pump performance curve at constant speed . . . . . . . . . . . . . . . . . 37
Fig. 2.11 Pump performance curves: fixed speed versus variable
speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Fig. 2.12 Compressor performance curves: fixed speed
versus variable speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Fig. 2.13 Typical pump curves (reproduced, with permission,
from Niu and Xiao (2022)) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Fig. 2.14 Typical compressor curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Fig. 2.15 Composite performance curves . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Fig. 2.16 System resistance curve and process dynamics . . . . . . . . . . . . . . 42
Fig. 2.17 Operating point on performance and resistance curves . . . . . . . . 43
Fig. 2.18 Best efficiency point and efficiency ellipse . . . . . . . . . . . . . . . . . . 46
Fig. 2.19 Cause-and-effect relationship among speed, head, and flow . . . . 49
Fig. 2.20 Flow, head, and power under affinity laws . . . . . . . . . . . . . . . . . . 51
Fig. 2.21 Impellers and Euler’s equation . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Fig. 2.22 Reverse relationship between head and flow . . . . . . . . . . . . . . . . 56
Fig. 2.23 Reverse relationship between head and flow for radial
blades . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Fig. 2.24 “Basic slope” and performance curves . . . . . . . . . . . . . . . . . . . . . 58
Fig. 2.25 System resistance and process flow components
Fig. 2.26 Flow and system resistances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Fig. 2.27 Slope of the performance curves . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Fig. 2.28 Compressor surge cycle (reproduced, with permission,
from Niu and Xiao (2022)) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Fig. 3.1 A typical compression process for upstream E&P facility
Fig. 3.2 Control handles for centrifugal compressor control . . . . . . . . . . . 69
Fig. 3.3 Capacity control with centrifugal machines . . . . . . . . . . . . . . . . . 71
Fig. 3.4 Capacity control with reciprocating machines . . . . . . . . . . . . . . . 71
Fig. 3.5 Three flows in pumps and compressors . . . . . . . . . . . . . . . . . . . . 72
Fig. 3.6 Process flow, machine flow, and recycle flow . . . . . . . . . . . . . . . . 73
Fig. 3.7 Capacity control with inlet guide vane . . . . . . . . . . . . . . . . . . . . . 74
Fig. 3.8 Capacity control with centrifugal machines . . . . . . . . . . . . . . . . . 74
Fig. 3.9 A two-speed hair dryer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Fig. 3.10 Flow turndown below surge limit . . . . . . . . . . . . . . . . . . . . . . . . . 76
Fig. 3.11 Compressor operating point and envelope . . . . . . . . . . . . . . . . . . 78
Fig. 3.12 Compressor stonewall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Fig. 3.13 Flow through an orifice plate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
List of Figures xxix
Fig. 3.14 DP versus flow in an orifice plate and head versus flow
in a compressor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Fig. 3.15 Pump startup with a throttling valve . . . . . . . . . . . . . . . . . . . . . . . 83
Fig. 3.16 Trajectory of operating point during cold startup . . . . . . . . . . . . . 84
Fig. 3.17 Trajectory of operating point during shutdown . . . . . . . . . . . . . . 85
Fig. 3.18 Process flow of a 2 × 2 compressor network . . . . . . . . . . . . . . . . 87
Fig. 4.1 Layers of process safety protections for pumps
and compressors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Fig. 4.2 Compressor operating point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Fig. 4.3 Control target and handles for centrifugal compressor
control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
Fig. 4.4 Supply-driven versus demand-driven control . . . . . . . . . . . . . . . . 99
Fig. 4.5 Capacity control of reciprocating machines . . . . . . . . . . . . . . . . . 100
Fig. 4.6 Overall control strategy for a general E and P process,
supply driven (reproduced, with permission, from Niu
and Xiao (2022)) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Fig. 4.7 Overall control strategy for a general E and P process,
demand driven (reproduced, with permission, from Niu
and Xiao (2022)) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Fig. 4.8 Supply-driven control, speed versus throttling valve . . . . . . . . . . 101
Fig. 4.9 Capacity turndown, fixed speed versus variable speed
Fig. 4.10 Compressor operating envelope . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Fig. 4.11 Sizing of recycle valve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Fig. 4.12 Recycle surge volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Fig. 4.13 Auxiliary performance lines for centrifugal machine
control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Fig. 4.14 Overall control strategy for pumps and compressors . . . . . . . . . . 112
Fig. 4.15 Compressor control as multivariable control
with interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Fig. 4.16 Unintegrated capacity control and anti-surge control . . . . . . . . . 115
Fig. 4.17 Integrated capacity control and anti-surge control . . . . . . . . . . . . 115
Fig. 4.18 Load balancing improves energy efficiency . . . . . . . . . . . . . . . . . 117
Fig. 4.19 Load balancing among parallel compressors . . . . . . . . . . . . . . . . 117
Fig. 4.20 Load balancing among compressor stages . . . . . . . . . . . . . . . . . . 118
Fig. 5.1 Compressor design conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Fig. 5.2 Operating scenarios on performance map . . . . . . . . . . . . . . . . . . 124
Fig. 5.3 Impact of gas molecular weight . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Fig. 5.4 Polytropic head versus mass flow . . . . . . . . . . . . . . . . . . . . . . . . . 126
Fig. 5.5 Polytropic head versus volumetric flow, an invariant
coordinate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Fig. 5.6 Performance curves under equivalent coordinate systems . . . . . . 132
Fig. 5.7 Surge line and surge reference line . . . . . . . . . . . . . . . . . . . . . . . . 133
Fig. 5.8 Surge reference line: flow versus flow squared . . . . . . . . . . . . . . 134
xxx List of Figures
Fig. 5.9 Slope of the operating point versus the slope of the surge
reference line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Fig. 5.10 Surge reference lines with one to four parameters . . . . . . . . . . . . 141
Fig. 5.11 Compressor performance curve . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Fig. 5.12 Surge reference lines calculated with Excel . . . . . . . . . . . . . . . . . 146
Fig. 5.13 Surge reference lines calculated with CPACS tool . . . . . . . . . . . . 147
Fig. 5.14 Example of performance map . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
Fig. 6.1 A typical process control solution with a centrifugal pump
(reproduced, with permission, from Niu and Xiao 2022) . . . . . . 152
Fig. 6.2 Performance curve for a centrifugal pump . . . . . . . . . . . . . . . . . . 152
Fig. 6.3 A typical process control solution for a centrifugal pump
with a fixed flow setpoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Fig. 6.4 Performance curve for a centrifugal fixed-speed pump . . . . . . . . 154
Fig. 6.5 A cascade control scheme for a centrifugal pump . . . . . . . . . . . . 154
Fig. 6.6 Demand-driven capacity control for a fixed-speed pump . . . . . . 155
Fig. 6.7 Supply-driven capacity control with demand override . . . . . . . . 155
Fig. 6.8 A typical process control solution with a centrifugal pump . . . . 156
Fig. 6.9 Performance curve for a centrifugal variable-speed pump . . . . . 156
Fig. 6.10 A typical process control solution with a centrifugal pump . . . . 157
Fig. 6.11 Minimum recycle flow control for centrifugal pump . . . . . . . . . . 158
Fig. 6.12 Minimum and maximum flow limits for a centrifugal pump . . . 160
with variable flow setpoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
Fig. 6.14 Minimum flow controller with a fixed ASP setpoint . . . . . . . . . . 163
Fig. 6.15 Minimum flow control and end-of-curve control
for a centrifugal pump . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Fig. 6.16 Capacity control and end-of-curve control . . . . . . . . . . . . . . . . . . 165
Fig. 6.17 A complete pump control solution . . . . . . . . . . . . . . . . . . . . . . . . 166
Fig. 6.18 A simple compression process with a single-stage
compressor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Fig. 6.19 Supply-driven capacity control with throttling valve . . . . . . . . . . 167
Fig. 6.20 Supply-driven capacity control with speed . . . . . . . . . . . . . . . . . . 167
Fig. 6.21 Demand-driven capacity control with speed . . . . . . . . . . . . . . . . 168
Fig. 6.22 Supply-driven capacity control with recycle . . . . . . . . . . . . . . . . 168
Fig. 6.23 Supply-driven capacity control with demand override . . . . . . . . 169
Fig. 6.24 Anti-surge control scheme, with flow measurement
on suction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
Fig. 6.25 Anti-surge control scheme, with flowmeter at discharge . . . . . . . 171
Fig. 6.26 Anti-surge control scheme, with differential pressure DP
for flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Fig. 6.27 Control target and handles for centrifugal compressor
control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Fig. 6.28 Essential components of a control loop . . . . . . . . . . . . . . . . . . . . 174
Fig. 6.29 The compressor API design data . . . . . . . . . . . . . . . . . . . . . . . . . . 176
Fig. 6.30 Performance map for a compressor . . . . . . . . . . . . . . . . . . . . . . . . 176
List of Figures xxxi
Fig. 6.31 Orifice-based flowmeter datasheet . . . . . . . . . . . . . . . . . . . . . . . . 179

Fig. 6.32 Compressor curves with maximum and minimum flow
lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Fig. 6.33 Anti-surge trip conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Fig. 6.34 Segregated safeguarding and control in implementation . . . . . . . 182
Fig. 6.35 Basic capacity and anti-surge control with fixed-speed drive . . . 184
Fig. 6.36 Basic capacity and anti-surge control with variable-speed
drive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
Fig. 6.37 Reciprocating machine with capacity and protective control . . . 185
Fig. 6.38 Minimizing surge volume to increase speed of response . . . . . . 187
Fig. 6.39 Surge volume with hot recycle . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Fig. 7.1 Capacity turndown for centrifugal machines . . . . . . . . . . . . . . . . 192
Fig. 7.2 Capacity turndown for reciprocating machines . . . . . . . . . . . . . . 193
Fig. 7.3 A generic 2 × 2 compression process . . . . . . . . . . . . . . . . . . . . . . 193
Fig. 7.4 A minimal control design for capacity control . . . . . . . . . . . . . . . 194
Fig. 7.5 Trajectory of operating point for a primitive capacity
control design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Fig. 7.6 An improved control design for capacity control . . . . . . . . . . . . . 196
Fig. 7.7 Trajectory of operating point with improved capacity
control design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Fig. 7.8 The split-range control in compressor capacity control . . . . . . . . 198
Fig. 7.9 An integrated capacity control solution . . . . . . . . . . . . . . . . . . . . 200
Fig. 7.10 Load of different compressors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Fig. 7.11 Load balancing for parallel compressors . . . . . . . . . . . . . . . . . . . 202
Fig. 7.12 Load balancing among compressor stages . . . . . . . . . . . . . . . . . . 203
Fig. 7.13 Load balancing among compressor stages (example) . . . . . . . . . 204
Fig. 7.14 Components in a fully integrated compressor control
solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Fig. 7.15 A complete control design for the 2 × 2 compression
process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Fig. 7.16 Process overview diagram of an E&P surface facility
(reproduced, with permission, from Niu and Xiao 2022) . . . . . . 209
Fig. 7.17 Plant-wide control: normal regulatory control (reproduced,
with permission, from Niu and Xiao 2022) . . . . . . . . . . . . . . . . . 210
Fig. 7.18 Load distribution with two compressors in parallel . . . . . . . . . . . 211
Fig. 7.19 Pressure profile with three compressors in series . . . . . . . . . . . . . 213
Fig. 8.1 Data flow in open control loop for factory acceptance test . . . . . 221
Fig. 8.2 Data flow in closed control loop for site-acceptance test . . . . . . . 223
Fig. 8.3 Startup sequence of a variable-speed compressor . . . . . . . . . . . . 231
Fig. 8.4 Startup sequence of a fixed-speed compressor . . . . . . . . . . . . . . . 231
Fig. 8.5 Control handles for centrifugal compressor control . . . . . . . . . . . 233
Fig. 8.6 Real-time monitoring of the operating point . . . . . . . . . . . . . . . . 236
Fig. A.1 Data requirements for anti-surge parameter calculation . . . . . . . 243
Fig. A.2 CPACS main menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Fig. A.3 ASP calculation menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
xxxii List of Figures
Fig. A.4 Manual digitization of compressor performance curve . . . . . . . . 248

Fig. A.5 Digitization of compressor performance curve with CPACS . . . 249
Fig. A.6 Performance curve: head versus flow . . . . . . . . . . . . . . . . . . . . . . 251
Fig. A.7 Performance curve: pressure ratio versus DP/P1 . . . . . . . . . . . . . 252
Fig. A.8 Performance curve: ratio versus DP/P1 with limits . . . . . . . . . . . 252
Fig. A.9 Pressure ratio versus equivalent flow coordinates . . . . . . . . . . . . 253
Fig. A.10 Performance curves under pressure ratio versus reduced
flow coordinate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
Fig. A.11 Performance curve with N2 gas . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
Fig. A.12 Anti-surge parameter (ASP) formula . . . . . . . . . . . . . . . . . . . . . . 255
List of Tables
Table 1.1 Comparison of centrifugal and reciprocating machines . . . . . . . 5

Table 1.2 Common complex control loops . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Table 1.3 Third-party solutions versus in-house solutions . . . . . . . . . . . . . 22
Table 2.1 Compression cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Table 2.2 Ratio of specific heat for some common gases . . . . . . . . . . . . . . 31
Table 2.3 Examples of process gas compositions . . . . . . . . . . . . . . . . . . . . 32
Table 3.1 Cause-and-effect table (CET) for pumps and compressors . . . . . 70
Table 3.2 Operating points of a hairdryer . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Table 3.3 Cause-and-effect table for a hairdryer . . . . . . . . . . . . . . . . . . . . . 75
Table 3.4 Flow turndown in a centrifugal machine . . . . . . . . . . . . . . . . . . . 77
Table 4.1 Cause-and-effect table for pump and compressor operation . . . . 95
Table 4.2 Typical control handles for pump and compressor
operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Table 4.3 Typical control strategies for pump and compressor
operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Table 4.4 Typical requirements and means of protective control . . . . . . . . 104
Table 4.5 Cause-and-effect table for protective control . . . . . . . . . . . . . . . . 105
Table 5.1 Compressor API datasheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Table 5.2 Changes in inlet conditions with constant volumetric flow . . . . . 125
Table 5.3 Potential variables for Y-axis in alternative coordinate
system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Table 5.4 Potential variables for X-axis in alternative coordinate
systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Table 5.5 Potential variables for describing head versus flow
characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Table 5.6 Example: compressor performance data . . . . . . . . . . . . . . . . . . . 143
Table 5.7 Example: compressor performance data, with compression
ratio and equivalent flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Table 5.8 Example: compressor performance data, with anti-surge
parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
Table 6.1 Minimum flow calculation for pump . . . . . . . . . . . . . . . . . . . . . . 161
xxxiii
xxxiv List of Tables
Table 6.2 Minimum flow calculation for pump, with anti-surge

parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Table 6.3 Typical process data for compressor control . . . . . . . . . . . . . . . . 173
Table 6.4 Data requirements for anti-surge control design . . . . . . . . . . . . . 174
Table 7.1 Capacity control with a split-range scheme . . . . . . . . . . . . . . . . . 197
Table 7.2 Power efficiency at different pressure ratios . . . . . . . . . . . . . . . . 215
Table 8.1 Process control deliverables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Table 8.2 Differential pressure from DP sensor and flow rate in DCS . . . . 225
Table 8.3 Flowmeter testing and calibration . . . . . . . . . . . . . . . . . . . . . . . . . 225
Table 8.4 Valve stroking from field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Table A.1 Engineering units in CPACS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
Chapter 1
Introduction to Pumps and Compressors
Rotating equipment, including pumps and compressors, plays a critical role in the
process industry by enabling material movement for continuous operation and pro-
duction. However, due to their high capital cost and energy consumption, safe and
efficient operation is essential for both operational and economic reasons. To achieve
this, it is necessary to have a thorough understanding of the process flow configuration
and operating requirements. This chapter provides an overview of pumps and com-
pressors, specifically from the process control perspective. Additionally, the chapter
introduces a process-centric control concept, an alternative approach to the tradi-
tional equipment-centric method, to better facilitate the understanding and control
of the equipment in various industrial processes.
1.1 Type of Pumps and Compressors
Pumps and compressors are essential members of the rotating equipment family.
They are in direct physical contact with the material flow they work on and thus are
of primary concern to process control.
1.1.1 Rotating Equipment
Rotating equipment is ubiquitous. They typically include, but are not limited to,
motors, engines, compressors, turbines, pumps, generators, blowers, and gearboxes,
as illustrated in Fig. 1.1. Pumps, fans, blowers, and compressors have the so-called
“wet end” with direct contact with the process fluid; Motors, engines, and gearboxes
are at the “drive end” to provide the energy for operating the “wet end.”
The prevalent examples of rotating equipment are pumps and compressors, as
shown in the shaded boxes in Fig. 1.1. Pumps and compressors impart energy to a
fluid (liquid, gas, and less commonly, solid) to cause the fluid to flow forward or raise
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 1
S. S. Niu, Process Control for Pumps and Compressors, Advances in Industrial Control,
https://doi.org/10.1007/978-3-031-43122-7_1
2 1 Introduction to Pumps and Compressors
Fig. 1.1 Type of rotating equipment by principle of work
its energy to a higher level. For clarity and convenience, we will follow the below
conventions:
• Rotating equipment. We will use the term rotating equipment to refer to all the
equipment listed in Fig. 1.1.
• Centrifugal machines. Centrifugal machines refer to centrifugal pumps, centrifugal
compressors, fans, and blowers that work on centrifugal principles.
• Centrifugal pump. We will focus our discussion of pumps on centrifugal pumps
and much less on reciprocating pumps. We refer to centrifugal pumps if a pump
is mentioned without any modifier.
• Centrifugal compressor. When a compressor is mentioned without a modifier, we
refer to a centrifugal or axial compressor. Reciprocating compressors work on
different principles and their control is relatively straightforward.
1.1.2 Pumps and Compressors
Pumps and compressors are responsible for the continuous flow of process materials,
including gas, liquid, and solids (granules) in an operating plant, from a low energy
level to a higher one or from one location to another. A pump moves an incompressible
fluid (e.g., liquid) whose volume does not (significantly) change with pressure and
temperature. A compressor moves a compressible fluid (e.g., gas) whose volume is
strongly influenced by pressure and temperature.
Pumps and compressors are extremely widely used and consume enormous energy
to operate. Stoffel (2015) reports that electric motor-driven systems (EMDS) account
for 43–46% of the global electricity consumption, of which 39% is by compressors,
19% by pumps, and 19% by fans. The various industries are responsible for 64%
1.1 Type of Pumps and Compressors 3
Fig. 1.2 Type of compressors by capacity
of the consumption. In addition, a 2001 report by the Department of Energy of the

United States claims that “… pumping systems account for nearly 20% of the world’s
electrical energy demand and range from 25 to 50% of the energy usage in certain
industrial plant operations” (DOE 2001, 2006). The reliable and efficient operation
of pumps and compressors is thus of great economic incentive.
The selection of pumps and compressors for a particular operation is determined
by many factors, among which pressure and capacity are the most important consid-
erations. Capacity refers to the amount of fluid conveyed per unit of time and can be
expressed in volumetric or mass flow rate. Figure 1.2 illustrates the operating range
of compressors in terms of pressure and flow (Elliott and Bloch 2021; Hanlon 2001;
McMillan 1983).
The mechanical and hydraulic aspects of pumps and compressors are extensively
studied and published (Boyce 2003; den Braembussche 2019; Forsthoffer 2005a, b;
Elliott and Bloch 2021; Forsthoffer 2017; Stewart 2019). Some examples of the
mechanical aspects are vibration, seal compatibility, bearing selection, casing con-
figuration, metallurgical suitability, and radial thrust and shaft deflection, which are
of little interest to process control and will not be discussed. Our discussions in this
book will focus exclusively on the control aspects of the equipment, including the
variables describing the machine behaviors, the means of influencing the machine
behavior, and the dynamic cause-and-effect relationship between the variables.
1.1.3 Centrifugal and Reciprocating Machines
The two main types of pumps and compressors are positive displacement and
dynamic. A positive displacement machine pressurizes the liquid by confinement.
Fig. 1.3 Reciprocating compressor and bicycle pump
It increases the fluid pressure by operating on a fixed volume in a confined space.

Reciprocating, sliding-vane, rotary lobe, and screw machines are examples of pos-
itive displacement machines. In contrast, a dynamic machine pressurizes the fluid
by acceleration. It increases the pressure of the fluid by using blades to increase the
velocity. Typical dynamic machines are axial and centrifugal machines.
A centrifugal machine consists of a stationary inlet casing, a rotating impeller, and
several fixed diverging passages. It uses rotating vanes or bladed discs to sequentially
accelerate the fluids (gas or liquid) to increase their kinetic energy and then decelerate
them to trade a portion of the kinetic energy for potential energy (See Fig. 1.3a).
Standard centrifugal machines include centrifugal pumps and compressors, electric
blowers, and fans, as shown in Fig. 1.1.
Axial machines behave similarly to centrifugal machines, although they differ in
applicability.1 A jet engine is an example of a complex axial compressor. The internal
flow is in the axial direction, parallel to the shaft, as opposed to the radial exit direc-
tion in centrifugal machines. The dynamic characteristics of axial and centrifugal
machines are very similar; thus, control analysis and design for centrifugal machines
apply to axial machines almost without change.
Reciprocating machines work on the same principle as bicycle pumps. A human
heart is probably the most sophisticated reciprocating pump operating at variable
speeds (Fig. 1.4). A fixed volume of gas or liquid is moved from a low-pressure to
a high-pressure location on each stroke. The reciprocating machine is insensitive
to system changes. It operates at approximately the same delivered volume flow
1Axial flow compressors are more suitable for larger engines with smaller frontal areas (and drag)
and can achieve 3–4% higher efficiency for the same duty than centrifugal compressors. For very
small compressors with low-flow rates, the efficiency of axial compressors drops sharply, and
blading is small and difficult to make accurately. Centrifugal machines are best suited for constant-
pressure and variable-flow applications. Axial compressors are more suitable for constant-flow and
variable-pressure applications.
1.1 Type of Pumps and Compressors 5
"Non-Return Valves" "Non-Return Valves"
"Pump" "Pump"
Fig. 1.4 Example: human heart as a reciprocating machine
regardless of the fluid type or operating condition; see Fig. 1.3b for an illustration.
Due to the reciprocating action, the flow is intermittent and usually requires a damper
vessel downstream to minimize pulsation if a smooth flow is desired.
Centrifugal machines have fewer moving parts than reciprocating machines and
offer many advantages, such as simple construction, relatively low cost, low mainte-
nance, quiet operation, and excellent reliability. They are often preferred over other
machine types. On the other hand, the operation and control of centrifugal machines
are more complex due to the surge and choke limits, which is a central topic of this
book. Reciprocating machines have more moving parts than centrifugal machines
and are typically less reliable; however, their operation and control are much more
straightforward (see Table 1.1 for a summary).
Table 1.1 Comparison of centrifugal and reciprocating machines

Type Advantages Disadvantages
Centrifugal machine Wide operating range Surge at low flow
Low maintenance Moderate efficiency
High reliability
Axial machine High efficiency Low pressure ratio
High flow Expensive blading
High-speed capacity Narrow flow range
Reciprocating machine Head not affected by gas properties Limited capacity
High efficiency (at low speed)
1.1.4 Fans and Blowers
Fans and blowers are low-pressure compressors commonly used to supply air or
gas to dryers, furnaces, and HVAC systems. According to the American Society of
Mechanical Engineers (ASME), a device with a compression ratio of up to 1.11
is called a fan. A blower has a compression ratio of 1.11–1.2. Machines with a
compression ratio above 1.2 are known as compressors.
Blowers and fans work on the same principle as centrifugal compressors; there-
fore, most of the discussions on centrifugal compressors in this book also apply to
blowers and fans. Fans play a crucial role in our discussions because the renowned
fan laws (Sect. 2.3) were based on the observation of fan operations.
1.2 Applications of Pumps and Compressors
A process stands for a series of unit operations to produce a material in large quan-
tities, either continuously or in batches. Typical industrial processes include oil and
gas production and refining, petrochemical and chemical processing, biochemical,
pharmaceutical, pulp and paper manufacturing, power generation plants, and food
and beverage. Pumps and compressors are critical and integral components of the
process flow configuration to move oil, gas, and other fluids to specified destinations
and at the desired pressure or flow rate.
1.2.1 Liquid Pumping
An elemental liquid pumping process includes the liquid supply from the pump
suction side and the liquid destination on the discharge side, as depicted in Fig. 1.5.
The largest pumps are typically used in oil pipeline operations. For example,
Fig. 1.6 illustrates the main oil line (MOL) in one of the Gulf countries. Around one
million barrels of crude oil are transported daily from the interior to the coast. The
MOL (size 38–42”) runs through mountainous terrain, reaching a “high point” of
∆
Fig. 1.5 A simple pumping process

1.2 Applications of Pumps and Compressors 7
Fig. 1.6 Large pumps in pipeline operation
670 m above sea level, before arriving at the oil terminal, which is over 100 km away
at the coast. Two boosting stations, equipped with pumps powered by gas turbines
or electric motors, provide the necessary energy to move the oil over the mountain.
The discharge pressure is at around 50 bar(g).
The critical requirement is maintaining a minimum pressure at the “high point”
to prevent a vacuum inside the pipeline that could cause it to collapse. The terminal
operates at near-atmospheric pressure. Arriving pressure must be controlled, and
excessive pressure must be avoided. In addition, large pressure fluctuations can cause
uneven flow and excessive turbulence, leading to emulsion formation in crude oil with
the risk of failing export requirements. Therefore, the reliable and stable operation
of the booster pumps is crucial for successfully transporting the oil.
A complex pumping process can involve multiple pumps operating in parallel or
series and different modes. For instance, two machines operate in parallel to increase
capacity or provide redundancy. Figure 1.7 shows a surface processing facility in
Fig. 1.7 A complete surface facility with pumps and compressors (reproduced, with permission,
from Niu and Xiao (2022))
upstream oil and gas production. Multi-phase fluid from wells is sent under pressure
to the separators, where it is separated into gas, oil, and water, then transferred
downstream via compressors and pumps for further processing or sales.
1.2.2 Gas Compression
Compressors play a critical role in many sectors of industry. Bloch (2006) has pro-
vided a comprehensive list of compressor applications in the process industry. Some
typical applications include wet gas compressors for FCCU, refrigerant compressors
in liquefied natural gas (LNG), CO2 injection, booster compressors for gas pipelines,
instrument air compressors, and vapor return compressors for LNG ship loading.
An elemental gas compression system includes the gas source, a suction scrubber,
a single-stage compressor, a discharge cooler, a discharge knockout drum, the recycle
line and valve, and the destination of the gas, as shown in Fig. 1.8.
1. Gas source. The gas to be compressed can come from various sources, depending
on the type of operation. For example, the gas supply is usually from upstream
separators in a typical E&P facility and cracked gas in an ethylene plant.
2. Suction cooler, inter-stage cooler, and discharge cooler. As the gas is compressed
to a higher pressure, its temperature increases simultaneously. For practical rea-
sons such as metallurgy, lubrication, and efficiency, the discharge temperature
must not exceed a specific limit, typically up to 150 ◦ C (Hanlon 2001). The gas
must be cooled at appropriate points along the flow path before being sent down-
stream for further compression or processing.
3. Suction and discharge scrubber. Liquids, solids, and other contaminants can prove
troublesome for the compressor operation and must be removed before entering
the compressor. The suction/discharge scrubber serves this purpose; the entrained
liquids/solids are “scrubbed” from the bottom.
Fig. 1.8 A typical gas compression process (reproduced, with permission, from Niu and Xiao
(2022))
1.2 Applications of Pumps and Compressors 9
4. Driver system. The driver provides mechanical power to the compressor. The type
of the driver depends on the power and torque requirements and can be a steam
turbine, turbo expander, electric motor with fixed-speed drive, variable-speed
gearbox (VSGB), or variable frequency drive (VFD). In most modern compres-
sors, the driver is an electric motor.
5. Compressor. The compressor is the central component of the compression system.
Each compressor operates within a specific operating range (compressor operating
envelope). Within this range, the compressor receives energy from the driver. It
then converts the energy to the gas’s potential energy (defined as the head or
pressure) and kinetic energy (defined by flow rate).
6. Recycle line and recycle valve. When a centrifugal compressor’s throughput (also
called the load or capacity) is lower than a specific limit due to low supply or
demand, the compressor becomes starved, and backflow may result. This abnor-
mal phenomenon is called compressor surge, characterized by rapid switching
between backward and forward flow. Compressor surge could harm the com-
pressor’s mechanical integrity and must be prevented. The standard solution to
prevent surge is circulating a portion of the compressed gas back from the dis-
charge side to the suction side of the concerned stage. Typically a special anti-
surge recycle valve is installed at each stage to adjust the amount of recycle flow.
With an air compressor, because air is unlimited in supply and not harmful to
the environment, it is more convenient to blow off part of the compressed air to
the atmosphere at the discharge side instead of recycling it back to suction to
re-compress.
7. Suction or discharge throttling valve. The throttling of the suction or discharge
valve varies the system resistance and thus changes the split of the added energy
between potential energy and kinetic energy.
8. Discharge check valves. Backflow is undesirable, especially in the case of an
emergency shutdown. Check valves (non-return valves) are installed at the com-
pressor discharge to prevent backflow.
9. Gas destination. Like the gas source, the back pressure at the gas destination also
significantly impacts the compressor control strategy.
Complex compression processes may involve multiple compressors working

together to achieve a higher pressure (via serial arrangement) or more flow through-
put (via parallel configuration). A compressor system with two compressors, each
with two stages, is shown in Fig. 1.9. This 2 × 2 configuration is sufficiently repre-
sentative to show the concept of a complex compression system and is also flexible
enough for scaling up or down to accommodate different needs. It will be used as
the process configuration in the later discussions.
Machines operating in parallel share the same suction and discharge headers. As
a result, all machines receive fluid from the same suction header, and all the running
trains share the same pressure ratio.
Figure 1.10 is an example of a depletion compression process. Gas wellhead
pressure depletes over time, from 25 bar(g), 10 bar(g) to the eventual 3 bar(g)
before permanent shut-in of the wells. Compressors are added in phases to cope
Fig. 1.9 A compressor system with two compressors in parallel
Fig. 1.10 Multi-phase depletion compressors
with the decreasing pressures of the gas supply. The startup and operation of
such a complex compressor network are quite challenging. Meanwhile, the many
ways of configuring and operating the system provide opportunities for overall
optimization.
Figure 1.11 shows the process flow of a typical ethylene plant. A five-stage cracked
gas compressor (CGC) and many pumps and other compressors move the cracked
1.3 Operation and Control of Pumps and Compressors 11
Fig. 1.11 A typical ethylene plant with compressors
gas and condensed liquids. The high compression ratio and wide variation in the
cracked gas composition pose a considerable challenge to operation and control.
1.3 Operation and Control of Pumps and Compressors
Pumps and compressors are crucial to the process operation, with primary concerns
on the safe and efficient operation of both the process and equipment (Liptak 2006;
King 2016; McMillan 1983; Watterson 2018).
1.3.1 Operating Requirements and Control Objectives
The primary objective of pumps and compressors in an industrial setting is to safely

and efficiently move the process fluid (liquid for pumps and gas for compressors)
from one location to another as required by continuous process operation. A large
pump or compressor is characterized by some unique properties:
1. Large capital investment. A large pump or compressor can easily cost millions of
dollars to purchase and install.
2. Critical to operation. The liquid pumping and gas compression system is essen-
tial to the operation. A trip or shutdown of a critical pump or compressor often
significantly interrupts production, resulting in production loss or deferment.
3. High operating cost. Liquid pumping and gas compression constitute a consider-
able part of the operating cost. It is common to see compressors rated at over ten
megawatts in the oil and gas industry.2
4. Maintenance intensive. Pumps and compressors are also maintenance-intensive
and account for a considerable part of rotating equipment engineers’ maintenance
effort.
The rise in global energy consumption and environmental consciousness, coupled

with the depletion of fossil fuel reserves, has led to a surge in demand for new
compression systems. This demand is particularly evident in the increasing number
of gas-to-liquids (GTL), coal-to-liquids (CTL), carbon capture and storage (CO2
injection), and integrated gasification combined cycle (IGCC) facilities, which rely
heavily on extensive compression systems (Elliott and Bloch 2021; Shelley 2006).
Minimal energy consumption, maximum throughput, and low maintenance cost have
become the mandate. As a result, the efficient operation and reliable protection of
pumps and compressors are critical factors in process control design and operation.
1.3.2 Overall Control Strategy
A reliable process control solution helps achieve safe and efficient operation by the
following means:
1. Regulate the process flow to maintain the desired pressure and capacity during
normal operation. Due to inevitable disturbances and upsets, continuous adjust-
ment of the process flow rate is required, even if the steady-state material balance
is adequately satisfied.
2. Protect the process and equipment against unsafe and inefficient operation during
abnormal situations. Many engineering limits and operational constraints, such
as pressure, temperature, flow, and motor power, set the operation boundary of
the machines. Protective control is provided to keep the operating point within
the desired operating range.
3. Safeguard the process and machine against catastrophic failures. Although opera-
tors and automatic control are charged with keeping the process operation within
the operating limits, the effectiveness of control and the speed of response may
not be sufficient. The operating point may run away and go out of control. In this
2For example, a 10-megawatt compressor consumes more than 80 million kWh of power (10.0
MW × 1000 kW/MW) × 24 h/day × 365 day/yr = 87,600,000 kWh or over 8 million dollars per
year, assuming US$0.1/kWh for electricity.
Fig. 1.12 Process automation system (PAS) (reproduced, with permission, from Niu and Xiao
(2022))
case, the safeguarding system will proactively shut down3 the operation to avoid
more severe consequences.
4. Transition between different operating modes. A machine can operate in different
modes, and so is the process. The startup, shutdown, and on-stream modes are
essential. Reduced-capacity operation (also known as the turndown mode) is
commonplace. The transition from one mode to another can be associated with
many risks and can be highly complex. Sequence and logic are the primary means
for transitional control, but interaction with and support from the process control
scheme are also critical.
For a complex pumping or compression process, there are additional consider-
ations on the overall optimality of operation. For instance, the flow and pressure
distribution among multiple compressors can significantly affect operational effi-
ciency, which can be improved by rationalized operation or optimized control.
A plant automation system (PAS), typically a combination of distributed control
systems (DCS), safety instrument systems (SIS), and fire and gas system (FGS),
provides the infrastructure to implement the control solution. PAS typically consists
of three subsystems, as illustrated in Fig. 1.12.
1. Distributed control system (DCS). A DCS is the core platform and operator inter-
face for monitoring and control. The DCS system is typically also the human–
machine interface (HMI) for all other subsystems.
2. Safety instrument system (SIS). SIS is typically a PLC-based system running
condition-based logic and sequences. In case of an emergency, it proactively shuts
down the process and equipment, isolates hydrocarbon inventories, and switches
off the electric systems.
3An emergency shutdown is very disruptive with high risks and should be avoided as much as
possible.
3. Fire and gas system (FGS). The fire and gas system continuously monitors all
plant areas for abnormal conditions such as a fire or combustible/toxic gas release
via various fire and gas detection instruments. The system alerts the operator of
imminent danger if a hazardous situation occurs. Depending upon the hazard level,
the fire and gas system can also proactively shut down the process equipment and
activate automatic fire fighting systems to prevent incident escalation.
1.3.3 Process Control Technologies
Process control is a branch of automatic control vital to safe and efficient operation.
The pervasive control technologies range from simple and complex PID control
schemes to advanced control solutions. This section provides a highlight of the core
concept and a summary of the prevalent process control technologies for pumps and
compressors. For comprehensive discussions on process control, see King (2016),
Niu and Xiao (2022), Smith (2010).
Feedback control is one of the most commonly used mechanisms for automatic
control, as shown in the block diagrams in Fig. 1.13. The basic principle of feedback
control is
Measure → Compare → Correct . (1.1)
Manual control by the operator is the primitive form of feedback control. The
operator monitors the value of the controlled variable, compares it with the desired
value, and adjusts the manipulated variable (e.g., control valves) to keep the con-
trolled variable at or around its target value. This checking and adjusting keep on as
frequently as necessary.
This measure-compare-correct process is highly tedious and prone to
human errors. An automatic control solution replaces the operator to perform this
repetitive work more reliably and efficiently. Therefore, operator control and auto-
matic control are both feedback controls. They complement each other to achieve
the same control objectives.
A feedback control loop comprises five essential components, as illustrated in
Fig. 1.13, namely, the process to be controlled (➀), measurements and feedback
Fig. 1.13 Closed-loop control (feedback control)

mechanism (➁), final control element (➂), the control objective (➃), and the control
algorithm (➄).
1. Process. A process refers to the physical plant or equipment to be controlled.
Control is based on process dynamics, which is the transient behavior of a process
variable responding to a change in another variable. The critical information
required on the process is the dynamic cause-and-effect relationships between,
and transient behaviors of, these variables (See Fig. 1.14 for an illustration of
process dynamics and cause-and-effect relationship).
2. The measurement, monitoring, and feedback mechanism. The measurement pro-
vides a window into the process or equipment to provide an undistorted view
of changes in the process variables. The availability and credibility of process
measurements are crucial for process control.
A process measurement typically includes a sensor and a transmitter, with the
former providing the physical measurement and the latter transmitting the mea-
surement value to the controller as feedback. The four conventional process vari-
ables are flow, pressure, level, and temperature. See Fig. 1.15a for an illustration
of the flowmeter. For advanced control applications, inferential properties (soft
sensors) are widely used for those properties that are not directly measurable
or cannot be measured reliably. See Fig. 1.15b for an illustration of the surge
indicator calculation that is discussed in detail in Sect. 5.3.
3. Final control elements (FCE). The final control element is a manipulatable device
that can influence the process and cause the process output to change pre-
dictably. The control valve is the most common final control element, as shown in
Fig. 1.16a. A control valve is also called a throttling valve to distinguish it from an
on/off valve. The critical valve properties of concern to process control include
the speed of response, valve characteristics (see Fig. 1.16b.), and positioning
accuracy.
Other final control elements such as electric current, motor speed, or switches are
also commonly encountered.
∆
∆
Fig. 1.14 A process with noise and disturbance

Fig. 1.15 Process measurement components
Fig. 1.16 Valve characteristics
4. Control objectives. The control objectives serve the operating objectives and
dictate what process variable is to control and at what value it should be
controlled (see Fig. 1.17 for an illustration of operating point and operating enve-
lope).
5. Control algorithm or logic. The control algorithm or logic calculates the control
action to eliminate the control error and maintain the process value on the target.
The controller can be a simple on/off control logic, a standalone PID controller
with one control handle and one control target, or a complex control scheme con-
sisting of multiple PID controllers. The selection of control algorithms depends
on many factors, and the critical considerations include performance, reliability,
and the life-cycle ownership cost.
Fig. 1.17 Control objectives and constraints (reproduced, with permission, from Niu and Xiao
(2022))
Fig. 1.18 On/off control and PID control
On/off control is the simplest controller and thus is also the least expensive in
life-cycle ownership. The manipulated variable can assume just two values, e.g.,
open/close for a valve, on/off for a switch, or run/stop for a pump. Figure 1.18a
shows an example of a simple on/off switching control, represented by LS, where S
stands for switching, with an on/off valve LSV-101.
On/off control is a reliable and cost-effective option for control problems where
sloppy control performance is acceptable, such as with levels of large storage tanks,
the temperature in room heating, and pressure in instrument air. The major downside
Fig. 1.19 Structure of a PID control loop
of an on/off control scheme is that it cannot provide the required granularity in control
moves to maintain a tight setpoint. For example, a home A/C system usually has an
air compressor with on/off control, which can only maintain the room temperature
within a set range (e.g., ±1 ◦ C).4
The majority of industrial process control is by PID control loops. A PID con-
troller, as shown in Fig. 1.19, offers three types of control actions:
1. Proportional action is a linear function of the control error e(t) and provides a
quick initial response; For example, for temperature control, the control action
for a 20 ◦ C control error is twice that for a 10 ◦ C error. However, proportional
action requires a non-zero control error to sustain the control action. This non-zero
control error results in a steady-state offset, a significant drawback of proportional
control.
2. Integral action eliminates steady-state errors; The control action is proportional to
the time integral of the control error. For example, for temperature control, as long
as there is a difference between the actual temperature and the desired temperature,
the integral action will continue changing the heat input in the direction of reducing
the control error.
3. Derivative action provides anticipatory action for better disturbance rejection. It
calculates the rate of change in the control error and asserts a control action pro-
portional to it for quick response. However, noises and disturbances are prevalent
in the process variables. The derivative action amplifies noises and may result
in too aggressive control actions. For this reason, most controllers in the process
industry are P or PI controllers. D action is rarely used or used with carefully
designed signal filters.
Most industrial control loops are simple PID control loops (Åström and Kumar
2014; Lee and Weekman 1976; Shinskey 2001; Smith 2010), also called standalone
PID control loops. They typically consist of only one variable to control (the control
target) and one variable to manipulate (the control handle).
Complex control problems consist of multiple controlled and manipulated vari-
ables. These variables must be considered together due to the interactions among
4Variable-frequency A/C system is becoming increasingly affordable and can provide better tem-
perature control with less power consumption for improved comfort.
Table 1.2 Common complex control loops

Single control target Multiple control targets
On/off control Cascade control
Simple PID control Ratio control
Protective control Override control
Split-range control Selective control
Dual controller control Decoupling control
Fan-out control Model-based predictive control
Feedforward control
them. A natural solution is to have multiple simple controllers working together to

address the complexities.
Table 1.2 lists the common process control loops by the input/output structure.
Cascade control, split-range control, fan-out control, and override control are the
most widely used complex control schemes in pump and compressor control. For
more in-depth discussions, please refer to Niu and Xiao (2022), Smith (2010).
1.3.4 Monitoring and Safeguarding
Performance monitoring is crucial to the pump and compressor system’s sustainable

and efficient operation. Performance monitoring serves two primary users, the control
room operators and the process control engineers.
Pump and compressor vendors typically offer comprehensive monitoring tools to
ensure that the mechanical integrity of the machine is monitored, abnormal conditions
are reported, and the operators are alerted. However, they are much less interested
in the overall condition of the process operation, which requires intimate knowledge
of both the process and control.
A process is designed to operate under different setpoints and limits, including
regulatory control setpoint, protective control setpoint, alarm limits, and trip limits.
If automatic control fails to maintain the control target, an alarm alerts the operator
for manual intervention. If operator intervention fails to bring the operating point
back to safety, instrumented safeguarding logic kicks in to shut down the operation.
The different layers of control and protection are achieved by operating at different
setpoints and limits in a staggered fashion, with the setpoints and alarm limits properly
aligned and spaced to allow ample time for the next line of defense to respond.
Figure 1.20 shows the distinct layers of safety protection in an operating plant. Pro-
cess control is the first layer of protection against process fluctuations and upsets, fol-
lowed by the safeguarding layer against unsafe operations. Close integration between
process control and safeguarding is critical in an integrated design to avoid unnec-
essary process disruptions and shutdowns.
Fig. 1.20 Layers of process safety protections (reproduced, with permission, from Niu and Xiao
(2022))
1.4 Practical Considerations
The lifespan of rotating equipment is very long, typically spanning multiple decades.
Many of them are required to operate 24/7 for several years nonstop. The design and
control philosophy must consider this long lifespan and high up-time requirements.
1.4.1 Process-Centric Versus Equipment-Centric Solutions
The operation and control of a piece of rotating equipment must be considered within
the context of the process configuration. Rotating equipment is part of the process
and must coordinate with other components, such as vessels, exchangers, furnaces,
distillation columns, piping, and valves to meet the operating requirements. Together,
they serve the overall process control objectives. For this reason, the process control
1.4 Practical Considerations 21
solution must be process-centric aiming at overall integrity and optimality, rather

than an equipment-centric solution focusing on the equipment alone. A thorough
understanding of both the process and equipment is essential.
Equipment control is traditionally monopolized by equipment manufacturers
or their affiliates/partners, emphasizing equipment safety and mechanical integrity
rather than the overall process control performance. However, the equipment manu-
facturer is unlikely to have sufficient knowledge of the process operation or process
control. It is challenging for them to design an equipment control scheme that seam-
lessly integrates into the overall process control solution that they are unfamiliar
with. Process control engineers with a holistic view of the operating requirements
are thus better suited for designing the equipment control solution as an integral part
of the overall control solutions.
1.4.2 In-House Versus Third-Party Solutions
The protective control of pumps and compressors has traditionally been using pro-
prietary technology and dedicated hardware and software (Elliott and Bloch 2021,
Chap. 13). The downside of this approach is the high cost of life-cycle ownership.
Over-reliance on a single third party for hardware, software, and services can result in
unnecessary service delays and functionality limitations. Compressor control solu-
tion providers do not have a competitive edge over mainstream control systems (DCS
and PLC) vendors in providing a reliable and versatile control system infrastructure
to support the control solutions.5
As technology advances, especially in the control systems hardware (DCS and
PLC), compressor control solutions can now be implemented in standard control sys-
tems similar to other process control solutions. The entire process control solution,
including equipment control, can be designed, implemented, operated, and main-
tained by competent in-house process control engineers and instrument technicians.
The in-house solution has many advantages over the third-party solutions. Table 1.3
lists the advantages and disadvantages of the two types of solutions.
It is encouraging that the number of solutions based on standard process control
technology on open platforms is rapidly increasing (Zelenov 2013). Some commer-
cial vendors have also started to offer solutions in standard control systems instead of
their own (CCC 2021; Elliott and Bloch 2021; Vermillion et al. 2023). With sufficient
knowledge and experience, process control engineers can take ownership of a large
part, if not all, of the life-cycle activities, from design, implementation, operation,
and maintenance.
5The recent acquisition of Compressor Control Company (CCC) by Honeywell, announced on April
26, 2023, marks the beginning of the end of this proprietary black-box approach for compressor
control (Honeywell 2023).
Table 1.3 Third-party solutions versus in-house solutions

In-house solution Third-party solutions
Objectives Overall optimal operation Surge prevention + Others
Philosophy “Process-Centric” “Compressor Centric”
Technology Open, Transparent Vendor proprietary
Architecture Layered approach Proprietary, “One for All”
Infrastructure Standard DCS/SIS, “White-Box” Proprietary, “Black-Box”
CAPEX Existing DCS/SIS Vendor Hardware/Software
OPEX In-house support Vendor service
Warranty Via Standard DCS/SIS On proprietary hardware/Software
Service In-house support Vendor/Third party
Maintenance By In-house team Vendor service/Third party
1.5 Summary
Pumps and compressors are essential equipment in the process industry. The sig-
nificant investment, large energy consumption, and costly maintenance demand an
effective and integrated solution to control, protect, and safeguard their operation.
As part of the process flow configuration, the control solution for the equipment
must be based on a holistic view of the entire process operation instead of the equip-
ment in isolation, resulting in a process-centric solution instead of the traditional
equipment-centric one.
References
Åström KJ, Kumar PR (2014) Control: a perspective. Automatica 50(1):3–43

Bloch HP (2006) Compressors and modern process applications. Wiley
Boyce MP (2003) Centrifugal compressors – a basic guide. PennWell Corporation, Tulsa, USA
den Braembussche RV (2019) Design and analysis of centrifugal compressors. ASME
CCC (2021) CCC and Yokogawa R&D Cooperation. Tech. rep., Compressor Con-
trol Company (CCC), https://www.cccglobal.com/wp-content/uploads/2022/10/CCC-and-
YokogawaRD-cooperation.pdf
DOE (2001) Pump life cycle costs: a guide to LCC analysis for pumping systems. Tech. Rep. DOE-
GO-102001-1190, Hydraulic Institute, Euro-Pump, and the US Department of Energy’s Office
of Industrial Technologies (OIT)
DOE (2006) Improving pumping system performance — a sourcebook for industry. Tech. rep., US
Department of Energy’s Industrial Technologies Program (ITP) and the Hydraulic Institute (HI)
Elliott H, Bloch H (2021) Compressor technology advances — beyond 2020. Walter De Gruyter
Forsthoffer MS (2017) More best practices for rotating equipment. Elsevier Inc
Forsthoffer WE (2005a) Forsthoffer’s rotating equipment handbooks, vol. 2: pumps. Elsevier Ltd
Forsthoffer WE (2005b) Forsthoffer’s rotating equipment handbooks, vol. 3: compressors. Elsevier
Ltd
Hanlon PC (2001) Compressor handbook, 2nd edn. McGraw-Hill
References 23
Honeywell (2023) Honeywell to acquire compressor controls corporation, driving the energy
transition through leading automation and controls portfolio. https://www.honeywell.
com/us/en/press/2023/04/honeywellto-acquirecompressor-controls-corporationdrivingthe-
energytransitionthrough-leading-automation-and-controls-portfolio
King M (2016) Process control: a practical approach, 2nd edn. Wiley
Lee W, Weekman VW (1976) Advanced control practice in the chemical industry. AICHE J 22(27)
Liptak BG (2006) Process control and optimization, instrument engineers’ handbook, vol II, 4th
edn. Taylor and Francis
McMillan GK (1983) Centrifugal and axial compressor control. Instrument Society of America,
http://compressorcontrolstudent.modelingandcontrol.com
Niu S, Xiao D (2022) Process control – engineering analyses and best practices. Advances in
industrial control. Springer
Shelley S (2006) The challenge of mega syngas plants. Turbomach Int Mag 8–11
Shinskey F (2001) Process control: as taught vs. as practiced. Ind Eng Chem Res 41
Smith CL (2010) Advanced process control – beyond single-loop control. Wiley
Stewart M (2019) Surface Production Operations, Volume IV: Pumps and Compressors. Elsevier
Inc
Stoffel B (2015) Assessing the energy efficiency of pumps and pump units. Elsevier
Vermillion L, Gracia J, Ilchenko M (2023) Solutions spotlight: integrating process and turbomachin-
ery control. Control Mag. https://www.controlglobal.com/podcasts/article/21546058/yokogawa-
electric-corporation-solutionsspotlight-integrating-processand-turbomachinery-control
Watterson JM (2018) A simple guide to understanding compressors. Momentum Press
Zelenov A (2013) White paper – CCS implementation of surge prevention control system on yoko-
gawa stardom PLC. Tech. rep, Continuous Control Solutions (CCS)
Chapter 2
Characteristics of Pumps and
Compressors
Pumps and compressors operate very differently from static equipment like reactors
and distillation columns. However, when it comes to process control, the informa-
tion required for controlling rotating equipment is not vastly different from that of
static equipment. The analytical approach also remains fairly similar. The key is to
identify the manipulated and controlled variables, the causal relationships among
these variables, and their transient behaviors. This chapter presents the characteris-
tics of pumps and compressors from a process control perspective, with a focus on
a unified approach to understanding machine behavior and extracting the necessary
information for control. A basic understanding of thermodynamics is also helpful for
applying fundamental principles and processes to analyze equipment characteristics.
2.1 Basic Properties of Fluids
A fluid can exist in any of the three states: gas, liquid, or solid. The form and volume
of fluid in the gas state can change under external influences, while the form and
volume are both fixed in a solid state. In a liquid state, the form can change, but the
volume cannot.
Pumps deal with liquids, and compressors handle gases. There are numerous
theories and physical laws describing gas and liquid properties. It is not the objective
to provide systematic and in-depth coverage of the fluid properties in this book.
Instead, a few fundamental concepts and core principles are presented here to help
understand the behavior of liquid pumping and gas compression.

https://doi.org/10.1007/978-3-031-43122-7_2
26 2 Characteristics of Pumps and Compressors
Fig. 2.1 Conservation of mass
2.1.1 Basic Principles of Fluid Dynamics
Fluid dynamics is also called hydrodynamics. It is the study of the movement of

liquids and gases. Critical to the process analysis and control are the conservation
laws and Bernoulli’s equation.
The foundational axioms of fluid dynamics are conservation laws, specifically,
the conservation of mass, momentum, and energy. Conservation of mass states that
the mass in a control volume can neither be created nor destroyed. Its implication to
pump and compressor operation is that the mass flow rate remains the same during
pumping or compressing, while the volumetric flow rate may change with external
conditions. In equation format:
Fm = ρ1 A1 v1 = ρ2 A2 v2
= ρ1 Fv,1 = ρ2 Fv,2 (2.1)
where Fm is the mass flow rate, Fv is the volumetric flow rate, ρ is the fluid density, v
is the flowing velocity, and A is the cross-sectional area. The subscripts “1” and “2”
denote any two points along the flow path, e.g., a compressor’s suction and discharge
side, as shown in Fig. 2.1.
The conservation of momentum states that any change in the momentum of the
fluid within a control volume is due to the net flow of momentum into the volume and
the external forces acting on the fluid inside the volume. The change in momentum
can be determined by the change in the product of mass and velocity or by the external
force acting on the fluid and is reflected in the changes in velocity or flow direction
inside the pump or compressor. The conservation of momentum governs the transfer
of mechanical energy to potential energy at the tip of the impellers in centrifugal
machines.
Similarly, conservation of energy claims that energy can be converted from one
form to another, but the total energy in a closed system remains constant. For pumps
and compressors, energy is transferred from the external driver (e.g., an electric
motor) to the fluid inside the machine as either potential energy (head or pressure)
or kinetic energy.
2.1 Basic Properties of Fluids 27
Fig. 2.2 Bernoulli’s

equation for flow
The primary interest with rotating machines is how energy is transferred and
converted. For instance, an increase in the speed of the fluid coincides with a decrease
in static pressure or a decrease in the fluid’s potential energy.
The conservation laws can be summarized with Bernoulli’s Equation (Fig. 2.2),
which relates the different forms of energy as follows1 :
1
W =P+ ρ v2 + ρ g h (2.2)
2
1 1
P1 + ρ1 v12 + ρ1 g h 1 = P2 + ρ2 v22 + ρ2 g h 2 , (2.3)
2 2
where W is the energy added from the external source, v is the fluid velocity, P is
the pressure, and h is the elevation.
2.1.2 Basic Principles of Thermodynamics
Thermodynamics is a branch of physics that deals with heat, work, temperature, and
their relation to energy and entropy. Classical thermodynamics is characterized by
three fundamental laws, from the first to the third. The first law of thermodynamics
states that the changes in the system’s internal energy follow the energy conservation
laws when energy passes into or out of a system (as work, heat, or matter). The second
law states that in a natural thermodynamic process, the sum of the entropy of the
interacting thermodynamic systems never decreases. In other words, heat does not
spontaneously pass from colder to warmer bodies. The third law states that a system’s
entropy approaches a constant value as the temperature approaches absolute zero.
The laws of thermodynamics govern the working of a compressor. The first law
implies that the total energy remains constant. The external driver imparts energy
onto the fluid inside the machine, and the energy is converted to potential and kinetic
energies (and losses). The second law helps explain the different compression pro-
cesses, where the inevitable heat loss in gas compression renders most compression
processes irreversible.
1Bernoulli’s equation is only strictly valid for incompressible fluid but can be modified to apply to
compressible gas for compressor applications.
The state of a fluid under a given set of physical conditions is described by state
variables such as pressure, volume, temperature, or internal energy. The state vari-
ables can be related by a thermodynamic equation known as the equation of state,
which describes the properties of both pure substances and mixtures in liquids, gases,
and solid states. The general form of an equation of state may be written as
f (P, V, T ) = 0, (2.4)
where P is the pressure, V is the specific volume (volume per mole), and T is the
temperature.
Although the equation of state can be used to describe solids, liquids, and gases, it
is most frequently used to describe gas properties. Gas can be characterized as ideal
or real. The equation of state for an ideal gas, or the ideal gas law,2 relates the gas’s
pressure, temperature, and volume as follows3 :
PV =mRT Ideal gas law (2.5)

f (P, V, T ) = P V − m R T = 0 Equation of state for an ideal gas, (2.6)
where m is the number of moles of a substance, and R is the gas constant (R =

8.31446 m3 · Pa/K · mol).
An ideal gas is convenient for analysis but exists only in theory. An empirical
correction factor is applied to make the analyses valid when dealing with real gases.
This correction factor is the gas compressibility Z , with Z = 1 for an ideal gas and
Z < 1 for real gases.4 Subsequently, the ideal gas law is modified to include the
compressibility factor as
P v = Z R T. (2.7)
The most crucial gas property concerning a compressor’s operation and control
is the gas density ρ. The gas density is determined by the gas composition (molec-
ular weight, compressibility, and ratio of specific heat) and the operating conditions
(pressure and temperature) and can be represented as
⎛ kg N kg ⎞
kPa
PM ⎜ kmol = m2 kmol ⎟ ,
ρ= ⎝kg/m =
3
⎠ (2.8)
Z RT J N·m
K K
mol · K mol · K
2 The equation of ideal gas law is based on the laws of Charles, Boyle, Gay-Lussac, and Avogadro.
3 Petroleum engineers call the pressure, volume, and temperature the PVT properties, which describe
the physical properties of reservoir fluids and variations in the volume and phase state that occurs
during oil production.
4 Under extremely high pressure, the gas compressibility Z may exceed 1.0.
where M is the molecular weight of the gas. The density of a liquid is determined
mainly by the type of liquid (molecular weight) and does not change markedly with
pressure and temperature.
The core difference between liquid pumping and gas compression is that gas is
compressible while a liquid is incompressible. In other words, the density of a liquid
can be assumed constant during pumping, while the density of gases decreases when
compressed.
Example 2.1 Gas density calculation. Figure 2.3 shows part of an orifice-based
flowmeter datasheet. A flowmeter is a critical element in compressor control, requir-
ing gas density as a crucial gas property for design or calibration.
The operating condition and gas property are given by the datasheet as follows:
Pressure: P = 23 bara = 23 × 100 kPa = 2, 300 kPa

Temperature: T = 83.5 ◦ C = 83.5 + 273.15 K = 356.65 K
Molecular Weight: M = 59.643 kg/kmol
Compressibility: Z = 1.0.
The gas density can be calculated and validated with Eq. 2.8:
PM
ρ=
Z RT
2300 × 59.643
=
1.0 × 8.314 × 356.65
= 46.26 kg/m3 , (2.9)
where R = 8.314 J/mol/K is the gas constant. The calculated gas density agrees with
the value given in the datasheet.
2.1.3 Gas Compression Process
The pressure and volume of a gas obey the following relationship when being com-
pressed:
P vn = const, (2.10)
where P is the gas pressure, and v is the specific volume. The polytropic exponent
n indicates the type of compression process. For instance, the condition of n = 0
represents an isobaric process, n = 1 for an isothermic process, and n = ∞ for an
isochoric process. See Table 2.1 for a summary of all possible compression processes.
An isobaric process does not cause pressure change; thus, no work is done. It is
thus not much useful. On the other hand, an isochoric process maintains the volume
Fig. 2.3 Gas density calculation in a flowmeter datasheet
Table 2.1 Compression cycles

Polytropic Exponent Process Equation Description
n =0 Isobaric p = const Constant Pressure
n =1 Isothermic p v = const Constant Temperature
n =κ Isentropic p v κ = const Reversible and Adiabatic
n ∈ (1, ∞) Polytropic p v n = const Reversible but non-adiabatic
n =∞ Isochoric (isometric) v = const Constant Volume
constant while working on the fluid, coinciding with the pumping of an incompress-
ible liquid. The polytropic process is a generic term covering many compression
processes.
An isentropic process is an ideal compression process that only exists in theory.
However, it is frequently used as a benchmark process to facilitate analysis. An
isentropic compression process is both adiabatic and reversible (without heat loss).
The polytropic exponent of an isentropic process equals the ratio of specific heat
κ = C p /Cv , where C p is the specific heat under constant pressure, and Cv is the
specific heat under constant volume. The ratios of specific heat for some common
gases are given in Table 2.2. The process gas is usually a mixture of different pure
Table 2.2 Ratio of specific heat for some common gases

Gas Formula Mole Weight Cp Cv κ
kg/kmol kg/(kg.K) KG/(kg.K) -
Air 28.96 1.01 0.72 1.40
Hydrogen H2 2.016 14.32 10.16 1.41
Nitrogen N2 28.02 1.04 0.74 1.40
Oxygen O2 32.00 0.91 0.66 1.40
Ammonia NH3 17.03 2.19 1.66 1.31
Methane CH4 16.04 2.22 1.70 1.30
Ethane C2 H6 30.07 1.75 1.48 1.19
Ethylene C2 H4 28.05 1.53 1.23 1.24
Acetylene C2 H2 26.04 1.69 1.37 1.23
Propane C3 H8 44.10 1.67 1.48 1.13
Carbon Dioxide CO 28.01 0.84 0.60 1.29
Carbon Monoxide CO2 44.01 1.02 0.72 1.40
gases. It can be as “light” as hydrogen or as “heavy” as cracked gas. See Table 2.3
for some examples.
When a gas is compressed, the gas properties can vary in response to changes in
external conditions, such as pressure and temperature. Figure 2.4 illustrates a recip-
rocating compressor’s pressure and volume changes under different compression
cycles.
2.1.4 Liquid Transport Process
A liquid is usually assumed to have a constant density during the pumping process.5
Pumps work by the same principle as compressors, but the assumption of constant
density makes the analysis and control much more straightforward.
Although not precise in theory, the discussion of liquid pumping and gas com-
pression can be unified by assuming that pumping an incompressible liquid can be
treated as a polytropic compression process with n = ∞, i.e., an isochoric process
that does not involve volume changes in the fluid, see Fig. 2.4. This book will focus
primarily on the analysis and control of compressors. Once compressor operation
and control are understood, most of the analysis can readily apply to pumps in a
simplified form by forcing n = ∞.
5 All fluids are compressible to an extent; that is, changes in pressure or temperature cause changes
in density. However, with liquid, the changes in pressure and temperature are sufficiently small that
the changes in density are negligible. In this case, the flow can be modeled as an incompressible
flow. Otherwise, the more general compressible flow equations must be used.
Table 2.3 Examples of process gas compositions

Composition Natural Gas Cracked Gas Hydrogen Air
% % % %
Argon Ar 0.9
Hydrogen H2 1.05 18.0 99.0
Nitrogen N2 78.1
Oxygen O2 20.9
Carbon Dioxide CO2 1.06 0.03
Methane CH4 90.00
Acetylene C2 H2 0.1
Ethylene C2 H4 52.6
Ethane C2 H6 5.37 14.1
Propylene C3 H6 8.3
Propane C3 H8 1.70 3.7
Butene C4 H8 1.9
n-Butane C4 H10 0.33 0.6
i-Butane C4 H10 0.27
n-Pentane C5 H12 0.09
i-Pentane C5 H12 0.06
n-Hexane C5 H12 0.07
Others 0.7 1.00 0.07
Total 100.0 100.0 100.0 100.0
Fig. 2.4 Compression cycles

2.2 Description of Pumps and Compressors 33
Fig. 2.5 Three basic variables for pump performance description
2.2 Description of Pumps and Compressors
Pumps and compressors are machines that convert and transfer the mechanical energy
from the driver to the potential and kinetic energies of the fluid. The general and
fundamental principles of pumps and compressors can be found in, for instance,
Bachus and Custodio (2003), Cumpsty (1989), Forsthoffer (2005), Elliott and Bloch
(2021), Giampaob (2010), Gresh (2001), Watterson (2018), although much of the
information is of no interest to process control. From a process control perspective,
the primary concerns are the variables that describe the machine characteristics and
the cause-and-effect relationships among them. These can be summarized as three
variables, two curves, and one point, described with an invariant coordinate system.
2.2.1 Three Variables: Speed, Head, and Flow
Pumps and compressors work on similar principles: the external driver, such as an
electric motor and turbo expander, imparts energy to the fluid (gas or liquid) inside
the machine. The energy is then converted to the fluid’s potential and kinetic energies,
reflected by the increase in the head and flow.
The energy transfer and conversion can be conveniently described with three
basic variables: speed, head, and flow, as shown in Figs. 2.5 and 2.6 for pump and
compressor, respectively. The speed indicates the amount of energy transferred to
the fluid, while the head and flow reflect the split between the potential and kinetic
energy added to the fluid.
The characteristics of, and the relationship between, the three variables are
critical in understanding and controlling pumps and compressors. For centrifugal
machines, the impeller design determines the machine’s characteristics, and the rela-
tionships between the three variables (see Fig. 2.7) are complex. For reciprocating
machines, on the other hand, the relationships among the three variables are more
straightforward.
Fig. 2.6 Three basic variables for compressor performance description
Fig. 2.7 Impellers and diffusers in a centrifugal machine
Speed N
The machine speed refers to the shaft speed measured in revolutions per minute
(rpm) for a centrifugal machine or the frequency of piston strokes for a reciprocating
machine (1/s).
With a radius of r2 , the impeller tip speed U2 in a centrifugal machine (see Fig. 2.8)
is related to the rotating speed N via U2 = N · r2 .
For both centrifugal and reciprocating machines, the machine speed is an inde-
pendent variable determined by the external driver and can be constant or varying,
depending on the type of driver. Correspondingly, the machine can be operated in
fixed-speed or variable-speed mode.
Flow Fv
For a centrifugal machine, the actual velocity that the fluid leaves the impeller is
given by both the impeller tip speed U2 of the machine and the relative exit speed
W2 of the fluid (see Fig. 2.8):
Fig. 2.8 Three basic variables: head, flow, and speed
→ → →
[Absolute Exit Velocity] V2 = [Speed] U2 + [Relative Velocity] W2 , (2.11)
The flow rate is the fluid volume measured in, e.g., cubic meters per hour (m3 /h)
or gallons per minute (GPM):
⎧
⎨[Volume Flow] = [Exit Area] × [Exit Speed] × 3, 600
(m 3 / h) (m 2 ) (m/s) (s/ h) (2.12)
⎩F = A · V2 .
v
A reciprocating machine’s flow rate is the piston volume multiplied by the stroke
frequency. That is
[Volume Flow] = [Stoke Volume] × [Stroke Frequency] × 3, 600. (2.13)

(m 3 / h) (m 3 ) (1/s) (s/ h)
Head H
The head H of a machine is the energy required to deliver one mass unit of the fluid
at a given temperature from one energy level (e.g., suction pressure) to another (e.g.,
discharge pressure). The fluid can be liquid, gas, or even solid, depending on the
conditions of the fluid at that time.
Head as an energy variable has an engineer unit of kJ/kg. For legacy reasons, the
pump head is typically represented as the height h of the liquid in meters or feet.
The pump head can be intuitively interpreted as the maximum height that the pump,
at a fixed speed, can raise the liquid to. A compressor works on gas that is much
Fig. 2.9 Head and pressure
“lighter” than liquid, so the concept of the head in terms of height is not as intuitive
as the pump head. As a result, the compressor head is more commonly (and rightly)
represented as energy, e.g., in kJ/kg.
Head in elevation (meters) and energy (kJ/kg) can be confusing when pumps and
compressors are discussed together. For example, ethylene can be in either liquid or
gas form. If it is a liquid, a pump is needed to raise its pressure, and the pressure
increase is usually represented as a head in meters or feet. If the condition renders the
ethylene stream a vapor, then a compressor must be used to compress it to a higher
pressure, and the pressure increase is measured as head in an energy unit such as
kJ/kg.
However, the two units can be readily converted between each other by multiplying
with or dividing by the gravitational constant g = 9.81 m/s2 since they are both
potential energies, as shown in Bernoulli’s equation (Eq. 2.3):
H = g · h, (2.14)
(J/kg=m 2 /s 2 ) (m/s 2 ) (m)
where h is the head in height (meters), and H is the head in energy unit (kJ/kg). See
Fig. 2.9 for an illustration of the differences.
For example, to pump water to a height of 10 m (h = 10 m), a head of H = g h =
9.81 × 10 = 98.1 J/kg is required, which is translated to a pressure of Pd − Ps =
ρgh = ρ H = 1, 000 × 9.81 × 10 = 98, 100 Pascal, or 9.81 kPa, approximately 1
atmospheric pressure.6 In other words, the pump must generate 98.1 kPa of pressure
to lift the water to 10 m high, equivalent to a head of 98.1 J/kg.
6A rule of thumb is: 1 bar (1 bar = 100 kPa) or approximately one atmospheric pressure (1 atm ≈
101.3 kPa) is roughly 10 m of water height, or 1 m of water generates roughly 10 kPa of pressure.
A Mollier diagram that shows the relationship between pressure and energy at
various temperatures can help understand the concept of the fluid head, which will
not be explained in detail here.
2.2.2 Two Curves: Performance and Resistance Curves
Pumps and compressors are part of the process flow configuration. The performance
curve describes the head-flow relationship inside the pump or compressor, while the
system resistance curve describes the pressure and flow profile in the process piping.
The two curves together dictate how the machine operates.
Equipment Performance Curves
The performance curve is a graphical representation of the relationship between the

three basic variables: head, flow, and speed, under the head-flow coordinate system.
They are usually provided by the manufacturers to describe the flange-to-flange
performance of the machine. See Fig. 2.10 for a comparison of the typical curves of
centrifugal, axial, and reciprocating machines.
Performance curves of centrifugal machines typically exhibit an inverse rela-
tionship for centrifugal machines (see the basic slope in Sect. 2.3.3 for a detailed
explanation). At a constant speed, the head decreases as the flow increases. Axial
machines behave similarly to centrifugal machines but have steeper performance
Fig. 2.10 Pump performance curve at constant speed

Fig. 2.11 Pump performance curves: fixed speed versus variable speed
curves.7 Nevertheless, unless mentioned otherwise, most discussions on centrifugal

machines are also valid for axial machines.
The performance of a reciprocating machine can be described with the same head-
flow coordinate system, but the head and flow relationship is drastically different. The
flow rate is determined by the volume of the cylinder and remains almost8 constant
if the stroking speed of the piston is kept constant. The head is determined mainly
by the back pressure and motor power limit, independent of the flow. The resulting
performance curve is close to a vertical line, as shown in Fig. 2.10.
Depending on the driver system, a pump or compressor can operate at a fixed or
variable speed. For example, a fixed-speed motor driving a compressor will result
in a fixed-speed compressor. The same fixed-speed motor driving the same com-
pressor through a variable-speed gearbox (VSGB) can achieve continuously varying
speeds (within a reasonable range) and thus make the compressor a variable-speed
compressor.
The performance of a fixed-speed machine is described with only one curve (see
Figs. 2.11a and 2.12a), while a variable-speed machine with multiple performance
curves (see Figs. 2.11b and 2.12b). The speed change can be continuous9 or discrete,
depending on the driver.
The mechanical construction of the machine determines the performance curve.
Once the machine is built, the curves remain unchanged, barring mechanical modi-
fications, fouling, or wear and tear.
The vendor-provided performance curves typically have additional performance
information in addition to the head-flow relationship, such as head/efficiency/power
against mass/volume flow. Typically the operating limits of the machine speed
and flow rate are also shown on the performance curves. An example of pump
7 Steeper performance curves are better suited for constant-flow applications than for constant-
pressure applications (Liptak 2006).
8 The curve slightly bends to the left at high head values due to higher leakage and losses at higher
pressures.
9 Even if the machine speed can vary continuously, the performance data (head-flow) are supplied
at only a few typical speeds. Those in between can be readily inferred via interpolation.
Fig. 2.12 Compressor performance curves: fixed speed versus variable speed
Fig. 2.13 Typical pump curves (reproduced, with permission, from Niu and Xiao (2022))
performance curves is provided in Fig. 2.13, while Fig. 2.14 is an example of com-
pressor performance curves.
Multiple machines can operate in parallel or series to meet higher demands on the
flow or head. Machines share the same pressure ratio when operating in parallel and
have the same (mass) flow rate when operating in series.10 Figure 2.15 illustrates the
composite performance curves of two compressors in parallel and series. The same
concept can be extended to pumps and reciprocating machines (Golden et al. 2002).
A comparison of Fig. 2.11 with Figs. 2.12 and 2.13 with Fig. 2.14 reveals that
the performance curves of centrifugal compressors and pumps are very similar. For
this reason, we will treat compressors and pumps with a unified approach. Many
discussions on compressors can be applied to pumps. On the other hand, there are
10 There may be liquid knockouts or side streams between compressor stages.

Fig. 2.14 Typical compressor curves
sufficient differences between the two types of machines, and separate discussions
are offered as necessary.
System Resistance Curves
The performance curve defines the head-flow profile on the machine side, while
the system resistance curve describes the pressure-flow profile on the process side.
The term “system resistance” refers to the “resistance to flow,” caused by all the
components in the flow line. The resistance curve defines how the pressure is affected
by the flow-resisting components.
Since the head and pressure ratio can be easily converted from one to another (see
Sect. 2.2.4), the pressure and flow relationship can be plotted on the same head-flow
Fig. 2.15 Composite performance curves
coordinate system as the performance curves. This representation shows a parabolic

shape, as depicted in Fig. 2.16.11
Unlike the equipment performance curve, which is fixed once the machine is built,
the resistance curve is determined by the process flow resistance components and
can be adjusted by changing the flow resistance. Some components, such as piping
and equipment, are fixed once the plant is built.12 In contrast, other components, like
the control valves and inlet guide vanes, are meant to be manipulated and changed.
Therefore, the resistance curve can be “moved” by adjusting the system resistance.
There are several ways to adjust the system resistance, including:
1. Throttling the suction or discharge valve. Closing the discharge valve increases the
system resistance, causing the resistance curves to move to the left.
11 The resistance curve is theoretical and imaginary, assuming ideal conditions. It is rarely exactly
known in a practical setting.
12 The resistance from some flow components, such as filters and exchangers, may change with
time due to fouling and aging.

Fig. 2.16 System resistance curve and process dynamics
Opening the valve decreases the system resistance, allowing more flow to pass
through and causing the resistance curve to move to the right. See Fig. 2.16b.
2. Opening the recycle valve. Opening the recycle valve reduces the system resis-
tance, allowing more flow to pass through the machine (but not necessarily to the
downstream process).
3. Changing the back pressure. The source and destination pressures of the fluid can
cause changes in system resistance, thus affecting the flow rate.
Changing the system resistance is a primary means of compressor operation and
control and will be discussed in detail in later chapters.
2.2.3 One Point: Operating Point
The operating objective of a pump or compressor is to deliver the required flow at

the desired head (or pressure). The head and flow constitute the operating point.
The machine’s performance curves dictate where the operating point can be, while
the resistance curves, imposed by the process, determine where the operating point
should be. They jointly define where the operating point can operate and what tra-
jectory the operating point can follow.
The operating point can be viewed as the equilibrium between the energy required
by the process and the energy produced by the machine. When a pump or compressor
is connected to the process, the operation must meet the requirements of both. As a
result, the actual operating point is always at the intersection of the performance and
resistance curves.
The performance curve remains unchanged once the machine is constructed, while
the system resistance curve can shift left and right in response to resistance changes
in the process components on the flow path. As a result, the operating point of a
fixed-speed machine can only move along a single performance curve via resistance
Fig. 2.17 Operating point on performance and resistance curves
changes (see Fig. 2.17a). The operating point of a variable-speed machine can move
along both the resistance curve and the performance curve, responding to changes
in the machine speed and system resistance. See Fig. 2.17b. Changing the machine
speed and system resistance together can “move” the operating point to anywhere
on the performance map (subject to constraints and limits).
Due to the reverse relationship between the head and flow in the machine
(described by the performance curve) and the direct relationship between the pres-
sure and flow in the piping (described by the resistance curve), the operating point of
a centrifugal machine is somewhat self-regulating. With a change in the head or flow,
the operating point will be pulled in opposite directions because of the opposite signs
of the slope of the two curves. The machine will eventually reach a new equilibrium
and settle at a new operating point by itself, provided that the new point is still within
the operating limits and constraints.
The operating point is described by the three basic variables: speed, head, and flow.
The three variables are correlated. Any two of them can fully define the operating
point on the head-flow-speed performance map. On the other hand, the operating
point can be “forced” to move to a new location by changing any one or two of
the three basic variables using the available control handles. This basic principle of
pump and compressor operation is discussed in detail in Chap. 3.
2.2.4 † Other Variables and Curves
Several other variables supplement the description of machine performance. These

variables will be used in various contexts in later chapters.
Power W
Power is defined as work per unit of time, with a typical engineering unit of kilowatt
(kW) or horsepower (hp). The power consumption for compressing a given gas flow
can be calculated by integrating the volumetric flow over the pressure:
d
W = Fv d P, (2.15)
s
where “s” means suction and “d” discharge. The gas law relates the pressure P and
volume flow Fv of a polytropic process by Eq. 2.10 as
P · Fvn = Ps · Fv,s
n
= const, (2.16)
the gas power is then given by

d
W = Fv d P
s
1
d n
Fv,s Ps n
= dP
s P
1 d
P− n d P
1
= Fv,s Psn
s
n−1
n Pd n
= Fv,s · Ps −1 . (2.17)
n−1 Ps
A centrifugal pump can be treated as a special case with n = ∞ and Fv = Fv,s ,

which leads to
n−1
n Pd n n→∞
W = Fv,s · Ps −1 ≈ Fv,s · (Pd − Ps ) . (2.18)
n−1 Ps
Equation 2.18 is the same result as directly derived from Eq. 2.15 with a constant
volumetric flow Fv independent of the pressure P.
The power can refer to fluid power (hydraulic horsepower), shaft power (or brake
horsepower), or motor power (or drive horsepower), depending on what losses are
included in the calculation. Typically,
[Motor Power] > [Shaft Power] > [Gas Power] (2.19)

[Shaft Power] = [Gas Power] + [Mechanical Losses]. (2.20)
The gas power is the work done on the total flow, while the head is the work per
unit of mass weight. From Eq. 2.17:
W
Hp =
Fm
n−1
Fv,s n Pd n
= · Ps −1
Fm n − 1 Ps
n−1
1 n Pd n
= · Ps −1
ρs n − 1 Ps
n−1
n Ps Pd n
= −1
n − 1 ρs Ps
n−1
n Z s R Ts Pd n
= −1 (2.21)
n−1 M Ps
Ps M
with: ρs = .
Z s R Ts
For isentropic compression, the polytropic component n in Eq. 2.21 is substituted

with the isentropic component κ, which equals the ratio of specific heat of the gas
κ = C p /Cv :
κ−1
κ Z s R Ts Pd κ
Hisen = −1 . (2.22)
κ −1 M Ps
It is more accurate to use the average value (Z s + Z d )/2 in place of Z s . However,

the compressibility Z s and Z d are functions of pressures and temperatures and cannot
be measured directly. Fortunately, Z s and Z d remain reasonably constant within the
operating range of most applications (Mirsky et al. 2012).
Polytropic Efficiency η p
The power calculated with Eq. 2.15 is the net power transferred to the fluid and is thus
called the gas power in the case of gas compression. The power required is always
greater than what is transferred to the fluid due to hydraulic inefficiencies, driver
inefficiencies, and mechanical losses. The difference between the power delivered
by the driver and the power received by the fluid is represented by the efficiency term
η, a ratio of the ideal energy required to the actual energy delivered.
The best efficiency point (BEP) is illustrated in Fig. 2.18. It is desirable to design
or select the machine so that the normal operating point is at or near the best efficiency
point.13
13 In many applications, the optimal operating point tends to be close to the operating limits. In the
case of pumps and compressors, however, the primary operating limits are the surge points, which
differ from the best efficiency point.
Fig. 2.18 Best efficiency point and efficiency ellipse
The efficiency decreases as the operating point moves away from the BEP. The
constant efficiency line exhibits an elliptical shape, with its long axis approximately
parallel to the constant-resistance line. Therefore, it is sometimes also called the
efficiency ellipse. See Fig. 2.18.
The precise calculation of the polytropic efficiency for real gas is complicated and
requires equation of state (EOS) data. One practical approximation of the polytropic
efficiency without elaborate EOS data is as follows:

Hp κ −1 n−1
ηp = = , (2.23)
Hactual κ n
where κ is the ratio of specific heat (assuming to be known from gas analysis), and
n is the polytropic exponent in Table 2.1.
Equation 2.23 indicates that when the compression process approaches isentropic,
with n approaches k, the polytropic efficiency η p approaches 100% (which is not
achievable in reality).
Pressure Ratio (Pd /Ps ) and Pressure Rise (Pd − Ps )
The fluid head is not directly measurable. In practice, the pressure ratio (Pd /Ps ) or
pressure rise (Pd − Ps ) across the machine is usually used in place of the head for
calculation.
From Eq. 2.21, the pressure ratio is related to the head as follows:
n−1
n n−1
n
Pd n − 1 ρs n−1 M
= Hp + 1 = Hp + 1 . (2.24)
Ps n Ps n Z s R Ts
Pressure rise (Pd − Ps ) is another commonly used variable to describe energy.

The pressure rise is related to the pressure ratio simply by
Pd
Pd − Ps = Ps −1 . (2.25)
Ps
For centrifugal pumps, assuming an isochoric process with n = ∞, the pump

head is given from Eqs. 2.21 and 2.24 as

Ps Pd 1
H= −1 = (Pd − Ps ) (2.26)
ρs Ps ρs
Pd H
= ρs +1 (2.27)
Ps Ps
Pd − Ps = ρs H = ρs g h, (2.28)
where H is the head in kJ/kg, and h is the head in meters. Equation 2.28 is the same
as Eq. 2.14.
Polytropic Exponent n
For a centrifugal compressor, the polytropic exponent n is a critical variable indicating

the type of compression cycle the gas goes through (see Table 2.1) and thus is
indispensable in most analyses and calculations. The exact value is determined by
the machine’s mechanical construction and the gas’s physical property and is difficult
to know. However, an approximate value can be calculated from the pressure and
temperature of the gas before and after compression.
Assume a polytropic compression process, with Ps and Ts the pressure and temper-
ature at the suction side and Pd and Td the pressure and temperature at the discharge
side. The polytropic exponent n is given in Eq. 2.10 as
Pd
ln
Ps
n
Ps Fv,s = Pd Fv,d
n
⇒ n= . (2.29)
Fv,s
ln
Fv,d
From the law of mass conservation, assuming Ms = Md , we have

Fm = ρs Fv,s = ρd Fv,d ⇒
Pd Md
Fv,s ρd Z d R Td Pd Ts Z s
= = ≈ . (2.30)
Fv,d ρs Ps Ms Ps Td Z d
Z s R Ts
The polytropic exponent n is then given from Eq. 2.29 as
Pd
ln
Ps
n= . (2.31)
Pd Ts Z s
ln · ·
Ps Td Z d
The compressibility factor Z is introduced for real gases to account for the non-
ideal property, but the exact values are rarely available without rigorous state-of-
equation calculation. The result of n from Eq. 2.31 is typically very unreliable. In
practice, within the normal range of most applications, the value of Z does not change
drastically from suction to discharge. Therefore, a constant Z d /Z s = 1 is acceptable.
The polytropic exponent n is thus simplified to
Pd
ln
Ps
n= . (2.32)
Pd Ts
ln ·
Ps Td
The polytropic exponent can also be computed from the polytropic efficiency η
(if available) from Eq. 2.23 since polytropic efficiency is closely related to the type
of compression cycle:
κ ηp
n= . (2.33)
κ ηp − κ + 1
A related variable that is frequently used for convenience is the polytropic index
σ derived from Eq. 2.32 (Mirsky et al. 2012):
Td
ln
n−1 Ts
σ = = . (2.34)
n Pd
ln
Ps
2.3 Behaviors of Dynamic Machines 49
2.3 Behaviors of Dynamic Machines
The dynamic behavior of the machine refers to the cause-and-effect relationships

between, and the transient responses of, the machine variables, such as the head,
flow, and speed, which profoundly impact the operation and control strategy. In
addition, because the pumps and compressors are integral parts of the process flow
system, the dynamics include both the machine and process dynamics. Therefore, it
is crucial to understand the machine’s behavior in the process context (Boyce 1993,
Gresh 2001).
It is worth noting that the perspective and needs of a process control engineer
differ significantly from that of a process engineer or equipment engineer, and the
essential knowledge of the pump and compressor characteristics is thus also quite
different.
2.3.1 Relationship Among Head, Flow, and Speed
Inside a dynamic machine, the mechanical energy from the driver is transferred to the
fluid and then converted to potential energy (as pressure increases) and kinetic energy
(reflected by flow increases). The energy transfer and conversion can be conveniently
described by the three basic variables: head, flow, and speed.
Speed is an independent variable determined by the external driver system.
Changes in machine speed can simultaneously affect the head and flow of the
machine. However, the impact depends on whether the machine is reciprocating
or centrifugal. For instance, the volumetric flow rate of a reciprocating machine is
a function of the machine’s speed and is independent of the head. That is, a fixed-
speed reciprocating machine will deliver a fixed volumetric flow rate regardless of the
fluid type and pressure/temperature. In contrast, the volumetric flow of a fixed-speed
centrifugal machine is determined by the machine’s speed and also the energy split
between the potential (head) and kinetic (flow) energy. See Fig. 2.19 for a simple
illustration of this cause-and-effect relationship.
Fig. 2.19 Cause-and-effect relationship among speed, head, and flow

The relationship between a centrifugal machine’s power, head, flow, and speed
can be described by the so-called affinity laws, popularly known as the fan laws.14
The relationship among the variables can be expressed as:
For a given centrifugal machine with a fixed-diameter impeller, the capacity will be directly
proportional to the speed, the head will be directly proportional to the square of the speed,
and the required power will be directly proportional to the cube of the speed.
The affinity laws indicate that if the machine speed changes, the flow, head, and
power vary by the speed ratio’s first, second, and third power. In equation format:
3 2
W2 N2 H2 N2 Fv,2 N2
= , = , = , (2.35)
W1 N1 H1 N1 Fv,1 N1
where W is the power, N is the speed, H is the head, and Fv is the volume flow rate.
Subscripts “1 ” and “2 ” denote two operating points. In another format:
2 3
Fv1 Fm1 N1 H p1 N1 W1 N1
∝ ∝ , ∝ , ∝ . (2.36)
Fv2 Fm2 N2 H p2 N2 W2 N2
A re-examination of Fig. 2.16 and Eq. 2.16 demonstrates the logic of this law. For
example, if the speed of the dynamic machine is increased by 10% (i.e., N2 /N1 =
1.10), the changes in flow, head, and power are given by15
Volumetric flow increases by 10%: Fv,2 /Fv,1 = 1.10

Head increases by 21%: H2 /H1 = 1.12 = 1.21
Power consumption increases by 33%: W2 /W1 = 1.13 = 1.33.
The effect of the affinity laws on flow, head, and power is illustrated in Fig. 2.20.
Due to the different responses by head and power, many energy-saving opportunities
exist to optimize the operation of a complex system consisting of multiple pumps or
compressors. See Chap. 7 for some examples.
14 Note that the term “fan laws” is a misnomer. The term law means a principle proven to be true
for all cases. Unlike other physical laws, fan laws are only an approximation with limitations.
They are well applicable to low-head single-stage compressors, including fans and blowers. The
accuracy can be significantly affected by many factors, such as the molecular weight of the gas,
high discharge pressure, the backlean angle of the impellers, and the number of compression stages
in the compressor (Brown 1991).
15 A common revamping activity for centrifugal machines is re-wheeling, which involves reducing
the diameter of the impellers. A change in the impeller diameter has the same effect as a change in
speed since the tip speed of the impeller is proportional to its diameter.
Fig. 2.20 Flow, head, and power under affinity laws
2.3.2 † Suction and Discharge Relationship
Many changes occur when the fluid travels from suction to discharge inside a pump
or compressor. These changes are reflected in critical process variables such as head,
flow, pressure, and temperature.
By the law of conservation of mass, the mass flow rate remains the same from
suction to discharge. The volumetric flow can also be assumed to be the same for
pumps working on incompressible liquids since the density remains constant. For
compressors handling compressible gas, however, a significant increase in density
occurs during compression; thus, the volumetric flow rate becomes significantly
lower from the suction to discharge, even though the mass flow remains the same.
The changes are closely related to the gas property and operating condition:
Fm = Fm,s = Fm,d = ρs Fv,s = ρd Fv,d (2.37)

Ps M
− n1 − n−1
1
Fv,d ρs Z s R Ts Ps Td Pd Td
= = ≈ = = , (2.38)
Fv,s ρd Pd M Pd Ts Ps Ts
Z d R Td
where Z s = Z d is assumed, which is not exactly true but usually acceptable. The
accurate profile of the compressibility value inside the compressor is challenging to
obtain.
A related flow definition is the standard flow rate, the volumetric flow rate at a
standard condition of P0 = 101.3 kPa and T0 = 15 ◦ C = 288.15 K. The standard
flow is a volumetric flow. However, it is the same concept as mass flow since the flow
condition is fixed. When describing the processing capacity or calculating power
consumption, it is customary to use either mass flow or standard flow since they are
independent of the specific operating condition. See Niu and Xiao (2022) for the
different types of flows.
The discharge pressure can be related to the suction pressure via the temperature
ratio (see Eq. 2.10):
n
Pd Td n−1
= . (2.39)
Ps Ts
The suction and discharge pressures can also be related via the polytropic head
with Eq. 2.24:
n−1
n
n−1 M
Pd = Ps · Hp +1
n Z s R Ts
σ1
ρs
= Ps · σ H p · +1 . (2.40)
Ps
The calculation of discharge pressure Pd from suction pressure Ps requires the

value of the polytropic exponent n or polytropic index σ , which in turn requires
the ratio of specific heat κ and efficiency η. As discussed later, the unavailability
or inaccuracy in the polytropic index n is a significant challenge in compressor
performance analysis and control design. An acceptable approximation is needed.
For a centrifugal pump, assuming an isochoric process with n = ∞ (i.e., σ = 1),
Eq. 2.40 reduces to
ρs
Pd = Ps · H +1
Ps
= Ps + ρs · H
= Ps + ρs · g · h, (2.41)
where h is the head in height. Equation 2.41 is the same as Bernoulli’s equation in
Eq. 2.3, assuming the same flow velocity from suction to discharge.
Pressure rise causes the temperature to increase. For centrifugal compressors,
temperatures at the suction and discharge side are related by
n−1 n−1
Td Pd n Pd n
= → Td = Ts · . (2.42)
Ts Ps Ps
Example 2.2 Changes in pressure and temperature during centrifugal compression.

For natural gas with n = 1.4, assuming a compression ratio of 3, the volumetric flow
rate at discharge is given by Eq. 2.38 as
− n1
Fv,d Pd
=
Fv,s Ps
= 3− 1.4 = 0.456 = 45.6%,
1
(2.43)
which is less than half of the suction flow rate, although the mass flow rate is the
same.
The primary concern for a higher compression ratio is the excessively high tem-
perature at discharge which can cause damage to the device and piping. Typically
the discharge temperature should not exceed 180 ◦ C (or 350 ◦ F).
As shown in Eq. 2.42, discharge temperature increases with pressure, which is
primarily caused by inlet temperature, compression ratio, and gas composition. Cor-
respondingly, the most practical measures for limiting discharge temperature are
providing adequate gas cooling and limiting the compressor ratio.
Assume the suction temperature is at Ts = 60 ◦ C. A compression ratio of 3 will
result in a discharge temperature as
n−1
Pd n
Td = Ts ·
Ps
1.4−1
= (60 + 273.15) × (3) 1.4
= 456.0 K = 182.8 ◦ C. (2.44)
If we would like to limit the discharge temperature to 150 ◦ C, the maximum

compression ratio can be calculated from Eq. 2.39 as
n
Pd Td n−1
=
Ps Ts
1.4
150 + 273.15 1.4−1
=
60 + 273.15
= 2.31. (2.45)
2.3.3 ‡ Euler’s Equation and Slope of Performance Curve
The machine’s performance is determined by its geometric construction, which is

dictated by fundamental laws like Euler’s equation. The shape and slope of the per-
formance curve are critical factors in assessing the machine performance, with a
profound impact on the machine’s behavior, impacting its operation and control.
Euler’s Equation
The fluid energy, represented by the fluid head, can be calculated from the change in
angular momentum of the fluid entering and exiting the impeller. Based on Newton’s
third law of motion, the fluid’s rate of change in angular momentum equals the net
torque imposed on the fluid by the rotor (Liptak 2006; Elliott and Bloch 2021):
Fm
τ= (r1 V1u − r2 V2u ) , (2.46)
g
where τ is the shaft torque, Fm is the mass flow, V1u and V2u are the tangential
components of the fluid velocity, r1 and r2 are the impeller-eye radii and outside
diameter hub radius. g is the gravitational constant.
The energy E transferred from the rotor to the fluid is thus given by
E =τω
Fm
= ω (r2 V2u − r1 V1u )
g
Fm
= (U2 V2u − U1 V1u ) (2.47)
g
with U1 = ω r1 , U2 = ω r2 ,
where U1 is the inlet tangential velocity, and U2 is the exit tangential velocity at
the impeller outside diameter. Equation 2.47 is the basis of nearly all performance
characteristics for all forms of turbo-machinery and is known as Euler’s Equation. It
states that the energy transferred between the rotor and the fluid can be accounted for
by the difference between the product U1 · V1u at the entry and the product U2 · V2u
at the exit of the impeller. In other words, the rate of change in angular momentum
of the fluid equals the net torque imposed on the fluid by the rotor.
The head, defined as the energy produced per unit of mass flow, is given by
E 1
H =− = (U2 V2u − U1 V1u ) . (2.48)
Fm g
Usually, the impeller is so designed that the fluid entry is axial, i.e., V1u = 0 and
V1w = V1 (see Fig. 2.21), then Euler’s Equation would reduce to
1
H= U2 V2u . (2.49)
g
The head produced is a function of the impeller tip speed and the gas exit velocity
relative to the blade, which is proportional to the volumetric flow rate.
For an impeller with back lean angle β2 , the component of the fluid velocity V2u
is given by
V2u = U2 − V2w cot(β2 ), (2.50)
therefore,
1
H= U2 (U2 − V2w cot(β2 )) , (2.51)
g
where V2u is the exit velocity of the fluid and is related to the volumetric flow Fv,2 via
V2w = Fv,2 /A2 , with A2 being the cross-sectional area at the outlet. We then have
1 Fv,2
H= U2 U2 − cos(β2 ) . (2.52)
g A2
This equation shows that at a constant speed, Euler’s head is a function of the vol-
umetric flow rate and nothing else. In other words, for a given diameter impeller at
a given speed, a fixed amount of energy (kJ/kg) is transferred to each mass unit of
fluid, regardless of the density of the fluid.
Note that if the backlean angle β2 = 90◦ , and cot β2 = cot 90◦ = 0, the head
generated would be a function of the blade tip speed alone:
1 2
H= U . (2.53)
g 2
Basic Slope
The performance curves we have shown so far all have a negative slope under the
head-flow coordinate system, which is not always true. The slope of the curve is dic-
tated by the underlining mechanical design and physical principles, with the impeller
being the determining factor.
Each centrifugal machine can have multiple sections, each section has multiple
stages, and each stage has multiple impellers. An impeller consists of multiple rotat-
ing vanes to impart mechanical energy to the fluid. In theory, the performance curve
is per impeller, and the curve differs from impeller to impeller. In practice, however,
composite performance curves are generally provided for each stage. Because no heat
is removed between impellers, the composite performance curves are acceptable as
a good trade-off between complexity and accuracy.
There are many different designs of impellers. One critical factor affecting the
machine’s performance is the backlean angle β2 , as shown in Fig. 2.21a. This angle
determines the slope of the performance curves.
For different β2 values, the so-called basic slope, which is the ideal or theoreti-
cal slope of the head-flow relationship without considering any losses, is shown in
Fig. 2.21b.
For radial exit with β2 = 90◦ , the basic slope is zero, corresponding to a flat
head-flow line. With a backlean angle β2 > 90◦ , the basic slope would be positive,
indicating that more flow will generate more head. See Fig. 2.21b for an illustration.
With a backward lean angle β2 < 90◦ , also called “back-sweeping,” the perfor-
mance line between the head and flow has a negative slope, which indicates that the
more flow through the machine, the less head it will generate. In addition, a smaller
angle will result in steeper curves, and vice versa.
This relationship can be revealed from the illustration in Fig. 2.22, where it is
seen that more flow implies a higher relative speed from W2 to W2 . The absolute
exit speed moves from V2 to V2 , resulting in a smaller value in V2u . According to
Fig. 2.21 Impellers and Euler’s equation
Fig. 2.22 Reverse relationship between head and flow
Euler’s equation, the generated head is the product of the tip speed U2 , and the flow
component V2u . Therefore, a larger flow rate will produce a smaller head value of H
at the same speed. See Fig. 2.22a.
Similarly, less flow throughput implies a lower exit speed, and thus, a more promi-
nent flow component for the head calculation. The result is a higher head with less
flow, as shown in Fig. 2.22b.
This phenomenon is as expected. With a fixed blade tip speed, the head is deter-
mined by the relative velocity of the gas. A higher flow implies a higher gas velocity
Fig. 2.23 Reverse relationship between head and flow for radial blades
exiting the impeller. The higher the velocity, the less time the gas is in contact with
the impeller blade and the less energy it will pick up from the blade. As a result, a
lower pressure will be developed.
Most centrifugal machines in the process industry choose backward-leaning
blades (with β2 between 55 and 75◦ ) because their characteristic curves with a neg-
ative slope provide a better differential head and a wider stable range for process
control (Boyce et al. 1983).
For comparison, Fig. 2.23 shows the head and flow for impellers with radial exit.
With β2 = 90◦ , the gas exits the impeller in the radial direction. The flow component
V2u remains constant with changing flow velocities. Therefore, the head generated
becomes a function of the tip speed only, unaffected by the flow.
Actual Performance Curves
In reality, there are inevitable aerodynamic and mechanical losses16 during the energy
transfer from the rotor to the fluid, including, for instance, fluid separation from the
blade surface causing a decrease in exit whirl velocity or slip, fluid-wall friction, and
turbulence (Boyce et al. 1983).
This separation of the fluid from the blade’s exit angle is called slip. As a result,
the actual performance curve deviates significantly from the theoretical curves in
Fig. 2.22, especially at both ends. It resembles what is illustrated in Fig. 2.24.
16Mechanical losses affect the power input to the machine, but do not influence the shape of the
head-flow curve.
Fig. 2.24 “Basic slope” and performance curves
Moreover, the head/flow lines derived from the Euler equation assume that the
fluid velocity is uniform across the impeller cross section and exits along a path
guided by the blade’s exit angle. However, the fluid does not exit perfectly along
the blade’s exit angle since flow separation between the fluid and the blade’s surface
takes place long before the fluid exits.
It is virtually impossible to know the exact shape of the actual performance curves.
The qualitative description based on the manufacturer-provided performance curves
is sufficient for most applications.
2.3.4 ‡ System Resistance and Slope of Resistance Curve
The process around the pump and compressor comprises two parts: the suction system
and the discharge system. The objective of the machine is to remove (or “pull”) the
fluid from the suction vessels at the same rate as it enters that vessel. Similarly, the
discharge system “pushes” the fluid against the discharge system resistance to the
downstream operation to achieve the desired discharge pressure. The discharge head
corresponds to the external energy required and is the net effect of the total resistance
from both the discharge and suction systems.
Figure 2.25 illustrates the resistance components of a simplistic single-
stage single-train compression process, including ➀ piping, heat exchangers, fil-
ters/strainers, and non-return valves; ➁ control valves; ➂ back pressure; and the
line-up of the flow path. The diffuser of the compressor is another major flow-resistant
component.
For the fluid to flow forward, a pressure gradient must exist. Assume any two
points on the flow path, with pressure P1 and P2 , respectively (see Fig. 2.26a), and
the flow resistance is lumped into a single component represented by an imaginary
Fig. 2.25 System resistance and process flow components (reproduced, with permission, from Niu
and Xiao (2022))
Fig. 2.26 Flow and system resistances
choke valve or equivalent orifice. The relationship between the flow rate and pressure
gradient is given by

Fm = C ρ (P1 − P2 ) (2.54)
where Fm is the mass flow rate, and ρ is the fluid density.

Flow constant C is a relative measure of the efficiency of the flow-resisting com-
ponent between the two points on the flow path.
To achieve a flow rate of Fm , the pressure difference between the two points must
meet the following requirement:
1 Fm2
P2 − P1 = . (2.55)
C2 ρ
In many operations, the pressure difference P1 − P2 is not large enough to produce

the desired flow rate. For example, in an oil and gas processing facility, the fluid often
must be processed and transferred from a lower pressure point to a higher pressure
point, e.g., the main oil line in Fig. 1.6. The required pressure gradient is provided
by a pump for liquid or a compressor for gas, as shown in Fig. 2.26b, where Ps and
Pd are the suction and discharge pressures of the pump, respectively, with Pd > Ps .
Assume that the suction and discharge systems are simplified into two equiva-
lent choke valves for convenience of understanding. By conservation of mass (see
Eq. 2.1), the pressure profile must meet the following requirements:
⎧
⎪ 1 Fm2
⎪
⎨ P1 − Ps = 2
C1 ρ1
(2.56)
⎪
⎪ 1 Fm2
⎩ Pd − P2 = 2 .
C2 ρ2
The pressure boost provided by the pump or compressor is a function of the source
pressure P1 , the destination pressure P2 , the fluid density ρ, and the flow constants C:
1 1
Pd − Ps = + Fm2 − (P1 − P2 ). (2.57)
C12 ρ1 C22 ρ2
This relationship indicates that the pressure rise in the pump or compressor is
approximately a quadratic function of the flow rate. The source and destination
pressure affect the relationship via the P1 − P2 term. At the same time, the resistance
changes in the flow components, such as the throttling valves, are reflected in the
flow constant C1 and C2 in Eq. 2.57.
For example, suppose a fluid with constant density ρ is transferred from one
storage tank to another under atmospheric pressure (P1 = P2 ). In that case, the pump
must provide the pressure gradient to overcome the flow resistance to produce the
desired flow rate of Fm :
1 1 1
Pd − Ps = + 2 Fm2 . (2.58)
ρ C12 C2
The pressure rise generated by the pump is a quadratic function of the flow rate,
similar to any flow component on the flow path shown in Eq. 2.55.
The system resistance curve is a theoretical concept. If the back pressure remains
constant and all valves (throttling and recycling) remain unchanged, the result is a
constant-resistance curve. An example of a constant-resistance curve is an electric
fan that draws air from the atmosphere and pushes the air to the atmosphere. Similarly,
the resistance curve resembles one constant-resistance curve if the compressed gas
is discharged to the main gas line of relatively infinite capacity.
In practice, however, the system resistance rarely remains constant. There are
almost always multiple factors affecting system resistance. As a result, the actual
2.4 Surge and Choke Phenomena 61
trajectory of the operating point is a combined effect. For instance, if the discharge
gas is sent to a confined space like a storage tank, such as buffering tank of instrument
air, the back pressure increases as more gas is received, the system resistance will
increase with time, and the resistance curve on the performance map will shift to the
left as the pressure increases. We will call the trajectory the actual-resistance curve17
to distinguish it from the constant-resistance curves.
2.4 Surge and Choke Phenomena
The flow through a centrifugal machine has an inherent operating range. When the
flow falls below a specific limit, the machine may go into surge, an unstable working
condition.
Centrifugal machines are also constrained by the maximum flow that they can
deliver. The maximum flow is attained when the flow reaches the velocity of sound
and is said to have choked.18 For centrifugal compressors, the choked flow limit is
also called the stonewall.
2.4.1 Surge and Choke Points
A surge in a centrifugal compressor is a result of flow separation from the flow

surface caused by low gas velocity. It can occur anywhere on the flow path in the
compressor. Centrifugal pumps experience the same separation phenomenon at low
flows, causing the liquid to vaporize, resulting in cavitation.
The surge point on the performance curve represents the minimum flow value
for stable operation. At a flow rate below the stability limit, the head produced by
the compressor cannot overcome the system resistance to maintain a forward flow,
resulting in temporary backflow. This resistance is imposed by the process and is
reflected by the pressure difference across the machine. The thrust reversal can lead
to violent vibrations to cause mechanical damage to the machine.
The surge point is located at the peak of the curve, as shown in Fig. 2.27a, and
marks the head/flow condition where the surge phenomenon may start. To the right of
the surge point, the slope of the performance curve is negative. The slope approaches
zero as the operating point approaches the surge point. Over the surge point, the slope
turns positive.
17 The trajectory of the operating point for a compressor sending gas to a confined space is also
called the integrative resistance curve in some literature, e.g., Boyce et al. (1983), Kurz et al. (2016)
since the pressure increases as more gas is brought in.
18 Choked flow at compressor discharge resembles critical flow through an orifice. When choked,
the volumetric flow does not increase with pressure increases (the mass flow can still increase
√ with
higher pressure or higher gas density). The speed of sound at the inlet is given by s = gκ Z RTs
where κ is the isentropic exponent, and Ts is the inlet temperature.
Fig. 2.27 Slope of the performance curves
Similarly, as the operating point moves to the right, the performance curve
becomes steeper (i.e., more negative slope). The performance curve becomes almost
vertical after a certain point, and the head-producing capacity falls off rapidly, with
low efficiency and unpredictable performance, thus deemed unsuitable for normal
operation. The cutoff point for this low-efficiency operation is the stonewall point or
choke point.
Vendor-provided performance curves customarily only provide the segment that
can be operated stably, i.e., from surge point to choke point. Therefore, on almost all
commercially available performance curves, the surge point is on the extreme left,
and the choke point is on the extreme right, see Fig. 2.27b. See also Fig. 2.14 for an
example of vendor-provided performance curves.
The positive slope on the head-flow coordinate system indicates resistance to flow,
while a negative slope indicates the opposite. The system resistance curve always has
a positive slope, while the slope of the performance curve can be positive or negative.
When the operating point is to the right of the surge point, the negative slope of the
performance curve and the positive slope of the resistance curve counteract each other
and provide a self-regulating condition for the operating point to reach equilibrium
by itself. Once the operating point moves left over the surge point, both the resistance
and performance curves have positive slopes; the machine loses the energy to counter
the system resistance. The operation quickly becomes unstable and is said to have
entered a surge condition.
The surge phenomenon results from the interaction between the compressor and
the process. A compressor in surge fails to meet the operating requirements, and
the rotor’s violent axial and radial movements can also cause severe damage to the
machine. Potential initiators of surge include startup and shutdown, trips, severe turn-
down, fouling, blockage, or drastic changes in operating conditions such as molec-
ular weight. Higher pressure and higher molecular weight applications can result
in more significant damage from a surge. Low-density applications such as hydro-
gen can cause surge damages that may only be detectable until the equipment is
2.4 Surge and Choke Phenomena 63
Fig. 2.28 Compressor surge cycle (reproduced, with permission, from Niu and Xiao (2022))
disassembled. One of the most impressive examples of compressor surge is when

the axial compressor on a jet engine goes into surge (Giampaob 2010).
A surge can be a mild surge or a deep surge. With a deep surge, an insufficient
head may cause the machine flow to reverse direction and flow backward. Once the
sufficient volume is rebuilt, the flow changes direction and moves forward again.
If the low-flow condition persists, the backflow will develop again, leading to rapid
fluctuations in flow and pressure and violent vibrations in the machine and associated
piping. Deep surges can reduce efficiency, create instability, and cause damage to
seals, bearings, and impellers. An accumulated surge of over 30 min is detrimental
to the components, with damage beyond acceptable levels (Day 2016; Hafaifa et al.
2014; Sundström et al. 2018).
Surge is a high-speed phenomenon; the flow reversal can occur in sub-seconds.
Figure 2.28 illustrates a typical surge cycle, where the cycle time is between 300 ms
and 3 s: A → C: 20∼50 ms, D → E: 20∼120 ms, A → · · · → E: 300 ms ∼ 3 s
(Mirsky et al. 2012).
Surge and choke phenomena are consequences of extreme gas velocity in the
machine. Surge is caused by low-flow turbulence, while choke is caused by high-
velocity friction. They can occur in dynamic machines, including centrifugal and
axial pumps and compressors (and blowers). However, there are no such concerns
with reciprocating machines.
An adequately specified compressor has its normal operating points near the best
efficiency point (BEP), well above the minimum flow limit. In other words, a surge19
19 Stall is another terminology related to surge. Reduced flow through a centrifugal machine will
increase the losses in its aerodynamic components, such as the diffuser and impeller, eventually
causing the separation of the fluid from the flow surface. As a result, part of the fluid will not exit
is an abnormal phenomenon under abnormal operational conditions (Gravdahl and

Egeland 1999; McMillan 1983; Hafaifa et al. 2014).
2.4.2 Surge Line and Choke Line
For centrifugal machines operating at multiple speeds, connecting all the surge points
at different machine speeds produces the surge line. Similarly, connecting all the
choke points produces the stonewall line or choke line (Hafaifa et al. 2014). See
Fig. 2.28.
The surge and choke lines are the dividing lines between stable and unstable
operating regions. The operation point is expected to be limited between the surge
and choke lines for stable operation.
The surge line is jagged due to errors and approximations in calculations or mea-
surements. Although the jaggedness poses little challenge to human interpretation,
automatic control requires a smooth line that can be represented with a simple math-
ematical formula, with the following two requirements:
1. A simple mathematical representation of the surge line that can be used by both
human operation and real-time control.
2. The representation must be valid for all the desired operating scenarios with
different pressure, temperature, and gas properties.
These two questions are central to pump and compressor control. The answer to
the first question is the so-called surge reference line, and the second is the concept
of the invariant coordinate system; both are discussed in detail in Chap. 5.
The different approaches to handling the above two questions mark the difference
in the control solutions provided by all commercial control solutions, as discussed
in later chapters.
2.5 Summary
Pumps, compressors, fans, and blowers are prevalent rotating equipment in the pro-
cess industry. They share many similarities in their working principles but differ in
many details.
Process control engineers view these machines differently from rotating equip-
ment engineers or process engineers. Process control concerns mainly the process
variables describing the behaviors of the machines, the cause-and-effect relationship
among the variables, and the dynamic transient responses of these variables.
the machine. Stall is a precursor to surge, although sometimes used interchangeably (Sundström
et al. 2018).
References 65
There are many commonalities in these machines’ dynamic analysis and con-
trol design from this perspective. The common characteristics can be extracted and
visualized with three variables (head, flow, and speed), two curves (performance
and resistance curves), and one point (the operating point). The three basic variables
describe the machine’s characteristics. The two curves represent the cause-and-effect
relationships among the variables. They form the basis for operation and control.
References
Bachus L, Custodio A (2003) Know and understand centrifugal pumps. Elsevier, London
Boyce MP (1993) Principles of operation and performance estimation of centrifugal compressors.
In: Proceedings of the twenty-second turbomachinery symposium, College Station, Texas
Boyce MP, Bohannan WR, Brown RN, Gaston JR, Meher-Homji C, Meier RH, Pobanz NE (1983)
Tutorial session on practical approach to surge and surge control systems. In: Proceedings of the
12th turbomachinery symposium, College Station, Texas, USA
Brown RN (1991) Fan laws, the use and limits in predicting centrifugal compressor off design
performance. In: Proceedings of the 20th turbomachinery symposium, Texas A&M University.
Turbomachinery Laboratories, pp 91–100
Cumpsty N (1989) Compressor aerodynamics. Longman Group UK Limited
Day I (2016) Stall, surge, and 75 years of research. J Turbomach 138(1)
Forsthoffer WE (2005) Forsthoffer’s rotating equipment handbooks, vol. 1: principles of rotating
equipment. Elsevier Ltd
Giampaob T (2010) Compressor handbook: principles and practice. The Fairmont Press
Golden S, Fulton SA, Hanson DW (2002) Understanding centrifugal compressor performance in a
connected process system. Petrol Technol Q
Gravdahl JT, Egeland O (1999) Compressor surge and rotating stall – modeling and control. Springer,
London Limited
Gresh MT (2001) Compressor performance: aerodynamics for the user, 2nd edn. Butterworth Heine-
mann
Hafaifa A, Rachid B, Mouloud G (2014) Modeling of surge phenomena in a centrifugal compressor:
experimental analysis for control. Syst Sci Control Eng 2:632–641
Kurz R, White RC, Brun K, Winkelmann B (2016) Surge control and dynamic behavior for cen-
trifugal gas compressors. In: Proceedings of Asia turbomachinery and pump symposium, Turbo-
machinery Laboratory, Singapore
McMillan GK (1983) Centrifugal and axial compressor control. Instrument Society of America,
Mirsky S, Jacobson W, Tiscornia D, McWhirter J, Zaghloul M (2012) Development and design of
anti-surge and performance control systems for centrifugal compressors. In: Proceedings of the
forty-second turbomachinery symposium, Houston, Texas
Sundström E, Semlitsch B, Mihaescu M (2018) Generation mechanisms of rotating stall and surge
in centrifugal compressors. Flow Turbul Combust 100(April):705–719
Watterson JM (2018) A simple guide to understanding compressors. Momentum Press
Chapter 3
Operating Requirements and Control
Objectives
Pumps and compressors are indispensable components in the process industry. They
play a vital role in maintaining the safety and efficiency of the overall operation.
To achieve optimal performance, it is essential to have a clear understanding of
the control targets and skillful use of the control handles. Equally important is a
good understanding of the machine’s characteristics and dynamic behavior. This
knowledge is vital in predicting how the machine will respond to changes in the
process. Moreover, pumps and compressors are not standalone components; they
are a part of a larger system. The process flow configuration dictates the higher
level operating objectives and strategies. Therefore, it is crucial to comprehend the
machine’s behavior within the process context to ensure it contributes to the overall
efficiency of the process. By considering these factors, operators can ensure that
pumps and compressors operate optimally, resulting in a safer and more efficient
process. With the use of advanced technologies and control systems, pump and
compressor operation has become more sophisticated, and the demand for skilled
operators has increased. This chapter aims to provide readers with the necessary
knowledge to operate these machines successfully.
3.1 Operating Objectives and Requirements
Pumps and compressors are critical elements of a process flow configuration. For
example, in the oil and gas processing facility illustrated in Fig. 3.1, the many pumps
and compressors are responsible for continuously moving the liquid and gas to the
specified location at the specified flow rate and the desired pressure.
1. Operating point. As discussed in Sect. 2.2.3, the actual operating point must
meet the requirements of both the machines and the process. The machine
characteristics dictate where the operating point is allowed, while the process
requirements determine where the operating point is desired.

https://doi.org/10.1007/978-3-031-43122-7_3
68 3 Operating Requirements and Control Objectives
Fig. 3.1 A typical compression process for upstream E&P facility (reproduced, with permission,
from Niu and Xiao (2022))
The operating point is typically specified as the operating pressure and flow rate.
The objective is to keep the operating point at the desired location on the perfor-
mance map.
There are two scenarios in the daily operation of the machine: proactive adjust-
ment of the operating point in response to operating needs and reactive adjustment
to counter the effect of load disturbances and process upsets. When the operating
point moves away from the desired location, it must be brought back by adjusting
the available control handles (LeBleu Jr. and Perez 2014).
2. Operating envelope. The operating point must be kept inside the various limits
and constraints imposed by both the machine and the process, i.e., the operating
envelope. During process upsets, the operating point may approach or breach these
limits, and immediate corrective actions are required to restore the operating point.
The most common constraints include the minimum/maximum flow limits and
minimum/maximum speed limits imposed by the machine, the power and torque
limit imposed by the driver, and the pressure and temperature limits imposed
by the piping and vessels. If the corrective actions are unsuccessful, the process
operation may go out of control; the machine or the processing unit may need to
be proactively shut down to avoid severe consequences.
3. Abnormal situation handling. A pump or compressor is expected to operate under
various modes, including startup, normal operation, crippled operation, turndown
operation, and shutdown, to name a few. Bringing the machine on-stream from
shutdown condition is challenging with high risk and requires careful attention
and extensive experience. Improper startup procedures can cause severe damage
to the equipment.
A partial shutdown of the operation is a common requirement that can occur
intentionally by operating request or unintentionally due to an emergency. During
a partial shutdown, the remaining units are expected to stay in operation, and the
partially operating unit is said to be in crippled mode. The smooth transition to
3.2 Maintaining the Operating Point 69
crippled mode and back to normal mode is a significant challenge for operation
and control.
Standard operation procedures (SOP) are essential to ensure a safe shutdown
with minimal disruptive effects. Automated startup and shutdown sequences have
attracted much attention in practical applications due to their potential to improve
operational safety and efficiency.
3.2 Maintaining the Operating Point
Normal operation requires maintaining the operating point at the desired values to
meet the quality and capacity requirements. The operating point typically refers to
the head and capacity of the machine corresponding to the pressure and flow of the
process. They are the control targets (dependent variables).
The control handles, or independent variables, are the means to influence the
control targets. The three types of control handle for a centrifugal compressor, include
the machine speed (➀), system resistance (➂ and ➃), and recycling (➄), are shown
in Fig. 3.2. An inlet guide vane (IGV, ➁) is another control handle for centrifugal
compressors, but not as widely used.
3.2.1 Operating with Single Control Handle
Most of the time, maintaining the operating point is to maintain the desired flow
throughput, referred to as capacity.
The basis of operation is the cause-and-effect relationships among the variables of
interest, represented by the cause-and-effect table (CET) in Table 3.1, derived from
the characteristics of pumps and compressors discussed in Chap. 2. The column on
Fig. 3.2 Control handles for centrifugal compressor control

Table 3.1 Cause-and-effect table (CET) for pumps and compressors

Centrifugal Machine Reciprocating Machine
Control Handles Head Flow Head Flow
Machine Speed ↑
System Resistance ↑ −
Recycling ↑
Inlet Guide Vane ↑ − −
Stroke Length ↑ − −
Clearance Pocket ↑ − −
the left lists the control handles, and the row on the top the control targets. The
up arrow ↑ indicates a step change in the control handle, while the arrows or
indicate the direction of the response of the control targets to the step change,
assuming only one control handle changes at a time.
All control handles in Table 3.1 may cause changes to the head and flow. Depend-
ing on whether the machine is centrifugal or reciprocating, the changes can be differ-
ent. For instance, increasing machine speed will cause increases in head and flow for
both types of machines. Increasing recycle will result in high flow and lower head.
Changing system resistance affects both flow and head of a centrifugal machine, but
only the head of a reciprocating machine. Inlet guide valve is specific to centrifugal
compressors, while variable stroke length or clearance pockets are for reciprocating
machines only.
Machine Speed
Adjusting the machine speed to change machine flow is the most efficient approach.
For instance, with performance curves illustrated in Figs. 3.3 (for centrifugal
machines) and 3.4 (for reciprocating machines), capacity change from 20.0 kg/s
to 15.0 kg/s can be achieved by reducing the compressor speed. The operating point
moves from the current point A to the new operating point C, along the constant-
resistance curve, assuming all other control handles remain the same.
Since the efficiency ellipse is roughly parallel to the constant-resistance curve,
moving the operating point along the resistance curve does not significantly affect
the operating efficiency.
System Resistance
Most machines operate at a fixed speed. The operating point is maintained by adjust-
ing the system resistance through a control valve on the machine suction or discharge.
Changing the system resistance moves the operating point along the constant-speed
line, illustrated by the trajectory from A to B in Fig. 3.3.
Changing machine capacity by varying system resistance is less efficient than
adjusting the speed because extra energy is needed to raise the pressure, which is to be
Fig. 3.3 Capacity control with centrifugal machines
Fig. 3.4 Capacity control with reciprocating machines
killed downstream. Moreover, moving along the constant-speed performance curve

causes the operating point to move away from the best efficiency point (BEP) (see
Fig. 3.3). However, capacity change with a throttling valve is simple and reliable, with
less capital investment and lower maintenance needs than variable-speed machines.
Changing system resistance has an insignificant influence on the flow of a recip-
rocating machine, as shown by the steep slope of the performance curve in Fig. 3.4.
The slope of the performance and resistance curves affects the easiness of oper-
ation and control. Once entering normal operation, a pump or compressor, whether
centrifugal or reciprocating, is surprisingly reliable and tolerant within a specific
range of operating conditions. However, the operation can be relatively less stable if
both the performance and resistance curves are flat.
Recycling
Operation with recycling is another approach to adjusting the operating point. Three
flows are essential to understanding and operating pumps and compressors: the pro-
cess flow, machine flow, and recycle flow, as shown in Fig. 3.5.
The machine flow is the sum of the process flow and recycle flow:
[Machine Flow] = [Process Flow] + [Recycle Flow]

Fs = F1 + Fr . (3.1)
The machine does not know, nor does it care, whether the flow into the machine
comes from the process or recycle line. For this reason, capacity adjustment can be
achieved by directly recycling a portion of the flow at the discharge side back to
the suction side, thus maintaining the desired flow rate through the machine while
reducing the flow from the process.
The effect of recycling is illustrated in Figs. 3.3 and 3.4 with lines A → D. At
the same machine speed and pressure ratio, the machine flow remains at 20.0 kg/s.
A recycle flow of 5.0 kg/s displaces the process flow and allows the process to
turndown to 15.0 kg/s. In extreme cases, such as during startup, the machine flow
can be entirely from the recycle flow with zero process flow (forward flow)! See
Fig. 3.6 for an illustration.
The recycling operation wastes energy to re-pressurize the recycled fluid, thus
inefficient. For this reason, capacity change with recycling is typically reserved as
a last resort for achieving capacity turndown beyond the minimum flow limit of
the machine, e.g., during startup and crippled operation. Nevertheless, recycling is
Fig. 3.5 Three flows in pumps and compressors

Fig. 3.6 Process flow, machine flow, and recycle flow
simple and reliable, and thus often used in many simple applications. For instance,
recycling is the preferred control strategy for reciprocating compressors due to its
simplicity.
For air compressors, bleeding the air into the atmosphere is typically used, similar
to recycling in terms of efficiency.
Inlet Guide Vane
Moving the operating point by adjusting the inlet guide vane (IGV) works differently
than changing speed, resistance, or recycling. From a process control standpoint, a
compressor with IGV can be viewed as having a variable geometry. Changing the
entry angle of the guide vane effectively changes the performance curve, which indi-
rectly causes the operating point to move. For instance, assume that the compressor
operates at 100% speed (of fixed speed). Reducing the guide vane angle causes the
operating point to move along line ➀ while reducing the speed makes the operating
point move along line ➁, as shown in Fig. 3.7.
3.2.2 Operation with Multiple Control Handles
Practical applications often require both the flow and head to be maintained on target.
Since flow and head are separate control targets, they require two independent control
handles, usually the compressor speed and system resistance.
Figure 3.8 illustrates how the machine speed and throttling valve work together to
maintain the head and flow on target. Suppose the head is to be kept constant while
the machine speed is reduced. The operating point would move in the direction of
A → C, which would cause the head to decrease. The discharge valve will then be
throttled to increase the resistance. The increased resistance will cause the head to
return to the target value (Point B). Speed reduction and valve throttling together
will eventually move the operating point to D. The actual trajectory from A to D
Fig. 3.7 Capacity control with inlet guide vane
Fig. 3.8 Capacity control with centrifugal machines
may differ depending on how quickly the speed is reduced and how fast the valve is
throttled.
Example 3.1 Operation of a hairdryer. Consider a two-speed hair dryer in Fig. 3.9a.
A hairdryer is an air blower and can be treated as a low-pressure air compressor.
Fig. 3.9 A two-speed hair dryer
The two-speed option provides discrete speed changes for capacity control. The
concentrator increases the system resistance to increase the exit speed of the warm air.
With the two-speed switch and the detachable concentrator, the hairdryer has four
possible operating points, as shown in Table 3.2 and Fig. 3.9b.
When the hair dryer is at the lower speed and without the concentrator, the operat-
ing point is at A. Switching to the higher speed moves the operating point to B along
the system resistance line. Attaching the concentrator increases the outlet resistance
and forces the operating point to move from B to D along the constant-speed line.
See Table 3.3 for a summary.
If the dryer is switched to the lower speed and the concentrator is removed simul-
taneously, the operating point would move from D to A. Whether this move is
through B or C depends on which action is taken earlier and which response is
faster: the speed switch or concentrator removal.
If both the speed and the resistance could change continuously, then the trajectory
from D to A would take any path between D → C → A and D → B → A, similar
to that in Fig. 3.8.
Table 3.2 Operating points

of a hairdryer Concentrator OFF Concentrator ON
Low Speed A C
High Speed B D
Table 3.3 Cause-and-effect table for a hairdryer

Control handle Method Operating point Flow Head Power
Machine speed Switch to higher speed A⇒B More Higher More
System resistance Add concentrator B⇒D Less Higher Less
Machine speed Switch to lower speed D⇒C Less Lower Less
System resistance Remove concentrator C⇒A More Lower More
3.2.3 † Capacity Turndown
Capacity is a different name for flow throughput. Two types of capacities concern us:
the process capacity, i.e., process flow, and the machine capacity, i.e., the machine
flow. The process flow is the control target at the process level, and the machine
flow is the control target at the machine level. These two flows are not always the
same since the machine flow has a turndown limit while the process flow does not.
In case of conflict, the target at the process level takes precedence over the target
at the machine level (except for protective actions), and additional handles must be
used to meet the process level requirements.
For centrifugal machines, the machine flow can be influenced by any of the con-
trol handles shown in Fig. 3.3. For instance, process flow turndown can start with
reducing the process and machine flows in tandem. After the machine flow reaches
its minimum flow limit, recycle flow is introduced to continue reducing the process
flow while maintaining the machine flow at the minimum flow limit.
This two-step turndown is illustrated in Fig. 3.10. From Point A to B is via valve
throttling, and from B to B is via recycling. Similarly, from A to C is by reducing
the speed, and from C to C is via recycling. With both machine speed and throttling
valve changing simultaneously, the operating point can be moved from A to D , after
which recycling is introduced to move the operating point from D to D. The amount
of recycling (the difference between F2 and F3) in these scenarios is significantly
different.
It is also possible to move the operating point from A to D by recycling only.
However, this is inefficient due to the unnecessarily large recycle flow.
'
'
'
Fig. 3.10 Flow turndown below surge limit

3.3 Protecting the Operating Envelope 77
Table 3.4 Flow turndown in a centrifugal machine

Process Flow Machine Flow Recycle Flow
Point A F1 F1 0
Point B , C , D F2 F2 0
Point B, C, D F3 F2 F2-F3
In other words, the capacity turndown can be achieved by any of the three types
of control handles (speed, resistance, and recycling) before reaching the minimum
flow limit. After that, recycling is the only control handle for further turndown. The
flow values at various points are illustrated in Table 3.4.
This minimum flow limit, or surge point, can be visualized on the performance map
and easily followed by a human. However, for automated flow control, the minimum
flow limit must be represented in a machine-understandable format suitable for online
implementation, such as a mathematical formula or look-up table. This quantitation
of the performance curve is the main topic of control design in Chap. 5.
3.3 Protecting the Operating Envelope
Disturbances and abnormal conditions are inevitable in actual operation. Respond-

ing to abnormal situations is more critical and challenging than normal operations
since operational safety is at stake. Keeping the operating point inside the operating
envelope is compulsory.
3.3.1 Operation Under Abnormal Operating Conditions
Typical process conditions that may drive the operating point to an inefficient or
unsafe region include process upsets, equipment failures, instrument malfunctions,
and turndown operation. As shown in Fig. 3.11, the operating envelope (shaded
area) is defined by various limits and constraints imposed by both the machine and
the process, such as minimum and maximum speed, minimum flow (surge) and
maximum flow (choke), motor power, pressure and temperature, and gearbox torque.
The minimum flow limit is crucial for centrifugal machines, and over-pressure for
reciprocating machines.
Operational safety requires that the machine operation not violate the operat-
ing limits. At the same time, operational efficiency demands that the machine be
allowed to operate as close as possible to the limits for maximum operating range.
For instance, minimizing recycle flow through better control significantly contributes
to energy efficiency because recycling is highly inefficient. However, excessive recy-
cling, common in many operating plants, is a huge energy waste.
Fig. 3.11 Compressor operating point and envelope
The process operating constraints vary with the actual process flow configuration.
For instance, process piping and equipment all have pressure and temperature ratings
that must not be exceeded, especially in reciprocating machines where the back pres-
sure determines the discharge pressure and is limited only by the power supply from
the driver. Low pressure at the suction side of a reciprocating pump may violate the
net suction pressure head required (NSPHR) and should be monitored and avoided.
On the machine side, pumps and compressors have inherent design limits. For
instance, a centrifugal machine is designed to operate within the rotation speed range.
Below the minimum speed, the machine may become unstable, while operating at
too high a speed may also put the mechanical integrity at risk. The typical operating
range is between 70% and 105% of the design speed. Enforcing the speed limits is
straightforward by adequately monitoring and limiting the rotating speed.
The surging phenomenon is a common threat to all centrifugal and axial machines,
which can potentially cause severe mechanical damage.1 Surge is caused by insuf-
ficient flow through the machine, commonly occurring during initial startup, severe
turndown, or significant load disturbances.
1On the other hand, some machines, usually smaller ones, can operate for long periods with inter-
mittent or continuous light surges without mechanical harm, although with significant impairment
of aerodynamic performance.
Fig. 3.12 Compressor stonewall
When the machine flow approaches the minimum stable flow limit,2 the recycle
valve should be opened to send a portion of the fluid at the discharge side back to
the suction via the recycle line to maintain the machine flow above the limit. In the
actual operation, the main challenge is getting real-time information on the minimum
allowable flow limit, which varies with speed and inlet condition.
Too much flow through a centrifugal pump or compressor is also a concern. As
the discharge flow approaches sonic speed, the flow can no longer increase. The
machine is said to have reached the “choked” condition, or stonewall, as shown in
Fig. 3.12. The choke phenomenon is not as threatening as a surge but can still cause
serious operational problems such as vibration and extremely low efficiency.
If a centrifugal machine is prone to approaching the choke line often, it may be
necessary to limit the machine flow. A restrictive orifice or throttling valve can be
installed on the discharge line to limit the flow. The challenge for choke prevention
is knowing where the choke line is in relation to the current operating point, the same
as surge prevention.
The driver for a pump or compressor is typically an electric motor or a gas turbine.
The motor can have a fixed-speed drive (FSD) or variable frequency drive (VFD).
2 Centrifugal pump operation has three minimum flow limits: minimum continuous thermal flow
(MCTF), intermittent minimum flow, and minimum continuous stable flow (MCSF). By minimum
flow, we typically refer to the minimum continuous stable flow (MSCF). MCSF ranges from roughly
10% to 80% of the best efficiency point flow depending on pump size and type, operating speed,
impeller suction geometry, liquid density, and other factors. The operation of centrifugal pumps
below their minimum flow requirements is the primary cause of premature pump failure. Hydraulic
instability occurs at low flows and can cause a surge, cavitation, and excessive vibration in the
pump. The extent and type of damage to the pump depend on how long the low-flow persists and
the magnitude of the generated forces and vibrations.
Fig. 3.13 Flow through an orifice plate
A fixed-speed motor can also be equipped with a variable-speed gearbox (VSGB)

to achieve variable-speed operation. These machines and devices have operating
constraints, such as electric current, power, and torque, that should not be violated.
3.3.2 ‡ Flow in System Resistance Components
A pump or compressor is connected to a suction and a discharge system; both can

be viewed as flow-resisting virtual orifices or choke valves. The machine “pulls”
fluid from the suction system and then “pushes” it through the discharge system to
reach downstream operation. From the process point of view, the pressure versus flow
behavior can be interpreted as a flow through a series of flow resistance components
(Fig. 2.25a).
A flow through a contraction in the piping results in a pressure drop, which
exhibits a quadratic function of the flow rate (Cunningham 1951; ISO-5167 2003).
For example, the pressure drop P across a standard orifice (Fig. 3.13) is proportional
to the squared flow rate Fv2 :

Fv = C P/ρ, (3.2)
Fv2
P = ρ . (3.3)
C2
C is a flow coefficient that remains a constant once the orifice plate is constructed.3
The pressure drop across a valve exhibits similar behavior, except that the flow
coefficient C varies with the valve opening.
3A carefully designed and constructed orifice plate is an inexpensive and reliable flow measurement
device. See Sect. 6.2.4.
In general, this quadratic relationship exists for all the components in the flow
line, including filters, strainers, coolers, and control valves. Each component has the
same mass flow but a different flow coefficient C.
The flow through an orifice becomes unstable when the flow rate falls below
a specific limit. A typical range is 10:1 in terms of differential pressure across the
orifice. For this reason, an orifice-based flowmeter√loses its accuracy at approximately
10% of the maximum DP value or one-third ( 10 : 1 = 3.16 : 1 ≈ 3 : 1) of the
maximum flow rate, as shown in Fig. 3.13b.
The surge and choke phenomena share many similarities with flow through a valve
or orifice. A compressor in stonewall operation is like an orifice with a critical flow
(Boyce et al. 1983). The choked flow through an orifice defines the maximum flow,
with the turndown limit at approximately 1/3 of the maximum flow. Coincidentally,
a centrifugal machine suffers from the surge problem at approximately 1/3 of the
maximum flow rate, defined by the choke flow, similar to an orifice plate.
The pressure difference Pd − Ps between the suction and discharge is defined by
the pressure rise across the impeller, and the pressure drop across the diffuser. The
former is a function of machine speed and flow rate, while the latter is determined by
the flow rate only. The fluid inside the machine reaches the maximum speed at the
tip of the impeller with the highest energy at the entrance of the diffuser (Fig. 2.7). A
pressure drop occurs when the gas goes through the diffuser. The pressure difference
and the flow rate through the compressor have a relationship similar to that of an
orifice (see Eq. 3.3):

Fv ∝ C (Pd − Ps )/ρ (3.4)
Pd − Ps ∝ Fm2 ∝ Fv2 . (3.5)
This analogy is made more prominent by flipping the X and Y axes of the flow
versus DP relationship in Fig. 3.13b. Assume Pmax is the pressure at the blade tip.
The relationship between the pressure drop (Pmax − P) and the flow in Fig. 3.14a
is similar to the head-flow relationship in the compressor performance curve in
Fig. 3.14 DP versus flow in an orifice plate and head versus flow in a compressor
Fig. 3.14b. This similarity helps explain and understand the performance and resis-
tance curves discussed in Sect. 2.2 and the surge indicators in Sect. 5.3.
This relationship agrees with Eq. 2.57. In other words, the pressure rise in the
compressor is roughly a function of the square of the machine speed, which in turn
is proportional to the machine flow, per fan laws (Boyce et al. 1983). As a result,
this pressure difference across the compressor is roughly proportional to the pressure
drop across an orifice on the same flow path:
Pd − Ps ∝ P. (3.6)
This conclusion is extremely qualitative and has included many approximations.

Nevertheless, this simplified relationship had been the basis for many early anti-surge
control solutions (Boyce et al. 1983; Liptak 2006; Staroselsky and Ladin 1979; White
1972).
3.4 Transitioning Between Operating Modes
Process operation can have many different operating modes. Transitioning from
one mode to another poses unique challenges, sometimes with severe risks. A
safe and smooth transition is compulsory, requiring a good understanding of the
startup/shutdown sequence and the transitional behavior of the machine. For exam-
ple, the startup and shutdown of rotating equipment are routine operations that are
complicated and potentially risky. Damages to pumps and compressors most often
occur during startup or emergency shutdown.
3.4.1 Startup of Centrifugal Machines
The startup can be cold or hot. For example, for the two-train compressor system
in Fig. 1.9, if both trains are started from a shutdown condition, it is a cold startup,
which is characterized by low pressures at both the suction and discharge headers.
On the other hand, assume train 1 is in normal operation, and train 2 is in shutdown
mode. When train 2 starts to join the first train, it is a hot startup. During a hot
startup, pressure headers at suction and discharge are also established by the running
machines.
The electric current required during startup can be several times higher than keep-
ing the machine running at the normal operating condition. The heat generated during
startup can be significantly higher than normal since the heat generated is propor-
tional to the current squared (I 2 R).
The startup of a centrifugal pump or compressor is a complex sequence. The
machine starts with the downstream block valve closed and the recycle valve fully
opened. Once the machine is started, the fluid quickly fills the recycle line; the
3.4 Transitioning Between Operating Modes 83
operation enters total-recycle mode. Depending on the recycle piping and valve size,
the operating point will typically be on the far right side of the performance curve.
As more process flow enters the machine, the discharge pressure (thus the head)
builds up.
For a hot startup, such as a machine joining an established header, the block valve
would be opened as soon as the discharge pressure is higher than the downstream
line pressure, and thus the transition is bumpless. However, the downstream may be a
long empty pipeline with virtually zero pressure during a cold startup. The discharge
pressure remains very low until the pipeline is filled with liquid. If the block valve (an
on/off valve!) is suddenly opened with virtually zero resistance to flow, the operating
point would abruptly move beyond the end-of-curve and trip the machine.
Example 3.2 Cold start of an injection pump. Deep water disposal (DWD) is an
integral part of a surface facility to inject the produced water (after treatment) back
into the underground reservoir. As shown in Fig. 3.1, the pump discharge connects
via a long pipeline to the deep disposal wells.
One particular challenge for a cold startup is the need for initial system resis-
tance. As the motor starts, the operating point rapidly moves to the far right on the
performance curve due to the low system resistance (low head) from downstream.
The high flow rate quickly overwhelms the motor and trips the motor by high power
or vibration.
The solution is to artificially create the required system resistance during a cold
startup. Two popular designs are as follows:
1. Discharge throttling valve. If available, the discharge control valve for capacity
control can create the initial system resistance during startup with the valve
initially closed. Once the pump reaches full recycle mode, the valve can be
manually and slowly opened to build up the back pressure in the downstream
pipeline. See the piping configuration in Fig. 3.15a.
2. A discharge control valve is typically not installed for a variable-speed pumping
process. Even if available, the control valve may not be rated for the type of
high-pressure drop across the valve in high-pressure pumping applications. For
this scenario, a working example is to utilize two block valves in parallel, one
large and one small. See Fig. 3.15b. At the initial phase of the startup, the large
valve is closed, with only the small valve open. The small valve provides enough
Fig. 3.15 Pump startup with a throttling valve

Fig. 3.16 Trajectory of operating point during cold startup
resistance to keep the pump running to prime the pipeline. Once the back pressure
in the pipeline reaches an acceptable level, the large valve opens. The startup
sequence is responsible for the opening and closing of the valves.
Example 3.3 Startup procedure of a centrifugal machine. The startup of a large

compressor is very challenging. Figure 3.16 shows the trajectory of the operating
point during a cold start of a produced gas compressor (PGC). The suction and
discharge header pressures are both under pressure control, and thus, the pressure
ratio is nearly constant.
1. Initial ramp-up of machine speed: ➀ → ➁. The machine starts with the recy-
cle valve fully open. It first rapidly ramps the machine speed to above the mini-
mum stable speed with the compressor in total-recycle mode. The operating point
quickly moves to the far right to the valve capacity limit. The flow increases as
the pressure inside the circuit builds up (variable resistance curve). The discharge
pressure increases as more gases are brought in but are still below the discharge
header pressure, and thus the non-return valve is still closed, with no process flow
(i.e., forward flow) going downstream. The compressor flow equals the recycle
flow.
2. Establishment of forward flow: ➁ → ➂. After the machine exceeds the minimum
stable speed, continue ramping up the machine speed but at a much slower pace
for stability reasons. At the same time, the recycle valve is slowly closed to build
up the discharge pressure.
Once the discharge pressure is above the discharge header pressure, the non-return
valve opens, and the gas starts to move forward. As the recycle valve slowly closes,
the operating point moves to and stabilizes at point ➂.
3.4 Transitioning Between Operating Modes 85
3. Normal operation: ➂ → ➃. The operating points move between point ➂ and ➃

during normal operation, responding to the changes in capacity demand. The
trajectory is approximately flat with a constant-pressure ratio since the suction
and discharge pressures are both under control. The compressor speed is adjusted
to maintain the desired capacity (compared to the trajectory in Fig. 3.8).
4. Shutdown: ➃ → ➄. A controlled shutdown starts by reducing the compressor
speed and opening recycle, which brings the operating point to the origin ➀.
The trajectory of the operating point for a fixed-speed compressor is different but can
be similarly analyzed based on what has been discussed in this section. Chapter 5
will provide more detailed discussions on the startup sequences.
3.4.2 Shutdown of Centrifugal Machines
Machine shutdown is the process of bringing the head and flow to zero. Shutdowns
can either be controlled (scheduled) or uncontrolled (emergency).
A scheduled shutdown follows the standard operating procedure (SOP). The
machine speed is slowly reduced to the minimum stable speed before cutting off
the power supply. This procedure ensures that the operating point reaches the origin
following a steady and controlled path, such as ➀ in Fig. 3.17.
An emergency shutdown is a disruptive process and poses many challenges to
equipment safety, with surge being one of the main risks. During an emergency
shutdown, the fuel supply to the gas turbine or electricity to the motor is cut off
instantly, causing the compressor to coast down on inertia. Since the drop in the head
(or compression ratio) is proportional to the square of the machine speed, and the
Fig. 3.17 Trajectory of operating point during shutdown

downstream and upstream piping imposes the pressure difference across the machine,
the compressor may enter a surge condition if the pressure cannot be relieved quickly
by the recycle valve.
The decrease in the head may be faster or slower than the reduction in the flow,
depending on the machine configuration and process flow condition. Thus, the trajec-
tory from the current operating point to the origin may exhibit significant variations,
as shown in Fig. 3.17. It is nearly impossible to know the exact trajectory since the
shutdown happens extremely fast4 and is influenced by many factors unless a high-
fidelity rigorous dynamic simulation is conducted. The primary concern is whether
the operating point would move too far to the left of the surge line and stay there
for too long during the shutdown for the machine to enter a deep surge, as shown by
line ➁ in Fig. 3.17. A typical and acceptable trajectory is similar to line ➂.
The most critical consideration is that the accumulated gas in the surge volume
must be minimized in the process design to increase the speed of response. For
instance, the recycle line’s take-off should be before the discharge cooler. Otherwise,
hot or cold bypass lines may need to be added.
3.4.3 † Crippled Operation
A complex process configuration often comprises multiple pumps or compressors.

It is common to run into a situation where a part of the operation goes down, and
the remaining unit is expected to stay in operation, leading to the so-called crippled
mode operation (partial shutdown). Like an emergency shutdown, the safe and smooth
transition to crippled mode can be very complicated and disruptive.
Consider the two parallel compressors in Fig. 3.18. Assume that the second com-
pressor is tripped to a shutdown. What would happen to the first compressor, and
how is it expected to respond to this abrupt change without tripping itself?
When the second train is tripped to shutdown, the process flow shared by the
two compressor trains is now forced through the first train alone, causing the suction
header pressure to increase. The higher suction pressure will demand a quick increase
in the machine speed. With proper responses by the operator or control scheme, the
first train may react fast enough to survive the disruption. Eventually, the operating
point will settle at a higher flow rate to accommodate the entire flow. Otherwise, the
first train may trip as well.
4Extremely high-speed data recorder is in increasing use to record data during a shutdown for
post-incident analysis.
3.5 Summary 87
Fig. 3.18 Process flow of a 2 × 2 compressor network
3.5 Summary
The operation of pumps and compressors is challenging. The goal is to achieve

safe and efficient operation, and the operation is based on a sound understanding
of the critical process variables such as speed, flow, and head, the familiarity of
the dynamic cause-and-effect relationships between the key variables, which are
discussed in Chap. 2. These are also the basis for automatic control to be discussed
in later chapters since process control is to automate manual operations. The tasks
of pump and compressor operation typically include:
• maintaining the operating point in terms of flow and head,
• protecting the operating envelope by observing both process constraints and
machine limits, and
• ensuring a safe and smooth transition between the different operating modes.
The operation of complex systems involving multiple machines is even more
challenging due to the infinite possibilities of process flow configuration. It is impos-
sible to be familiar with all the variations. Instead, applying the basic concepts and
principles to analyze and guide the operation is more effective and efficient.
References
12th turbomachinery symposium. College Station, Texas, USA
Cunningham RG (1951) Orifice meters with supercritical compressible flow. Trans ASME
73:625–638
ISO-5167 (2003) Measurement of fluid flow by means of pressure differential devices inserted in
circular cross-section conduits running full, 2nd edn. ISO
LeBleu Jr J, Perez R (2014) Operator’s guide to rotating equipment—-An introduction to rotating
equipment construction, operating principles, troubleshooting, and best practices. AuthorHouse
Liptak BG (2006) Process control and optimization, instrument engineers’ handbook, vol 2, 4th
Niu S, Xiao D (2022) Process control—engineering analyses and best practices. Advances in indus-
trial control. Springer
Staroselsky N, Ladin L (1979) Improved surge control for centrifugal compressors. Chemical Engi-
neering, pp 175–184
White MH (1972) Surge control for centrifugal compressors. Chemical Engineering, pp 54–62
Chapter 4
Overall Control Strategy
Rotating equipment are complex machines with extremely fast responses. Achieving
safe and efficient operation through manual control (Chap. 3) is nearly impossible.
Unscheduled shutdowns of these machines can result in significant interruptions to
production, reduced capacity, and equipment damage. Therefore, automated control
solutions are necessary to ensure reliable operation and protect the equipment. Since
process requirements vary across applications, there is no universally applicable
control design. A sound control solution is based on a holistic view of the process
operations and follows a process-centric approach that includes machine control
as an integral part. A structured approach and proven methodology are advised to
analyze the process flow, extract the cause-and-effect relationship, and design the
control solution.
4.1 Overall Control Strategy
Rotating equipment is instrumental to the safe and efficient operation of any indus-
trial processes, similar to static equipment such as chemical reactors or distillation
columns (Bahadori 2016; Borremans 2019; Giampaob 2010; Lieberman and Lieber-
man 2014; McMillan 1983). They deserve the same attention and treatment in the
overall process control design as received by static equipment.
4.1.1 General Control Philosophy
Equipment control has traditionally been the responsibility of equipment manufactur-

ers or their affiliated/certified partners. The focus is primarily on the safe operation
of the equipment and much less on the efficient operation of the entire process.

https://doi.org/10.1007/978-3-031-43122-7_4
90 4 Overall Control Strategy
The equipment control solution is usually designed and implemented with propri-
etary technology, hardware, and software, separately from the overall process control
solution. Consequently, equipment control is often treated as a black-box component
of the process control solution in both design and operation. This equipment-centric
design is usually poorly integrated with process control for overall optimality. Due
to its black-box implementation, it is also a maintenance nightmare (Botros et al.
2016).
On the contrary, a control strategy based on a holistic view of the process flow con-
figurations has a much higher chance of reaching overall optimality than piecemeal
control designs (Kurz and Brun 2017). From the process control perspective, a pump
or compressor is no more complex than a distillation column or chemical reactor.
Both can be analyzed by following the same standard analytic procedures, designed
with the same control methodologies, and controlled with the same standard process
control technologies we already possess (Niu and Xiao 2022).
Equipment control is an integral part of process control. The process and equip-
ment must be treated as a single integrated system when designing the optimal control
solutions. A flawed process control strategy may fail to maintain the proper mate-
rial balance and causes oscillations in the unit operation, making equipment control
more challenging. When this happens, even a perfectly working equipment control
solution may fail to deliver the desired performance.
Another perception is that pumps and compressors are high-speed rotating equip-
ment and require high-speed hardware and software to support the control solution.
As computer technology advances, the same speed that could only be offered by
proprietary hardware and software is now achievable with standard control systems
such as most DCS and PLC systems (Elliott and Bloch 2021; Honeywell 2023; Ver-
million et al. 2023). The standard control system infrastructure has the advantage of
established reliability, rich functionality, and complete transparency.
4.1.2 Overall Control Objectives
The operating objectives dictate the control objectives. Generally speaking, the over-
all control objective is to achieve safe and efficient operation, which can be summa-
rized as controlling one point, protecting two envelopes, as well as ensuring smooth
transitions between different operating modes:
1. Control the operating point. The pressure and flow define the operating point,
dictated by the machine’s performance curve and the process’s resistance curve.
Continuous regulatory control keeps the operating point on the target.
2. Protect the operating envelope. Design limits and operating constraints imposed
by both the process and machine constitute the operating envelope. The operating
point should be allowed to move close to the operating envelope for maximum
operating range. At the same time, the protective control functions prevent the
4.1 Overall Control Strategy 91
operating point from moving outside the operating envelope for safety and effi-
ciency.
3. Safeguard the safety envelope. Even the protective control functions may fail
during a major process upset or device failure, and the operating point may move
out of the safety envelope. The instrumented safeguarding logic will initiate a
proactive shutdown to avoid severe damage. Typical trip events include vibration,
surge, and high pressure.
4. Ensure smooth mode transitions. There are various operating modes in practical
applications, such as shutdown mode, normal operating mode, turndown mode,
and crippled mode. It is crucial to maintain safety and efficiency when the oper-
ation transits from one mode to another.
4.1.3 Layered and Integrated Design
A sound control design follows a layered yet integrated approach, with normal reg-
ulatory control, protective overriding control, and instrumented safeguarding as the
three lines of defense against abnormal situations. See Fig. 4.1 and compare with 1.20.
Fig. 4.1 Layers of process safety protections for pumps and compressors
1. Normal regulatory control, such as capacity control, continuously executes to

maintain the operating point on the target. It serves as the first line of defense
against abnormal conditions and reduces the probability of machine surges.
2. Protective overriding control like anti-surge control keeps the operating point
inside the operating envelope. Protective control remains dormant during normal
operating conditions and becomes active only when the operating point moves
outside the operating envelope. It serves as the second line of defense against
abnormal conditions. Anti-surge control is the most crucial protective control for
a centrifugal machine.
3. Safeguarding logic, such as anti-surge trip logic, monitors the operating point
against the safety envelope and proactively shuts the operation down if the safety
limits are violated. Anti-surge trip logic provides the third line of defense against
potentially damaging events like surge.
It is crucial to distinguish between normal regulatory control and protective over-
riding control in control design. Normal regulatory control is responsible for the
continuous and smooth operation of the entire process to maintain the desired mate-
rial and energy balance. Protective overriding control focuses more on localized
protection of a specific process area or equipment against abnormal operating con-
ditions. Complex control solutions can also become much easier to apprehend or
design by separating regulatory and protective control.
There are many significant advantages to this layered design. For example, the
control solution is unaffected by a single mode failure. The multiple lines of defense
against abnormal conditions provide more refined control action. Each layer can
be tuned separately with different levels of aggressiveness based on the urgency of
action.
4.1.4 Migration from Proprietary to Open-Platform Solutions
Many vendor-provided solutions are based on proprietary hardware and software

that is too difficult or costly to support the layered design in Fig. 4.1. Many reported
failures with third-party solutions are with the hardware/software infrastructure (e.g.,
with the computer motherboard) rather than the control solutions itself. Proprietary
hardware and software are becoming a liability that can no longer compete with the
full-featured and field-proven DCS and SIS control systems that have already existed
and have been running the entire plant.
The use of proprietary hardware and software for control solutions is becoming
increasingly difficult to justify. Specifically, the equipment control, particularly anti-
surge control, is shifting toward open solutions that rely on standard hardware and
software (CCC 2021; Elliott and Bloch 2021; Honeywell 2023; Vermillion et al.
2023; Zelenov 2013). The control solution can now be designed with existing process
control technologies we have all been familiar with, following the same standard
methodology and procedure as we do with other process control solutions. Moreover,
4.2 Regulatory Control: Capacity Control 93
the control solution is implemented on standard hardware and software in the existing
control systems we have been using. They can be operated and maintained like other
process control solutions in the control system.
This paradigm shift has a long-term impact on control and operation. The new
paradigm enables the control strategy to base on a holistic view of the process and
equipment with equipment control as an integral part of the overall control strategy,
focusing on the overall control optimality. At the same time, this new paradigm shift
demands increased knowledge of the machine characteristics and control strategy by
process control engineers and operation personnel.
4.2 Regulatory Control: Capacity Control
Normal regulatory control typically includes quality control and capacity control.
Quality control is an operating requirement at the process level and is usually not a
main consideration for pump and compressor operation. Capacity control is thus the
primary requirement for normal regulatory control (Giampaob 2010).
In practical operations, fluctuations and disturbances are inevitable. The operating
point must be regulated by continuously adjusting the process flow rate. Capacity
control is thus a regulatory control action to maintain the material balance between
supply and demand. The improved stability of the plant operation through capac-
ity control also benefits the protective control loops (e.g., anti-surge control) and
safeguarding functions (e.g., anti-surge trip).
We will explain the general capacity control strategy by examining the five essen-
tial components of a feedback control loop one by one within the context of capacity
control.
4.2.1 Capacity Control Objectives
The pumps and compressors provide the energy for the fluid to flow forward by
overcoming the resistance created by the piping, fitting, and other static equipment.
Automatic control ensures the process gas flows to the right place at the desired
pressure and flow rate. The primary objective of capacity control is thus to keep the
operating point at the desired pressure and flow, dictated by the material balance at
the process level (see Fig. 4.2).
A secondary control objective is optimizing the operation to achieve the best
efficiency, especially when multiple pumps or compressors are involved. All pumps
and compressors should have their best efficiency points (BEP) aligned with the
normal operating condition for best efficiency, which translates to maximum energy
savings.
Fig. 4.2 Compressor operating point
4.2.2 Process Dynamics and Cause-and-Effect Relationships
Process dynamics, including the supply-and-demand model, cause-and-effect rela-

tionship, and transient response behavior, are the basis of design for capacity control.
The supply-and-demand model dictates whether the operation is supply driven
(“supply pushing”) or demand driven (“demand pulling”). In a supply-driven oper-
ation, the upstream operation dictates the flow rate. The fluctuations in the supply-
and-demand balance propagate forward along the process flow path. In contrast,
the process flow throughput is dictated by downstream demand in a demand-driven
operation, and the fluctuations in demand propagate backward to upstream.
Supply driven : fluid supplied by upstream = fluid pushed to downstream

Demand driven : fluid demanded by downstream = fluid pulled from upstream.
The supply-driven and demand-driven operations often require drastically dif-

ferent capacity control designs. A systematic supply-and-demand analysis is thus
essential in complex process design before starting control design or troubleshoot-
ing.
The cause-and-effect analysis starts with defining the key process variables. The
controlled variables in pump and compressor operation are typically the head and
flow and can be influenced by machine speed, system resistance, and recycling. Speed
Table 4.1 Cause-and-effect table for pump and compressor operation

Control Handles Head Flow Head Flow
Machine Speed ↑
Recycle ↑
is an independent variable that directly impacts the head and flow. System resistance
and recycling affect the head and flow more complexly.
The cause-and-effect relationships are summarized in the cause-and-effect table
(CET) (see Table 4.1). The CET helps determine the targets to be controlled (con-
trolled variables), the control handles to be manipulated (manipulated variables),
and the control strategy to follow for meeting the control targets with the available
control handles.
The transient behavior between the variables is another critical factor to consider
in the control design since some variables can change extremely rapidly, requiring a
similar response speed from the controller.
4.2.3 Process Measurements and Controlled Variables
The control target for capacity control is the desired flow rate amid supply-and-
demand fluctuations. The imbalance between supply and demand is caused by flow
changes and is reflected by the variations in gas pressure or liquid level. The pressure
or level will increase if the incoming flow exceeds the outgoing flow. Conversely, the
pressure or level will decrease if the outflow exceeds the inflow. Instead of directly
and explicitly controlling the flow rate, regulating the pressure or level is more reliable
and convenient for capacity control. Thus the pressure and level are the controlled
variables and essential process measurements.
Multiple relay points are needed to propagate the supply-and-demand balance
forward along the process flow path. At selected control points (or relay points), the
pressure and level are indicators of the inventory level in the system. Any imbalance
in supply and demand is reflected in the changes in pressure (for gas) or level (for
liquid). At each relay point, the pressure is maintained by adjusting the incoming or
outgoing flow rate.
The typical pressure measurements around a compressor are the suction header
pressure P1 and discharge header pressure P2 . See Fig. 4.3. If the gas supply increases,
the pressure P1 increases. A pressure controller speeds up the machine to maintain
the target pressure. Higher speed causes the discharge pressure to increase, which
Fig. 4.3 Control target and handles for centrifugal compressor control
in turn triggers the discharge pressure controller to force more flow downstream to
bring the discharge pressure back on target.
The pressures at the key relay points are also the required process measurements.
4.2.4 Final Control Elements and Manipulated Variables
The final control elements (FCE) include the three major types of control handles
for compressor capacity control: the rotating speed, system resistance, and recycle
flow, as shown in the cause-and-effect table in Table 4.1. A list of the most common
control handles are as follows:
1. On/off switch. Switching the machine on and off to control the capacity is sur-
prisingly common, especially with reciprocating machines. On/off control is a
powerful and economical solution for process operations if the machine has high
starting reliability and can accept slack control performance. Many small-capacity
machines often operate on an on/off basis. For instance, a jockey pump is designed
to start and stop frequently; the home A/C unit is based on on/off control; and an
operating plant’s nitrogen gas or instrument air compressor is switched on and off
to maintain the gas supply pressure within the desired range in a storage vessel.
2. Rotation speed. A pump or compressor can operate at different speeds by design.
Changing the machine speed is an energy-efficient way to achieve the desired
pressure or flow. The variable-speed capability is usually provided by a variable-
frequency drive (VFD), variable-speed gearbox (VSGB), or gas expander.
3. Inlet guide vane (IGV). Centrifugal compressors can be equipped with adjustable
inlet guide vanes (IGV) on the inlet. The inlet guide vanes pre-rotate the gas
stream relative to the impeller rotation and effectively change the performance
curve. It is another efficient way of changing the compressor capacity.
Table 4.2 Typical control handles for pump and compressor operation
Control Handles Pump Compressor Pump Compressor
Machine Speed x x➀ x x
Inlet guide Vane x➁
Throttling Valve x x ➂, ➃ x x
Stroke length x x
Suction Unloading x x
Recycle Valve x x➄ x x
4. Throttling valves. The most common approach for capacity control for centrifugal
machines is changing the system resistance via control valves at either the suction
or discharge. Throttling the flow increases the pressure ratio that the compressor
“sees,” indicating the system resistance that the machine must overcome.
5. Recycle valve. Recycling a portion of the high-pressure fluid from the discharge
back to the suction side is a simple but inefficient way of adjusting the flow
through the machine (Kurz et al. 2016).
6. Blow-down. Discharging a portion of compressed air into the atmosphere is preva-
lent for air compressor control.
For a centrifugal compressor, the location of the potential control handles is illus-
trated in Fig. 4.3. Their applicability is summarized in Table 4.2, with the circled
number corresponding to the control handles in Fig. 4.3.
Some control handles are mutually exclusive and will not be installed/used on the
same machine. For instance, a fixed-speed centrifugal machine typically depends
on the suction or discharge throttling valve to adjust the capacity, but not both. The
throttling valve is typically unnecessary if the machine has a variable-speed driver.
If installed, the throttling valve is always at the discharge side for a centrifugal
pump due to the requirement of net suction pressure head (NSPHR). However, for a
centrifugal compressor, the throttling valve is preferably at the suction but can also
be on the discharge.1
Varying the machine speed for capacity control is more efficient than valve throt-
tling, which in turn is more efficient than recycling, as illustrated in Fig. 3.3.
1 Surge calculation is typically based on suction flow. If discharge flow is provided, it must be
converted to suction flow in the calculation. This conversion requires the polytropic exponent or
efficiency and thus makes an invariant coordinate less “invariant.” See Chap. 5.
Table 4.3 Typical control strategies for pump and compressor operation
Pump Compressor Pump Compressor
Fixed-Speed Valve Throttling + Recycle Recycle
Variable-Speed Speed + Recycle Speed
4.2.5 Capacity Control Algorithms
Capacity control is typically based on standard PID controllers. Depending on the

process flow configuration, the control scheme can be as simple as a single standalone
PID controller; it can also be very elaborated, comprising multiple PID controllers,
sequences, and logic. Even with the same controlled and manipulated variables, the
control strategy can vary significantly depending on the supply-and-demand model
and the cause-and-effect relationship.
Supply-Driven and Demand-Driven Control
Process engineering design ensures the steady-state supply-and-demand balance for

material and energy.2 However, supply and demand are never precisely balanced in
a dynamic operating environment due to fluctuations, or swings, in the supply or
demand:
Swing = Supply − Demand. (4.1)
Process design typically includes built-in buffering capacity or surge volume, such as
“line-packing”3 for gas and surge tanks for liquid. However, the inherent surge vol-
ume is limited. Swing streams or swing capacity often need to be explicitly provided
in the design to “absorb” the sustained imbalance in supply and demand.
The imbalance between supply and demand is reflected in pressure (for gas oper-
ation) or level changes (for liquid operation) at selected control points. For this
reason, capacity control is typically achieved indirectly via pressure or level control
rather than direct flow control. There are many variations in control design due to
different process configurations and personal preferences, including supply- versus
demand-driven operation, variable versus fixed-speed drive, and reciprocating versus
centrifugal machine. The typical control strategies are summarized in Table 4.3.
2 A persistent imbalance in steady-state supply and demand is a flaw in process design, often
resulting in continuous flaring in gas operations or overflow in liquid operations. Flaring used to be
a simple and popular “solution” to the material imbalance in gas plants but is no more viable due
to more stringent environmental regulations.
3 Line-packing refers to the transient condition where more gas is fed to the piping that is taken out,
usually leading to increased pressure.

The process capacity control determines the machine capacity control as part of
the overall process control strategy, where the supply-and-demand relationship is the
critical consideration. The supply and demand can be affected by any components
along the material flow path, including recycles and loop-backs. Hence, capacity
control must be considered within the overall process configuration based on a holistic
view of the complete process flow path.
An operation can be supply driven or demand driven, requiring significantly dif-
ferent control strategies (Liptak 2006; Niu and Xiao 2022). For supply-driven oper-
ations, the process flow capacity is typically maintained by controlling the header
pressure at the compressor suction or the tank level at the pump suction. Suppose
there is more supply than the current capacity. The suction pressure or tank level will
rise, and the pressure or level controller will increase the machine speed or open the
control valve to increase the throughput. The downstream operation must have the
swing capacity and swing-control mechanism to absorb the fluctuations in the flow.
See Fig. 4.4a for an illustration.
On the other hand, the discharge pressure is typically controlled for demand-
driven operation. When the demand increases, the discharge pressure will decrease.
The pressure controller at the discharge header will increase the machine speed or
open the discharge control valve to increase the throughput. Swing capacity and
swing control must exist upstream of the machine to accommodate the changes in
demand. See Fig. 4.4b.
The same supply-and-demand analysis for centrifugal machines applies to recip-
rocating machines. For example, an ordinarily supply-driven control is typically via
pressure control at the suction side (see Fig. 4.5a), while a demand-driven control is
via pressure control on the discharge side (see Fig. 4.5b).
Capacity control of reciprocating machines is achieved predominately with recy-
cling. Recycling is simple and reliable but inefficient, a price to pay for a simple
control solution.
Example 4.1 Overall control strategy for an upstream production facility. Figure 1.7
illustrates a complete surface processing facility for oil-and-gas production from the
wellhead to the final sales point. Multi-phase fluids from the production wells are
Fig. 4.4 Supply-driven versus demand-driven control

Fig. 4.5 Capacity control of reciprocating machines
Fig. 4.6 Overall control strategy for a general E and P process, supply driven (reproduced, with
permission, from Niu and Xiao (2022))
sent to the separators to be separated into gas, oil, and water, which are then sent to
the respective processing units with the help of pumps and compressors.
The operating objective is to maximize oil production; therefore, the production
rate from the wells determines the overall throughput (capacity), and the operation
is supply driven. The supply-driven control design is illustrated in Fig. 4.6. The
gas processing unit compresses the gas to the desired pressure and sends the gas to
downstream users ➁, following a supply-driven propagation. Similarly, the oil ➂ and
water ➃ processing units are also supply driven.
Assume the consumers of the produced gas have decided to dictate the gas flow
rate based on their actual needs. The gas processing unit then becomes demand
driven. The change in demand propagates back to the separator and the wells. The
well production must be able to adjust production rate to adapt to the changes.
This reversal from supply-driven to demand-driven operation requires a significant
modification in the process or control design. The re-designed overall control strategy
Fig. 4.7 Overall control strategy for a general E and P process, demand driven (reproduced, with
permission, from Niu and Xiao (2022))
Fig. 4.8 Supply-driven control, speed versus throttling valve
is shown in Fig. 4.7, with the stream ➁ being demand driven and the stream ➀ a
swing.
Fixed-Speed Versus Variable-Speed Control
Capacity control is typically achieved via speed change for variable-speed machines
or a throttling valve for fixed-speed machines, as shown in Fig. 4.8.
A fixed-speed machine is considerably cheaper in procurement, installation, and
maintenance than a variable-speed machine. However, the operation is less energy
efficient due to the throttling waste, and the overall operating cost can be significantly
higher in the long run. Moreover, a fixed-speed machine may require a larger capacity
motor to cope with the initial surge in electric current.
On the contrary, variable-speed compressors are more efficient than fixed-speed
ones, leading to significant energy savings, but investing in and maintaining a
variable-speed driver is considerably more costly, except for the inherently variable-
Fig. 4.9 Capacity turndown, fixed speed versus variable speed (reproduced, with permission, from
Niu and Xiao (2022))
speed turbo expanders. In addition, the variable-speed driver as another system (usu-
ally without backup) adds complexity to operation and vulnerability to failures.
Variable-speed control favors compressors with steep resistance curves, such as
pipeline compressors, while the throttling valve is more appropriate for relatively
flat resistance curves, such as off-gas compressors.
Split-Range Control for Turndown Operation
Operations require that the process flow be able to vary between zero and maximum
flow. However, centrifugal machines have a minimum flow limit imposed by the surge
phenomena. When the machine flow reaches the minimum flow limit, the recycle
flow must be introduced to keep the machine flow above the minimum flow limit
while reducing the process flow.
Combining the machine and recycle flow to turn down the process flow constitutes
a split-range control arrangement (King 2016; Niu and Xiao 2022). The machine flow
as the primary control handle reduces the process flow to the minimum flow limit; the
recycle flow is then introduced to reduce the process flow further to 0. The split-range
operation is illustrated in Fig. 4.9a.
The machine flow is the sum of the process and recycle flow (Fig. 3.5). The min-
imum flow limit is different at different speeds. Recycling can start at any operating
point. Recycling early provides more surge margin, but unnecessarily wastes energy.
On the other hand, introducing recycling too late can result in operating too close to
the surge line. See Fig. 4.9b.
There are other standalone regulatory control loops for rotating machines. For
example, temperature controllers at suction, discharge, and inter-stage coolers are
4.3 Protective Control: Anti-surge Control 103
Fig. 4.10 Compressor operating envelope
required to maintain the gas temperature4 to avoid excessively high discharge tem-
peratures. The scrubbers are typically on level control (on/off control or regular PID
control) to prevent liquid carry-over or gas carry-under. These standalone control
schemes are straightforward and will not be discussed further in this book (See Fig.
4.10 for an illustration of the operating envelope).
4.3 Protective Control: Anti-surge Control
The regulatory control discussed above assumes normal stable operating conditions.
In other words, normal regulatory control aims to maintain the operating point on
the target for stable operation. It is not designed and configured to respond swiftly
and decisively to prevent the operating point from moving away from the normal
operating region in case of a sudden process upset, which is inevitable in practical
applications. For this reason, protective control is provided to prevent the machine
from violating these operating limits. Protective control works separately from nor-
mal regulatory control. If the normal regulatory control fails to hold the setpoint, the
4In addition to the influence on gas temperature, the cooler will also cause a pressure drop of 35 to
70 kPa (Campbell 2004).
protective controller responds quickly and aggressively to bring the operating point
back inside the operating envelope (Botros et al. 2015; Kurz et al. 2016; Majumdar
1999; Niu and Xiao 2022; Mirsky et al. 2012).
Protective controls are also based on feedback principles. We will examine the
five essential components of feedback control loops one by one within the context
of protective control design.
4.3.1 Protective Control Objectives
Protective overriding control aims to keep the operating point inside the operating
envelope to avoid unstable or inefficient operations. Meanwhile, they should allow the
operating point to run close to the limits to maximize the operating region. Protective
control is thus a trade-off between safe operation and maximum operating range.
Pumps and compressors are designed to raise the pressure of the fluid to the desired
value, overcoming resistance from the piping, vessels, and equipment, which all have
design limits. The requirements of the process and the machine jointly determine the
operating envelope. On the process side, the pressures and temperatures must be
maintained within the design limits to protect the vessels and piping. While on the
machine side, the operation of centrifugal machines must stay clear from surge and
choke conditions. There may be other limits and constraints requiring additional
protective controllers. For instance, a high limit on motor current typically exists
for the safe operation of the motor. In the case of a variable-speed operation via a
variable-speed gearbox, the gearbox typically has a constraint on the torque.
The typical protective control requirements are listed in Table 4.4. However, the
actual implementation varies from application to application, depending on the spe-
cific process flow configuration; not all are required.
Table 4.4 Typical requirements and means of protective control

Limits and Constraints Pump Compressor Pump Compressor
Minimum Flow Recycle Flow Control Anti-surge Control − −
Maximum Flow End-of-Curve Control Anti-choke Control − −
Minimum Speed Speed Control
Maximum Speed Speed Control
Discharge Pressure Discharge Pressure Control
Discharge Temperature Discharge Temperature Control
Motor Power Motor Power Control
Gearbox Torque Gearbox Torque Control
Table 4.5 Cause-and-effect table for protective control

Control Handles Pressure Ratio Machine Surge Margin Pressure Ratio Machine
Flow Flow
Machine Speed ↑
Recycle ↑
4.3.2 Process Dynamics for Operating Envelope
Protective control is based on the process and equipment dynamics discussed in

Chap. 2. The dynamic characteristics of the machine near or at the operating envelope,
especially the dynamic cause-and-effect relationship between the control targets and
control handle near the surge line, are of primary interest, see Table 4.5 for the
cause-and-effect table.
Surge is the primary concern in a centrifugal machine’s operation. A surge occurs
when the machine flow (actual flow through the machine) is below the minimum
flow rate required by the centrifugal machine, e.g., due to low process flow (net
forward flow). Since the machine flow is the sum of the process and recycle flow, as
illustrated in Fig. 3.5, recycling is the most effective way to influence the machine
flow. Other means of influencing the machine flow include changing the machine
speed and varying the system resistance, but their impact is slower and less direct.
The choke phenomenon is not as damaging as surge and can be prevented by
discharge flow throttling or speed reduction.
In addition to surge and choke, the pumping and compression operation raises the
fluid pressure, accompanied by temperature rise. Protecting both the equipment and
piping against high pressure and temperature is compulsory.
4.3.3 Measurements and Controlled Variables
The objective of protective control is to keep the operating point away from the
limits and constraints imposed by the operation envelope, which requires reliable
measurements to indicate the real-time location of the operating point relative to
each operating limit. Some variables can be directly measured online, while others
can only be indirectly inferred from other variables based on engineering insight.
1. Machine flow. Fast and reliable flow measurement is the most important require-
ment for compressor and pump operation and control. A DP-based orifice plate
is typically used for flow measurement because of its simplicity, reliability, and
fast response.
The DP value is also a required measurement by most surge indicator calculations,

as shown in Chap. 5.
2. Pressure and temperature. Fast and reliable measurements of the pressure and
temperature at various points are also essential for protecting the machine and
piping.
3. Machine speed. Reliable speed measurement must be available to ensure the
machine does not operate outside of the stable speed range.
4. Motor power. The motor power or current must not exceed its maximum operating
limit. The motor power can be measured online or calculated from real-time
electric current and voltage measurements.
5. Gearbox torque limit. In the case of a variable-speed operation via a variable-
speed gearbox, the maximum torque limit must not be exceeded. Reliable online
measurement or calculation of the gearbox torque is needed protecting the gear-
box.
Surge avoidance is the most critical requirement for protective control. The primary
challenge is detecting when a surge is likely to occur and how much recycle flow
is needed to avert the surge. A surge indicator is typically used as the controlled
variable.
The surge indicator is not a variable that can be directly measured online. It
is usually calculated indirectly from other online measurements, such as flow and
pressure, based on engineering insight into the machine’s behavior. This type of
calculation is called inferential property or soft sensor, which is widely used in
process control. Its calculation can be rather involved and is thus deferred to Chap. 5.
Another approach to detecting surge conditions is monitoring the rate of change
in selected process variables such as machine flow, pressure, or temperature (Elliott
and Bloch 2021). For example, a rapid decrease in discharge pressure or an increase
in suction temperature may indicate an ominous surge. Although adding rate-of-
change-based protection can potentially increase reliability, the actual definition of
the surge signature and selection of the detection threshold can be quite tricky and
can result in many nuisance alarms or trips if not configured properly.
Motor power can be an inferential property as well. The motor driving the pump
or compressor has a rated limit on its power or electric current. A protective control
loop on the motor power can prevent the motor from unnecessary trips or damages.
The power consumption can be measured directly with a watt-meter. It can also be
calculated from online measurements of the electric voltage and current as follows,
assuming three-phase A/C power:
√
[Motor Power] = 3 × Current × Voltage × [Power Factor] (4.2)
Preal
[Power Factor] = . (4.3)
Papparent
The correct location of the flow and pressure meters contributes to the reliability
of measurements. For example, the flow transmitters should be placed above the
orifice plates with the minimum length of impulse lines to reduce response time. The
impulse lines should be self-draining to avoid liquid entrapment.
Redundant measurements are often pursued to improve the reliability of control
and safeguarding in some mission-critical applications. For control, three measure-
ments with a mid-of-three filter can considerably increase the measurement reliability.
For safeguarding, two-out-of-two (2oo2) or two-out-of-three (2oo3) voting logic
can improve the reliability and reduce nuisance trips.
4.3.4 Recycle Valve and Manipulated Variables
Protective overriding control shares control handles with the normal regulatory con-
trol. Based on the cause-and-effect table (CET) in Table 4.5, the three types of
control handles, including the machine speed, throttling valve, and recycle valve,
can all impact the surge margin. One significant difference among them, however,
is the speed of response. The recycle valve can affect the surge margin much faster
than the machine speed and throttling valve. Therefore, the protective control loops,
such as anti-surge control and discharge pressure control, use the recycle flow as the
primary control handle. A dedicated recycle valve is recommended for each pump
or each stage of a compressor,5 with the size, response time, and characteristics as
the most concerned specifications.
1. Valve size. The recycle valve is the most important control handle serving mul-
tiple purposes, including surge control, capacity control, and machine startup.
Therefore, an adequate size is crucial. Too small a valve does not provide suffi-
cient flow capacity to quickly empty the fluid entrapped in the surge volume (or
discharge volume) in case of a surge. Process engineers often habitually over-
specify the recycle valve for added assurance.6 An oversized valve increases the
5 In some older designs, a common anti-surge recycle line and valve may be shared by multiple
stages. The recycle line draws from the discharge of the last stage and ties in at the suction of the
first stage. This practice is discouraged because each stage has different characteristics and surge
margins. If a composite performance curve is used for anti-surge control design, the inter-stage
coolers significantly increase the uncertainties of the performance curves and negatively affect the
control performance. In addition, the large pressure drop across the recycle valve may result in a
drastic temperature decrease due to the Joule–Thomson effect. If the temperature downstream of
the values goes below freezing, ice may form and block the recycle line.
6 Typically, for large centrifugal compressors, a comprehensive dynamic simulation must be carried
out by the manufacturer to predict the operating point trajectory during emergency shutdown and
recommend the size and response time of the recycle valve. If it is deemed to be beyond the capacity
of the recycle valve to prevent surge during a shutdown, a second recycle line with an on/off valve
may be needed.
Fig. 4.11 Sizing of recycle valve
cost and also unnecessarily increases the response time, inadvertently affecting
the control performance.
The recommended size for the recycle valve is to have a maximum flow rate of
1.8 to 2.2 times the surge flow rate at the corresponding speed, as shown by line ➁
in Fig. 4.11.
2. Speed of response. The surge cycle is extremely fast, typically between tens of
milliseconds to a few seconds. The response of the anti-surge control loop must
be similarly fast to respond to the abnormal condition. Since the recycle valve
has the longest response time among all the components in a closed control loop,
selecting the recycle valve is thus of critical importance. A fast-responding valve
can become very expensive, so it is always a trade-off between the cost and
performance when deciding on the suitable valve for the job. A rule of thumb in
the industry is that in case of a surge, the valve should be able to open fully from
a closed position in less than two seconds.
In addition to selecting the proper valve, the piping configuration can help reduce
the response time. For instance, the surge volume of a centrifugal compressor is
defined as the gas volume between the non-return valves and the recycle valve
(Fig. 4.12). Once the valve is open, the accumulated gas inside this surge volume
must pass through the recycle valve as quickly as possible to reduce the discharge
pressure.7 Otherwise, a low flow and high head across the compressor may drive
the compressor into a surge.
A solenoid valve is usually added to the anti-surge control valve for emergency
shutdown (ESD) actions. Once the solenoid is de-energized, the fail-open recy-
7The minimum surge volume requirement also implies that the recycle line off-take should generally
be upstream of the discharge cooler (hot recycle) rather than downstream.
Fig. 4.12 Recycle surge volume
cle valve quickly opens and reaches its fully open position within two seconds,
propelled by the force of the spring.
3. Valve characteristics. A fast-open characteristic is preferred by protective control
for its fast response but is not ideal for capacity control that demands smooth
control performance.
The valve characteristics designated by manufacturers are based on a fixed pres-
sure across the valve and are called the inherent characteristics or simply the
manufacturer’s characteristics. When installed, the pressure across the valve is
typically not fixed. The pressure decreases as the valve opens. As a result, a linear
characteristic behaves more like a fast-open characteristic when installed. The
characteristics under actual operating conditions are called the installed charac-
teristics. To obtain fast-open installed characteristics, linear inherent characteris-
tics are normally required.
For reciprocating machines, there is no concern about surge. The recycle valve
is used for capacity control and discharge pressure protective control. The sizing of
the valve should match the operating range of the process flow.
4.3.5 Protective Control Algorithms
Protective control is typically based on standard feedback control schemes, such

as on/off control and PID control. It remains dormant during normal operations
and “kicks in” only when the operation runs into abnormal operating conditions.
For centrifugal machines, minimum flow control is the primary requirement for
Fig. 4.13 Auxiliary performance lines for centrifugal machine control
protective controls, also called anti-surge control in centrifugal compressor operation

and minimum recycle flow control in centrifugal pump operation.
Figure 4.13 summarizes the performance lines for controlling and safeguarding
centrifugal pumps and compressors.8 These lines use a surge indicator called the anti-
surge parameter (ASP) to indicate the surge margin, with a value from 0 to 100% (The
surge indicator and anti-surge parameter are discussed in detail in Chap. 5). These
lines will be frequently referenced when discussing control strategy and functional
design.
• Surge line. Connecting all the surge points at all machine speeds produces a line
that marks the boundary between the stable operating region (to the right of the
surge line) and the unstable surge zone (to the left of the surge line).
• Surge reference line (SRL). The surge line is a jagged line connecting all the surge
points. For the convenience of presentation and analysis, a smooth line with a
simple mathematical representation is desired in place of the jagged surge line.
This smooth line is called the surge reference line and typically takes the form of
a straight or parabolic line. The surge reference line is so defined that all operating
points to the left of the surge reference line are considered in the surge zone. The
surge reference line is set to ASP = 30%.
• Surge control line (SCL). The compressor is always operated a little distance from
the surge line for safety. This line is called the surge control line, typically 10%
(of flow) from the surge reference line at ASP = 33%.
8 Reciprocating machines work differently, and the performance lines are much simpler.
4.4 Control Integration and Optimization 111
• Capacity control line (CCL). Compressor capacity control requires that the oper-
ating point be further away from the surge reference line, typically 20% (by flow)
at ASP = 36%. This control line is called the capacity control line.
• Surge trip line (STL). The surge trip line defines the point at which the machine
is considered in a severe surge state and must be shut down. This line is typically
set halfway between the origin and the surge reference line, at ASP = 15%.
• Flow transmitter line. The flow transmitter lines show the minimum and maximum
flow lines corresponding to the 100% flow transmitter measurement range and the
30% flow turndown. The flow transmitter must provide reliable flow measurements
for the full operating range.
• Maximum flow line for recycle valve. The flow through the recycle valve can be
shown on the same diagram as the performance curve. When the recycle valve
is fully open, the pressure ratio versus flow rate is represented as a line similar
to the system resistance or surge reference lines. Note that the recycle flow is
very complex and depends on many factors; therefore, the line in Fig. 4.13 only
illustrates the concept.
• Choke line or stonewall. The choke line represents the maximum achievable flow
through the machine. The line is approximately at ASP≈ 100%.
The anti-surge control is a standard protective PID control, with the process value
being the ASP value. Anti-choke control, if implemented, uses the same ASP value
as controller input (PV) via an anti-choke controller. Because the stable operating
range of the centrifugal machine is from approximately 30% to 100% in terms of
ASP value, the anti-surge control keeps the ASP value above 33%, and a dedicated
anti-choke controller keeps the ASP value below 100%.
As rotating equipment raises the pressure and temperature, it is also essential to
protect the vessels, piping, and equipment against excessively high pressures and
temperatures.9 These protective controllers are standard protective PID controllers
and are generally straightforward. They are not discussed in detail here.
4.4 Control Integration and Optimization
A complete control solution for a process with pumps and compressors typically
includes one or more functionalities listed in Fig. 4.14. Normal regulatory control
and protective overriding control interact and potentially interfere with each other
and must work together to achieve better performance.
9An increase in head or pressure ratio implies an increase in discharge pressure, a decrease in
suction, or both. Therefore, protection against low suction pressure may be needed if the discharge
pressure is limited or controlled.
Fig. 4.14 Overall control strategy for pumps and compressors
4.4.1 Sequential Control: Mode Transition
A machine can operate in different modes, so is the process. Normal operation is one
operating mode. Shutdown mode is another. The reduced-capacity operation, known
as the turndown mode, is also pervasive.
The transition from one mode to another is associated with many risks, and a safe
and smooth transition is compulsory but rather complex. For instance, the startup
requirements and procedure can differ significantly with the machine type and process
configuration, affected by factors such as the machine speed, the position of the
recycle valve, and the throttling valves.
Sequence and logic are the primary means for managing the mode transition.
The startup sequence also depends on a reliably functioning process control scheme
(regulatory and protective control) to ensure the transition starts and ends in safe
conditions. A typical startup sequence from a practical application is provided in
Chap. 8.
4.4.2 Instrumented Safeguarding Against Failures
During normal operations, the pump or compressor is operated under regulatory con-
trol (e.g., capacity control) and protected by overriding control (e.g., minimum flow
control or anti-surge control). They serve as the first two lines of defense against
abnormal operations. In severe operating conditions, such as a major process upset,
the capacity and protective control may both fail to keep the machine inside the
desired operating envelope. In this case, the anti-surge trip mechanism kicks in
to proactively shut down the operation to avoid potentially severe damage to the
machine.
For centrifugal compressors, the anti-surge trip is based on the same surge margin
calculation as in anti-surge control. An anti-surge parameter value below the surge
trip limit (see surge trip line in Fig. 4.13) is considered a surge condition and triggers
a proactive shutdown of the compressor or the affected unit.
For reliability, the control and safeguarding functions require two separate and
independent sets of measurements and execution logic implemented in two segre-
gated systems, e.g., DCS and SIS. For critical operations such as liquidized natural gas
(LNG), redundant measurements with a voting mechanism are often implemented.
4.4.3 Online Performance Monitoring
Compressor performance monitoring is a crucial aspect of the long-term sustainable

operation of the control solution. Compressor vendors typically offer comprehensive
monitoring tools to ensure that the mechanical integrity of the machine is monitored
and abnormal conditions are detected to avoid catastrophic failures; however, they are
less interested or competent in monitoring process safety and operational efficiency.
Process control plays a key role in monitoring the process and control performance.
The display of the compressor map on the operator screen and the current real-
time operating point provides a visual cue of the status and health condition of the
compressor control system.
Continuous calculation and monitoring of the compressor’s key performance indi-
cators (KPI) and timely reporting of abnormal performance conditions are proactive
means of performance assurance for the process control solution. Chapter 8 provides
detailed discussions and some examples of performance monitoring.
4.4.4 † Integration of Capacity and Anti-surge Control
The capacity of a centrifugal machine can be controlled by changing the machine

speed, varying the system resistance, or introducing recycling. As shown in Fig. 4.15,
the capacity can be turned down from F1 to F2 by moving the operating point from A
to C through speed reduction, from A to B through throttling the suction/discharge
valve, or from A to D through recycling. The operating point can be moved from A
to anywhere on the map with flow F2 by combining all three control handles.
However, it is important to note that the three types of control handles differ
significantly in operating efficiency. The most efficient is adjusting the machine
speed, which forces the operating point to move along the constant-resistance line,
roughly parallel to the constant-efficiency lines (A to C). Valve throttling wastes
Fig. 4.15 Compressor control as multivariable control with interactions
energy because it increases the discharge pressure and then kills it at the valve.
Besides, it moves the operating point along the constant-speed lines (A to B) and away
from the best efficiency point (BEP), resulting in less efficient operation. Recycling
(A to D) is the least efficient due to the re-circulation and re-pressurization of the
fluid.
When multiple control handles are available, the most efficient ones should be
utilized before using the less efficient ones. An undesired scenario is running the
machine with maximum recycling while simultaneously throttling the control valve
or increasing the speed. For instance, as shown in Fig. 4.16, to bring the flow from
the current operating point A to the desired turndown flow, the recycle flow can be
introduced as early as point ➀ or as late as point ➄. The amount of recycling is
significantly different, so is the efficiency.
A straightforward solution integrating the capacity control and anti-surge control
is to bring the process flow from point A to point ➂; the anti-surge controller then
kicks in and aggressively opens the recycle valve to maintain the compressor flow
at point ➂ while the process flow continues to decrease. The abrupt transition from
smooth capacity control to aggressive anti-surge control when the ASP value crosses
the surge control line can potentially cause unacceptable operational disruptions.
An improved solution for integration between capacity control and anti-surge
control is shown in Fig. 4.17. When the process flow decreases during a smooth
turndown operation, the operating point at A moves left under capacity control to
reach point ➁, which is on the capacity control line (CCL). As a split-range control,
the capacity control then opens the recycle to further decrease the process flow (see
Fig. 4.5 for the split-range configuration). The machine flow remains at point ➁, and
the capacity controller determines the amount of recycling. The anti-surge control
Fig. 4.16 Unintegrated capacity control and anti-surge control
Fig. 4.17 Integrated capacity control and anti-surge control
would not be activated unless the capacity control fails, and the machine flow moves
to the right of point ➂. When the process flow recovers, the recycle flow will be
reduced first.
The capacity controller provides smooth capacity control during normal operation
with a split-range logic. Only when the capacity control is not fast enough will anti-
surge control temporarily come to the rescue as the second line of defense against the
potential surge. However, both actions manipulate the same anti-surge control valve
but with different levels of aggressiveness in response. See Sect. 7.1 for a design
example.
Integration between speed and recycling may not always be desirable, depending
on the steepness of performance curves and control trajectory. Similarly, variable-
speed drive (VSD) may not always be justifiable considering the costly investment
of a VFD motor or variable-speed gearbox (VSGB).
4.4.5 ‡ Load Balancing and Optimization
Multiple machines operating in parallel share the same head or pressure ratio, while
machines operating in series share the same mass flow. Due to these inherent con-
straints, the actual split of flows between each parallel train and the pressure profile
around each compressor stage can significantly impact the overall operating effi-
ciency and energy consumption (Jacobson et al. 2016; Nägeli et al. 1973).
Load-Balancing Among Parallel Compressors
Based on the affinity laws, power consumption increases by a cubic function of

machine flow, whereas head (compression ratio) increases by a quadratic function
of flow. Due to these differences in pressure and power response to flow, there are
optimal points for flow distribution (parallel operation) and pressure profile (serial
operation) that result in the highest power efficiency. See Fig. 4.18.
The load-sharing strategy aims to maximize the operating efficiency of a multi-
compressor network by optimally distributing the throughput among the multiple
machines and stages. Load sharing is achieved via load-balancing control, which
includes load-balancing control among multiple compressor trains (train balancing)
and among the multiple stages (stage balancing).
The load of a centrifugal machine typically refers to the machine flow. However,
forcing machines of different design to run at the same flow rate will not result in
the best efficiency. Figure 4.19 shows the performance curves of two machines on
the same head versus flow coordinate. The two machines have significantly different
capacities. Forcing them to run at the same flow rate can cause unnecessary early
recycling, resulting in a narrower operating range.
Surge prevention has a higher priority than operating efficiency, and balancing
on flow will not lead to the same surge margin during turndown. Therefore, load
balancing based on surge margin makes more sense than using flow rate or rotating
speed.
Balancing on surge margin, all stages of the compressors will operate at the same
surge margin (ASP value) during severe turndown operations. On the performance
map, all pumps or compressors stages will maintain the same distance to surge,
Fig. 4.18 Load balancing improves energy efficiency
Fig. 4.19 Load balancing among parallel compressors
resulting in minimum recycle flows and thus maximum efficiency. The total power
consumption can be considerably reduced.
Load balancing can be achieved by introducing a load controller for each train.
These load controllers receive the same surge margin request from the capacity
controller as the control target and manipulate the machine speed or throttling valve
to achieve the requested surge margin. For example, the two machines (A and B)
Fig. 4.20 Load balancing among compressor stages
in Fig. 4.19 will maintain the same surge margin (A1–A2 and B1–B2) under very
different flow rates (see Points A1 and B1). Section 7.2 presents a typical control
design for load-balancing control among different trains.
Load Balancing Among Multiple Stages
For multi-stage compressors on a common shaft, all stages will operate at the same
speed.
During severe turndown operations, a conventional capacity control design forces
all the recycle valves to open by the same amount as soon as any stage approaches
surge, even though some stages still have a healthy safety margin. Figure 4.20 shows
the performance curves of two machines A and B. To turn down the machine flow,10
machine A will run into surge condition first and open the recycle to reach point A2,
with a recycle flow of A1–A2. However, Machine B has ample surge margin to reach
point B1 before needing to recycle.
In practice, the different stage will not approach the surge line simultaneously.
Introducing the same amount of recycling as requested by Machine A (via the same
valve opening) unnecessarily wastes energy.
The improved strategy for capacity control among stages is to recycle only when
the machine reaches the surge limit at each stage. That is, each stage will start
recycling independently of each other and only when necessary. There are different
10Here, we use mass flow to illustrate the concept. All the stages share the same mass flow but may
have different surge margins.
References 119
ways of achieving this load-balancing control. A simple approach is calculating

the difference between the stages in real time. The surge margin of each stage is
individually adjusted by padding the surge margin with the difference between this
stage and the stage with the worst-case scenario. The opening of the recycle valve is
thus delayed until the adjusted surge margin reaches the surge limit. A typical design
is provided later in Sect. 7.2.
4.5 Summary
The pump and compressor control objective is maintaining one point and protecting
two envelopes. The one point is the operating point, and the two envelopes are the
operating and safety envelopes. The head and flow of the machine determine the
operating point. The operating envelope is defined by the limits and constraints
imposed by both the machine and the process. The safety envelope is defined by the
various design limits and safety constraints.
The control solution is typically implemented in a layered approach, including
regulatory control for maintaining operating points, protective overriding control for
maintaining the operating envelope, and instrumented safeguarding for guarding the
safety envelope. Together, they form the three lines of defense against abnormal
situations with increasingly aggressive and disruptive responses.
The capacity control design is based on the supply-and-demand model and the
cause-and-effect relationship of the entire process and can be rather complex. A well-
designed capacity control scheme can maintain the operating point at any desired
point within the operating envelope with carefully selected control handles, including
the machine speed, system resistance, and recycling.
Protective control is mainly concerned with surge avoidance. Surge is caused by
low flow through the machine and is prevented by recycling a portion of the flow
from discharge to suction to keep the machine flow above the minimum flow limit.
The primary challenge for control is that the minimum flow limit varies with the
machine speed and inlet condition and must be constantly re-calculated in real time.
Sequential control deals with the transition between operating modes, including
automated startup and shutdown. A safe and smooth transition is compulsory.
References
Bahadori A (2016) Essentials of oil and gas utilities process design, equipment, and operations.
Elsevier Inc
Borremans M (2019) Pumps and compressors. John Wiley & Sons Ltd
Botros K, Hill S, Grose J (2015) A new approach to designing centrifugal compressor surge control
systems. In: 44th turbomachinery and 31st pump symposia. Houston, TX
Botros K, Grose J, Hills S (2016) Centrifugal compressor surge control systems–fundamentals of
a good design. In: Asia turbomachinery and pump symposium. Singapore
Campbell J (2004) Gas conditioning and processing-the equipment modules, vol 2, 8th edn. John
M, Campbell and Company
CCC (2021) CCC and yokogawa R&D cooperation. Tech. rep., Compressor Con-
trol Company (CCC). https://www.cccglobal.com/wp-content/uploads/2022/10/CCC-and-
YokogawaRD-cooperation.pdf
Elliott H, Bloch H (2021) Compressor technology advances—beyond 2020. Walter De Gruyter
Giampaob T (2010) Compressor handbook: Principles and practice. The Fairmont Press
Jacobson W, Staroselsky S, Zaghloul M, McWirter J, Tolmatsky M (2016) Compressor load-sharing
control and surge detection techniques. In: Proceedings of the 45th turbomachinery and 32nd
pump symposia. Houston
Kurz R, Brun K (2017) Process control for compression systems. In: Proceedings of ASME turbo
expo 2017: Turbomachinery technical conference and exposition. Charlotte, USA
Kurz R, White RC, Brun K, Winkelmann B (2016) Surge control and dynamic behavior for cen-
trifugal gas compressors. In: Proceedings of asia turbomachinery and pump symposium. Turbo-
machinery Laboratory, Singapore
Lieberman NP, Lieberman ET (2014) A working guide to process equipment, 4th edn. McGraw-Hill
Education
Majumdar K (1999) Understand centrifugal compressor, equipment interaction. Hydrocarbon pro-
cessing, pp 55–66
McMillan GK (1983) Centrifugal and axial compressor control. Instrument society of America.
forty-second turbomachinery symposium. Houston, Texas
Nägeli JP, Spechtenhauser A, Aicher W (1973) Turbomachinary in base load natural gas liquidifi-
cation plants. In: Proceedings of the 12th world gas conference. Nice, France
Niu S, Xiao D (2022) Process control—engineering analyses and best practices. Advances in indus-
trial control. Springer
Vermillion L, Gracia J, Ilchenko M (2023) Solutions spotlight: Integrating process and turboma-
chinery control. Control magazine. https://www.controlglobal.com/podcasts/article/21546058/
yokogawa-electric-corporation-solutionsspotlight-integrating-processand-turbomachinery-
control
Zelenov A (2013) White paper-CCS implementation of surge prevention control system on yoko-
gawa stardom PLC. Tech. rep, Continuous Control Solutions (CCS)
Chapter 5
Invariant Coordinates and Surge
Indicators
So far, the description and analysis of pumps and compressors are all based on a
qualitative understanding of the head, flow, and speed relationship under a head-
flow coordinate system. This coordinate system has many distinct advantages for
conceptual understanding and theoretic analysis and has been the de facto standard
of describing machine performance. However, a primary drawback is that the head
and flow are not directly measurable and thus unavailable for real-time operation
and control. The actual coordinate system in practical applications must be based
on process measurements available in real time. Data availability is the major factor
that divides practical implementation from theoretical analysis, leading to different
coordinate systems used in theory and practice. This chapter explains the concept
of invariant coordinates and their significance for real-time control. It then shows
the different ways of defining and calculating the surge indicators, which are the
basis for capacity control and anti-surge control for practical applications discussed
in Chaps. 6 and 7.
5.1 Inlet Conditions and Invariant Coordinate Systems
A pump or compressor is expected to work under various operating scenarios, includ-

ing different types of gases and varying pressures and temperatures. The differ-
ent operating scenarios significantly impact the machine outlet condition and the
operation and control.
https://doi.org/10.1007/978-3-031-43122-7_5
122 5 Invariant Coordinates and Surge Indicators
5.1.1 API Datasheet for Compressor
When specifying the requirements of a centrifugal compressor, the process engineer

defines the expected machine behavior with the compressor datasheet. This design
datasheet is typically based on API Standard 617 and is thus also called the API
datasheet.
The API datasheet specifies the expected head and flow under different operating
scenarios, e.g., summer and winter conditions and light and heavy gases. The typical
data set must include the following:
• The gas property: molecular weight Mw , compressibility Z s , and the ratio of spe-
cific heat κs .
• The suction condition: pressure Ps and temperature Ts .
• The expected discharge condition in terms of the head/efficiency (H p /η p ) or
pressure/temperature (Pd /Td ), and the desired flow rate (mass flow Fm , standard
flow F0 , or volumetric flow Fv,s ).
Table 5.1 shows the API datasheet of five of the 13 different scenarios for a single-
stage compressor. When plotted under the head and flow coordinate system, the
design conditions of all the 13 operating scenarios in this example are shown in
Fig. 5.1.
Table 5.1 Compressor API datasheet

Variable Unit Case A Case B Case D Turndown N2
Suction Pressure bar(a) 65.8 65.8 65.8 65.8 5.00
Suction Temperature ◦C 16.30 16.50 13.50 21.70 30.0
Molecular Weight kg/kmol 201.9 20.21 19.19 19.72 28.1
Ratio of Specific Heat - 1.622 1.615 1.638 1.584 1.408
Compressibility - 0.809 0.817 0.829 0.830 1.000
Speed rpm 12,900 12,982 13,148 12,761 14,160
Polytropic Head kJ/kg 45,075 45,542 46,692 48,239 55,411
Polytropic Efficiency % 82.3 82.3 82.5 81.1 81.8
Suction Volumetric Flow m3 /hr 3,013 3,043 2,897 2,358 3,239
Mass Flow kg/s 57.10 57.10 51.40 41.76 5.00
Standard Flow MMSCFD 204.4 204.2 193.6 153.1 12.90
Power kW 3,191 3,224 3,070 2,549 364
Discharge Pressure bar(a) 101.5 101.5 101.5 101.5 8.7
Discharge Temperature ◦C 51.4 52.1 49.5 58.0 96.0
Ratio of Specific Heat - 1.621 1.615 1.580 1.561 1407
Compressibility - 0.831 0.840 0.852 0.852 1.000
Certified Yes
5.1 Inlet Conditions and Invariant Coordinate Systems 123
Fig. 5.1 Compressor design conditions
Although the mass flow through the machine is constant following the law of mass
conservation, the volumetric flow is heavily affected by the suction condition, defined
as the pressure Ps , temperature Ts , gas molecular weight M, and gas compressibility
Z s . This suction condition is called the inlet condition. The data that define the inlet
conditions are shown in the top 5 rows in Table 5.1.
The compressor manufacturer designs the impellers based on the API datasheet,
with the predicted characteristics of the machine described with the performance
curves (see Fig. 5.2), along with the design points. The machine’s operating envelope
is expected to align with the desired operating envelope of the process.1 For this
reason, the API datasheet can be deemed as the control objectives for the compressor
control.
The compressor manufacturer typically endorses a certified point (also called the
guaranteed point). As discussed later, this certified point is often used as the reference
condition in process control design. For this compressor, case B is the scenario with
performance guaranteed by the compressor manufacturer, as shown in Table 5.1.
5.1.2 † Impact of Inlet Conditions
A compressor is expected to work under a wide range of operating scenarios or inlet

conditions. For a centrifugal compressor, the performance curves are always tied to a
specific inlet condition since the inlet condition significantly impacts the machine’s
1Many times, compromises must be made to fill the gap between the desire of the process engineers
and the practicality of machine characteristics.
Fig. 5.2 Operating scenarios on performance map
performance. For example, an ethylene plant’s cracked gas compressor (CGC)

(see Fig. 1.11) compresses the off-gas from the cracking furnace. Depending on
the market demand, the cracking severity is frequently adjusted to meet the required
yields of ethylene, propylene, butene, and butadiene. The variation in the compo-
sition of the cracked gas is a change in the inlet condition that poses a significant
challenge to the operation and control of the cracked gas compressor.
The characteristics of a centrifugal machine are to increase the energy by working
on the fluid with blades; anything that can affect the fluid velocity or density at the tip
of the blades can impact the head generated or the flow produced. Most challenges in
analysis and design are in handling the variations in inlet conditions. An example of
the different scenarios is given in Fig. 5.2, where all the 13 scenarios in this example
can be visualized selectively or collectively by selecting the scenarios of interest
from the list of scenarios in the companion software tool CPACS.
From Euler’s equation in Sect. 2.3.3, the head is a function of the volumetric
flow and rotating speed. Although the changes in inlet conditions do not affect the
head/flow relationship, they significantly impact most other variables and relation-
ships. Table 5.2 shows the cause-and-effect relationship of the inlet condition with a
few key variables, assuming the volume flow rate remains constant.
The most prominent relationship is with inlet gas density. Higher molecular
weight, higher pressure, and lower temperature can all result in a higher density.
A higher gas density implies lower flow resistance, thus a higher mass flow and
compression ratio, resulting in an energy split favoring kinetic energy (flow) over
potential energy (head).
For this reason, it is often said that a heavier fluid (gas or liquid) is “easier” to
pump or compress, whereas “easier” means less power consumption to reach the
same pressure ratio. This behavior is illustrated in Fig. 5.3. The operating point
moves toward the bottom left as the gas molecular weight decreases (“lighter gas,”
Case A –5% MW) and moves toward the top right as the molecular weight becomes
higher (“heavier gas,” case A +5% MW). The shift of the operating point with a
Table 5.2 Changes in inlet conditions with constant volumetric flow

Head Mass Flow Pressure Ratio Gas Power
H Fm Pd /Ps W
Suction Pressure Ps ↑ - -
Suction Temperature Ts ↑ -
Molecular Weight M ↑ -
Compressibility Z ↑ -
Gas Density ρs ↑ -
Fig. 5.3 Impact of gas molecular weight
molecular weight roughly follows a parabolic line, similar to the constant-resistance

curve.
Similarly, it is “easier” for the same pump or compressor to operate in a Winter
environment than in Summer. In other words, winter operation has a lower “load” on
the machine than in Summer for the same throughput because of cooler weather.2
The impact of the inlet condition can be expressed in equation format and has been
briefly discussed in Chap. 2. The relevant equations are collected and reproduced
here for the convenience of reference:
2For this reason, controlling or limiting the suction temperature is compulsory. For compressor
operation, suction and discharge coolers are essential to maintaining the temperature.
n−1 σ
n Z s R Ts Pd n 1 Ps Pd
Hp = −1 = −1 (5.1)
n−1 M Ps σ ρs Ps
n−1
n σ1
Pd n−1 M ρs
= Hp +1 = σ Hp + 1 (5.2)
Ps n Z s R Ts Ps
Fv,s
Fm = ρs · Fv,s , Fv = (5.3)
ρs
σ σ
Fm Ps Pd Ps Pd
W = Fm · H p = − 1 = Fv,s · −1 (5.4)
ρs σ Ps σ Ps
Ps M
ρs = . (5.5)
Z s R Ts
For example, based on Eq. 5.1, increased compression ratio Pd /Ps , decreased
molecular weight M (“lighter” gas), or higher suction temperature Ts all cause a
higher head H p to be produced. At a constant speed, a higher head is traded with a
lower flow.
The relationship between polytropic head and volumetric flow does not change
with inlet conditions. Even though the inlet conditions are rather different, especially
in the nitrogen case, all the operating points fall on the nine speed lines, as shown
in Fig. 5.2. The surge points line up nicely to divide the performance map into two
distinct operating regions, the stable region to the right and the unstable region to
the left.
On the other hand, the relationship between the polytropic head and the mass flow
is drastically different with the inlet condition, as shown in Fig. 5.4. The surge points
are so scattered that no common feasible region can be defined on the performance
map under this polytropic head versus mass flow coordinate.
The relationships between other variables can be visualized similarly with the
CPACS tool by selecting the interested variable from the Y-Axis and X-Axis lists.
Fig. 5.4 Polytropic head versus mass flow

Fig. 5.5 Polytropic head versus volumetric flow, an invariant coordinate
The degree of scatteredness of the surge lines is significantly different under the
different Y-axis and X-axis choices.
5.1.3 Invariant Coordinate
All the performance curves and maps discussed so far are based on the head versus
flow coordinate system. The prominent advantage of the head versus flow coordinate
is that at a constant speed, the head is a function of the flow alone and is unaffected
by the type of gas, pressure, or temperature. On the performance map, the variations
in inlet conditions do not significantly affect the shape of the performance curves,
including the surge and choke points. As a result, all operating points fall nicely on
the individual speed lines, with insignificant deviations3 ; see Fig. 5.5. The surge lines
under the different inlet conditions converge to a narrow band. The head versus flow
coordinate is thus called an invariant coordinate, i.e., invariant to the inlet conditions
(Batson 1996; Bloch 2006; Mirsky et al. 2012).
If the most conservative surge lines are based to derive a reference line representing
all the required operating scenarios, then there will be no significant loss in the
operating range for any individual scenario. Therefore, the control design based on
invariant coordinates remains valid for all inlet conditions.
In contrast, with a coordinate system such as polytropic head versus mass flow,
as shown in Fig. 5.4, the scenario with nitrogen gas has a much different inlet con-
dition than the process gas and thus has drastically different performance curves.
3 The deviation becomes more prominent at higher flow or pressure.

It is impossible to find a surge reference line to represent all the scenarios without
significantly reducing the operating ranges of some operating scenarios. This poly-
tropic head versus mass flow relationship is said to vary with the inlet conditions and
is thus a non-invariant coordinate system.
Most commercial control designs vary primarily in their choice of invariant coor-
dinate (Hafaifa et al. 2014).
5.2 Equivalent Coordinates
Polytropic head versus volumetric flow is an invariant coordinate, thus the coordi-
nate of choice for conceptual understanding and theoretical analysis. However, in
practice, neither head nor flow is directly measurable for online applications; nor
can they be calculated online because the fluid density ρ that they depend on is not
directly measurable. Online control and monitoring must be based on an alternative
coordinate system, with the head and flow replaced by online measurements (Batson
1996).
5.2.1 † Alternative Variables for Polytropic Head
The performance map’s ordinate axis represents the fluid’s energy, with the fluid
head as the default variable. In addition to the head, the fluid energy can also be
represented by pressure in the form of reduced head, pressure ratio, or pressure rise,
as summarized in Table 5.3. The variables in pump operation (by assuming n = ∞)
are listed together for comparison and completeness.
The reduced head Hreduced is defined as the polytropic head, “reduced” by the
influence of suction pressure and gas density (Ps /ρs ):
Table 5.3 Potential variables for Y-axis in alternative coordinate system

Energy Type Centrifugal Compressors Centrifugal Pumps
Polytropic Head Hp H
σ
1 Ps Pd Ps Pd 1
Polytropic Head −1 −1 = [Pd − Ps ]
σ ρs P s ρs Ps ρs
1 Pd σ Pd
Reduced Head −1 −1
σ Ps Ps
Pd Pd
Pressure Ratio
Ps Ps
Pressure Rise Pd − Ps Pd − Ps
5.2 Equivalent Coordinates 129
σ σ
1 Ps Pd 1 Pd
Hp = −1 ⇒ Hreduced = −1 . (5.6)
σ ρs Ps σ Ps
The reduced head is a popular alternative to the polytropic head. It is used directly
or indirectly by most commercial anti-surge control solutions (Giampaob 2010;
Mirsky et al. 2012).
Pressure ratio Pd /Ps is the ratio between the (absolute) discharge pressure and
the (absolute) suction pressure. Since the pressures are directly available as online
measurements, the pressure ratio is the preferred alternative variable to the polytropic
head.
Pressure rise Pd − Ps is similar to pressure ratio, and is preferred in pump control.
5.2.2 † Alternative Variables for Volumetric Flow
The flow through a compressor is usually measured with a DP-based flowmeter,

such as an orifice plate or Venturi tube. The volumetric flow Fv is calculated from
the differential pressure P across the device as

P
Fv = C . (5.7)
ρ
Although the differential pressure P can be measured online, the fluid density
ρ cannot. Alternative flow representations must be used for online applications,
typically based on P. The common candidates include the reduced and equivalent
flows, as summarized in Table 5.4.
The reduced flow is similar to the reduced head. It is derived by removing the
same term Ps /ρs as in Eq. 5.6:
Table 5.4 Potential variables for X-axis in alternative coordinate systems

Flow Type Centrifugal Compressor Centrifugal Pump
Fv Fv
P P
Volumetric Flow C C
ρ ρ
√ √
Mass Flow C ρ P C ρ P
Differential Pressure P P
P P
Reduced Flow
Ps Ps
P P
Reduced Flow Squared
Ps Ps
Equivalent Flow Fe Fe

P Ps P Ps
Fv,s = C =C ⇒ Freduced = . (5.8)
ρs ρ
s Ps Ps
A related variable is the reduced flow squared:
Ps
Reduced Flow Squared = Freduced
2
= . (5.9)
Ps
The parabolic relationship between the reduced head and reduced flow becomes
linear with reduced flow squared. The theoretical SRL becomes a straight line instead
of parabolic.
The downside of using reduced flow squared is that the relationship is no longer
between head and flow, but between head and P. The reduced flow squared is no
longer a flow, and has no clear physical meaning in its value and engineering unit.
The preferred alternative is the equivalent flow, a reduced flow scaled with the
measurement range of the suction pressure, gas density, and mass flow. Suppose the
flowmeter produces a flow of Fm,r with a differential pressure of Pr at the design
condition of Pr . The equivalent flow under the actual suction condition of Ps and
from the actual DP of Ps is given by

Ps,r
Fe = Fm,r Freduced
Pr

Ps,r P
= Fm,r . (5.10)
Ps Pr
Unlike the reduced flow, the equivalent flow has the same engineering unit and
measurement range as the actual flow.4 It equals the true flow when the operating
condition matches the calibration condition of the flowmeter.5
5.2.3 † Equivalent Coordinate Systems
With the alternative variables for head and flow listed in Tables 5.3 and 5.4, the
possible choices for alternative coordinates are shown in Table 5.5. Evidently, not
all combinations are suitable for practical applications, as demonstrated in Fig. 5.4.
The head versus flow is called an invariant coordinate because the head and
flow relationship does not change with inlet conditions. The relationship remains
4 Caution should be taken that the equivalent flow is the uncompensated flow. It is different from
the true flow. See Niu and Xiao (2022) for discussions on mass flow, volumetric flow, actual flow,
standard flow, and flow compensation.
5 The so-called equivalent speed adjusts the speed for the inlet condition changes. All the points
with the same speed line up on the same equivalent-speed line rather than the actual-speed line.
5.2 Equivalent Coordinates 131
Table 5.5 Potential variables for describing head versus flow characteristics
invariant if both head and flow are scaled by the same value as in Eqs. 5.6 and 5.8.
The resulting coordinate system is deemed invariant as well. We will call it the
equivalent coordinate system. All equivalent coordinate systems are marked with
“≡” in Table 5.5.
Further simplification needs to exclude the σ term from the calculation, which
introduces further approximation. However, because the variation in σ is relatively
small and slow, the additional approximation introduced is usually within an accept-
able range. The resulting coordinate systems are regarded as almost invariant to
inlet conditions and are marked with “≈” in Table 5.5. All the equivalent coordinate
systems marked with ≡ or ≈ in Table 5.5 are being in use in practical applications.
However, due to approximations and errors, an invariant coordinate system is not
truly invariant to inlet conditions. Similarly, an equivalent coordinate is not truly
equivalent in the strict sense. Determining which one to use is always a trade-off
between complexity and accuracy.
Example 5.1 Performance map under equivalent coordinate systems. With the com-
pressor data in Fig. 5.2, the performance map can be plotted by selecting the Y-axis
and X-axis from the drop-down lists in the CPACS tool. The different performance
maps are shown in Fig. 5.6.
Figure 5.6a shows the performance map under the polytropic head versus volu-
metric flow coordinate system. Since the polytropic head is a fixed function of the
volumetric flow under a constant speed, unaffected by changes in inlet conditions,
all the operating points collapse nicely on the nine speed lines. All the surge points
line up neatly along a common surge line.
Figure 5.6b shows the performance map under the reduced head versus reduced
flow coordinate system. The performance curves are also invariant to inlet conditions.
However, they no longer align with the specific speeds since the variations in density
have caused the lines to shift. The shift is along the constant-resistance line in the
direction of speed change and does not affect the accuracy in indicating the division
between the stable and unstable regions.
Fig. 5.6 Performance curves under equivalent coordinate systems
Similarly, if the pressure ratio is plotted against the equivalent flow, the perfor-
mance map would appear as in Fig. 5.6c. The surge points also line up nicely despite
the different inlet conditions but with a shift along the resistance line.
Pressure rise (Pd − Ps ) versus flowmeter P is a primitive coordinate used by
some early commercial service providers. It has the advantage of simplicity, where
only two differential pressure measurements, one across the compressor and the other
across the flowmeter, are required. However, its deviation from the true invariant
coordinate (head versus flow) is often too significant to be reliable. Figure 5.6d
shows the equivalency where it is seen that the surge points can still be treated as
on the same line even though the performance curves with the nitrogen case have
shifted drastically toward the origin (due to low pressure and flow).
Anti-surge control is a significant part of process control for pumps and compres-
sors. The significance of choosing invariant coordinates for practical applications,
especially anti-surge control, will be explained later.
5.3 Surge Reference Line and Surge Indicators
Surge is an abnormal condition that must be avoided. However, to prevent the machine
from entering surge, the prerequisite is knowing the location of the surge limit and
the current operating point in relation to the surge limit, in real-time quantitative
terms.
5.3 Surge Reference Line and Surge Indicators 133
5.3.1 Surge Reference Line
The surge line is the dividing line between each operating scenario’s stable and
unstable operating regions. It is a jagged line due to errors and approximations in
calculation or measurements. The complexity of the shape depends on many factors,
with inter-stage cooling being a primary one. For instance, a composite surge line for
a multi-stage compressor can be a complex curve containing multiple breakpoints,
making the surge line deviate significantly from the expected parabolic line (see
Fig. 2.15) because of the heat removal by the inter-stage cooler.
Although a jagged line poses little challenge to human interpretation, automated
operation and control require a simple and smooth line. This smooth line is the surge
reference line (SRL), also called the surge limit line (SLL) (Mirsky et al. 2012). See
Fig. 5.7 for an illustration of the surge line and the surge reference line.
Each operating scenario (or case) has a separate surge line (see Fig. 5.5). The oper-
ating point is expected to stay a safe distance from the surge lines under all operating
scenarios. The surge reference line is thus dictated by the worst-case scenario, and
must be on the left side of all surge points. The invariant coordinate allows one surge
reference line to represent all the scenarios without significantly compromising the
feasible operating range.
There are many ways of defining the surge reference line. All definitions of the
surge reference line are approximations only since the exact form of the surge line
is never known. One common approach is based on the affinity laws (Eq. 2.36).
Assume everything remains constant. Changing the machine speed causes the
operating point to move along the constant-resistance curve. The ratio of head change
Fig. 5.7 Surge line and surge reference line

Fig. 5.8 Surge reference line: flow versus flow squared
is a quadratic function of the flow ratio for any two points on the constant-resistance
curve6 :
H p,2 Fv,2 2
≈ , (5.11)
H p,1 Fv,1
where H p,2 and H p,1 are the polytropic head of two arbitrary points along a system
resistance line; Fv,2 and Fv,1 are the corresponding volumetric flows.
Equation 5.11 represents a linear function between head H p and flow Fv2 , with
slope S:
H p = S · Fv2 , (5.12)
which is parabolic under the head (H p ) versus flow (Fv ) coordinate (Hafaifa et al.
2014). The relationship between the head and reduced flow squared (Fv2 ) becomes
a straight line,7 stretching from the origin through the operating point, with slope S,
as shown in Fig. 5.8b:
Hp
S= , (5.13)
Fv2
and S is thus an angular measure of the location of the operating point in an x-y-
coordinate.
Figure 5.9 shows the constant-slope line on the performance map of a centrifugal
compressor for any operating point. Each operating point on the performance map
falls on an imaginary line with its slope given by Eq. 5.13.
6 This quadratic function is an approximation. It shows an increasing deviation and non-linearity

as the compression ratio increases.
7 A straight line can be assumed as long as the deviation from fan laws is insignificant and the gas
composition remains relatively constant (Mirsky et al. 2012).

Fig. 5.9 Slope of the operating point versus the slope of the surge reference line
The surge reference line (SRL) is the most conservative constant-slope line with
all the surge points on the left of it. By definition, all points on the SRL have the
same slope. Let us define this slope as
H p,r
Sr = 2
, (5.14)
Fv,r
where H p,r and Fv,r are the head and flow of a selected surge point (a reference point).
The constant-slope SRL coincides with the constant-resistance line since they follow
the same DP versus flow relationship.
The location of the operating point in relation to the surge reference line is indi-
cated as follows:
⎧
⎪
⎨= Sr on the surge reference line
S < Sr on the right of the surge reference line (5.15)
⎪
⎩
> Sr on the left of the surge reference line,
where Sr is the slope of the SRL.

To ensure all the surge points stay on the left of the SRL, the surge point (H p,r ,Fv,r )
selected for SRL calculation in Eq. 5.14 must be the most conservative one.
The accuracy of the SRL determines how safely the compressor can operate. Its
accuracy also impacts the size of the operating envelope.
5.3.2 Surge Indicators
The SRL provides the basis for calculating the surge margin, defined in terms of the
current flow Fv in relation to the surge flow Fv,r :
Fv − Fv,r
Surge Margin = · 100% . (5.16)
Fv,r
In other words, a 10% surge margin leads to
Fv − Fv,r
10% = · 100% ⇒ Fv = (1 + 10%) · Fv,r . (5.17)
Fv,r
The SRL by itself is not a convenient surge indicator for practical applications.
A more intuitive variable is preferred for implementation and interpretation. This
variable is the surge indicator, a real-time indication of the surge margin.
With the surge reference line defined in Eq. 5.12, the surge indicator can be defined
as the ratio between the slope of the surge reference line Sr and the slope of the current
operating point S (Giampaob 2010; Mirsky et al. 2012):
Sr
SI = . (5.18)
S
The value of SI indicates the location of the current operating point as follows:
⎧
⎪
⎨= 1.0 on the surge reference line
SI > 1.0 on the right of the surge reference line (5.19)
⎪
⎩
< 1.0 on the left of the surge reference line.
Most anti-surge control technologies on the market are based on this concept but
differ in implementation.
With the slope definition in Eq. 5.13, the surge indicator for any operating point
on the performance map is given by
Sr
SI =
S
H p,r Hp
= 2
Fv,r Fv2
2
Fv Hp
=
Fv,r H p,r
Fv2
=C , (5.20)
Hp
where C is a coefficient that lumps all the constant values. Since there is only one
unknown coefficient, one surge point from the offline performance data is sufficient
to determine its value, e.g., with the most conservative surge point on the head-flow
performance map, similar to the calculation in Eq. 5.14.
Once the C coefficient is known, the real-time SI value of any operating point can
be calculated with the head H p and flow value Fv , if available online.
Surge Indicator Based on Reduced Head and Flow
The practical issue with Eq. 5.20 is that the head H p and flow Fv are not directly
available online, but the pressures and orifice DP values are. Therefore, the head and
flow need to be calculated from the pressure ratio Pd /Ps across the compressor and
the differential pressure P across the flow-measuring orifice, given by
⎧ σ
⎪
⎪ 1 Ps Pd
⎪Hp =
⎨ −1
σ ρs Ps
(5.21)
⎪
⎪ 2 P
⎪
⎩ Fv = C 2 .
ρs
With Eq. 5.21, the SI calculation is transformed to a function of the pressure ratio
and flowmeter DP:
Fv2
SI = C1
Hp
2
P
C2
ρ
= C1 s σ
1 Ps Pd
−1
σ ρs Ps
P
Ps
= C (5.22)
1 Pd σ
−1
σ Ps
2
Freduced
=C , (5.23)
Hreduced
where C is a coefficient that lumps all the individual constant terms ( C1 and C2 )
in Eq. 5.22. The denominator is “reduced” to the reduced head (Eq. 5.6) and the
numerator to the reduced flow (Eq. 5.8).
The surge indicator can be directly calculated using the reduced head and reduced
flow with Eq. 5.22. It can also be calculated indirectly from the head and flow with
Eq. 5.20, with the head and flow back-calculated from the reduced head and reduced
flow with 5.21 first.
Equation 5.22 is the generic formula from which most anti-surge control tech-
nologies are derived. This coordinate is made popular by the Compressor Control
Company (CCC), one of the key providers of compressor control solutions and ser-
vices (Bloch 2006; Honeywell 2023; Mirsky et al. 2012).8 However, many bells
and whistles are added to enhance the control performance in the CCC solutions.
8 CCC was acquired by the DCS vendor Honeywell in April 2023.

The Swiss-army-knife design style for its software makes its solution extraordinarily
versatile and, at the same time, very complicated.
Due to its large install base (over 14,000 applications per Honeywell 2023), it is
beneficial to understand this coordinate system to support the operation and migration
of the existing CCC applications efficiently.
One convenient feature is that an offset B is added as the control margin in the
actual applications to define a new controlled variable DEV. The control objective in
the CCC solution is thus to keep the DEV value above zero:
⎧
⎪
⎨= 0.0, on control line
DEV = SI − 1 − B1 > 0.0, above control setpoint (5.24)
⎪
⎩
< 0.0, below control setpoint.
Under the reduced coordinate of Hr versus Fr2 , the surge reference line represented
by the basic SI formula in Eq. 5.22 is a straight line from the origin, going through
the selected surge point. There is one single constant C to be determined from
the performance data, requiring a minimum of one surge point. The manufacturer-
provided performance curves typically have multiple speed lines and, thus, multiple
surge points. Due to non-linearities and measurement errors, all the surge points will
not produce the same C-values. A simple approach is taking the most conservative
point to calculate the C-value but will result in large inaccessible regions. The CCC
solution uses ad hoc characterizers to modify the slope calculation to approximate
the surge line more closely for improved accuracy:
f 2 (Z 2 ) f 3 (Z 3 ) f 1 (Hreduced ) Ps
Ss = K ·
P
f 1 (Hreduced )
= K · f 2 (Z 2 ) f 3 (Z 3 ) ·
P/Ps
f 1 (Hreduced )
= K · f 2 (Z 2 ) f 3 (Z 3 ) · 2
. (5.25)
Freduced
The characterizers f 1 , f 2 , and f 3 improve the curve fitting to the surge points to
arrive at a non-straight surge limit line, which provides a better fit to the jagged surge
lines to accommodate practical uncertainties in the data.
Surge Indicator Based on Pressure Ratio and Equivalent Flow
API 670 (API 2014) recommends calculating the surge indicator with flow, pressure,
and temperature (Elliott and Bloch 2021), similar to Eq. 5.22. However, the polytropic
index σ is a function of temperature, and the temperature measurement is typically
not fast and reliable enough for online calculation.
The dependence on σ can be removed without detrimental impact on the invariance
property of the coordinate system to inlet conditions. Taylor’s expansion is one way
of approximation:
σ 2 3
1 Pd Pd Pd Pd
− 1 ≈ C0 + C1 + C2 + C3 + ··· . (5.26)
σ Ps Ps Ps Ps
The generic SI formula based on the reduced head versus reduced flow in Eq. 5.22
becomes9 :
2
Freduced
SI = C
Hreduced
P/Ps
= 2 3 . (5.27)
Pd Pd Pd
C0 + C1 + C2 + C3 + ···
Ps Ps Ps
To further facilitate the understanding, the equivalent flow Fe is used in place of

the reduced flow in the numerator (see Table 5.4). The SI formula thus becomes

2
Freduced
SI = C
Hreduced
Freduced
=C √
Hreduced
Fe
= 2 3 × 100% (5.28)
Pd Pd Pd
C0 + C1 + C2 + C3 + ···
Ps Ps Ps

Ps,r P
with Fe = Fm,r , (5.29)
Ps Pr
where Ps,r , Fm,r , and Ps,r are the pressure, flow, and DP at the flowmeter design or
calibration condition.
With the square root, the numerator is transformed from quadratic to linear in
relation to the volumetric flow. This transformation allows for a more intuitive inter-
pretation of the surge indicator, which can now be viewed as a function of the flow,
proportionately scaled by a function of the pressure ratio. Anti-surge control, in turn,
becomes a matter of minimum flow control, consistent with pump operation.
Assuming that the surge reference line is assigned a value of 30%, the stonewall
will have a SI value of approximately 100%, owing to the 1/3 turndown ratio from the
9 There are many variations in simplifications. One variation is the same formula in Eq. 5.27 but
without the square root. Removing the square root in Eq. 5.27 still results in a valid (and popular)
ASP formula but tends to be less intuitive since the formula becomes a function of DP instead of
flow.
stonewall to the surge point (see Sect. 3.2.3). In other words, the range of the surge
indicator is from zero to the stonewall (0∼100%), with the surge point at 30%.10
From now on, the surge indicator (SI) in Fig. 5.28 will be called the anti-surge
parameter (ASP), with a re-defined range of 0 to 100%. The location of the operating
point can be indicated with ASP as follows:
⎧
⎪
⎪ < 30% on the left of the surge reference line
⎪
⎨= 30% on the surge reference line (minimum stable flow)
ASP (5.30)
⎪
⎪ > 30% on the right of the surge reference line
⎪
⎩
≈ 100% near the choke line (maximum flow).
The coefficients C0 , . . . , C3 are determined offline during control design with the
pressure and DP data of the surge points on the performance curves. The number of
coefficients to use is determined by the level of irregularity of the surge line, thus the
accuracy of the approximation.
Surge Indicator Based Directly on Orifice DP
The simplest solution requires one coefficient C0 and is based on two differential
pressures, one across the compressor and one across the flowmeter (orifice). If the
polytropic index σ remains constant and the polytropic head H p is not excessively
high, the ASP formula can be simplified from Eq. 5.28 as follows:
√
Fe P
ASP1 = or, ASP1 = C √ . (5.31)
Pd Pd − Ps
C0 −1
Ps
An example of the SRL with a single coefficient is shown in Fig. 5.10a.

The ASP formula with two coefficients (C0 and C1 ) provides the best trade-off
between accuracy and simplicity and is sufficient for most applications:

Fe P
ASP2 = or, ASP2 = . (5.32)
Pd C1 Pd + C0 Ps
C0 + C1
Ps
An example of the SRL with two coefficients is shown in Fig. 5.10b.

A formula with three or more coefficients is used for compressors with perfor-
mance curves that significantly deviate from the parabolic shape. The more coeffi-
cients are used, the better fit can be achieved, and the less operating area is wasted
10The stonewall is usually provided with less precision and is of less interest than surge point, we
simply set the surge reference line (SRL) at 30%, with the stonewall being roughly at 100%.
(see the shaded areas). Examples of SRL with three and four coefficients are shown
in Fig. 5.10c and d.
However, more parameters also mean higher complexity. It is thus a trade-off
between performance and complexity.
Surge Indicators for Pump Minimum Flow Control
In pump operation, the fluid is incompressible liquid with a constant density. The
equivalent flow is simplified to a mass or volume flow assumed to be directly mea-
surable. The single-coefficient ASP formula is derived from Eq. 5.31 as
⎧
⎪
⎪ Fm
⎪
⎪C√ , with mass flow measurement
⎪
⎪ Pd − Ps
⎪
⎪
⎨ Fv
ASP1 = C √ , with volumetric flow measurements (5.33)
⎪
⎪ Pd − Ps
⎪
⎪
⎪
⎪ P
⎪
⎪
⎩C , with DP measurements.
Pd − Ps
Fig. 5.10 Surge reference lines with one to four parameters

Generally speaking, the pump curves are much better-behaved than compressor
curves, as shown in Fig. 5.8, where the surge line closely resembles a parabolic line
(or straight line in head versus flow-squared coordinate). As a result, this single-
coefficient formula is sufficient for most pump control applications. However, if it
does not, the ASP formula with two coefficients can be used, simplified from Eq. 5.32:
⎧
⎪
⎪ Fm
⎪√
⎪ , with mass flow measurement
⎪
⎪ C 0 s + C 1 Pd
P
⎪
⎪
⎨ Fs
ASP2 = √ , with volumetric flow measurements (5.34)
⎪
⎪ C0 Ps + C1 Pd
⎪
⎪
⎪
⎪ P
⎪
⎪
⎩ with DP measurements.
C0 Ps + C1 Pd
5.4 Calculation of Anti-surge Parameters
The anti-surge parameter (ASP) provides a convenient indication of the location of

the operating point in relation to the surge limit. When the operating point is below
the ASP control setpoint, the anti-surge control opens the recycle valve to recirculate
a calculated flow from the discharge side to the suction.
The surge control line (SCL) is the actual control line by adding a sufficient control
margin to the SRL (Giampaob 2010; Mirsky et al. 2012). With the SRL at 30%, the
SCL is typically set to 33%, which is 10% away from the SRL (30%×(1 + 10%) =
33%, see Fig. 4.13). An insufficient surge margin can put the compressor into a
surge; an excessively conservative surge margin can result in unnecessary recycling
and energy waste (Elliott and Bloch 2021).
We will conclude this chapter with a practical example to demonstrate the concept
and calculation of the surge indicator. The example consists of a small subset of data
from an actual compressor control application; therefore, it is possible to complete
the calculation with pencil and paper. The full data set is overwhelmingly large for
manual calculation, requiring a dedicated software tool to facilitate the calculation.
See Chaps. 6 and Appendix A for a demonstration of the work process of anti-surge
parameter calculation using the companion software tool CPACS.
Example 5.2 Calculation of anti-surge parameter. The first step in anti-surge param-
eter calculation is collecting all the relevant performance data into a spreadsheet. The
required data include the compressor performance data and the flowmeter data from
the respective manufacturers. For this example, the performance curves under head
versus flow coordinate are shown in Fig. 5.11, which must be digitized as the first
step.
The manufacturer-supplied data may not be in the desired format or engineering
units as needed in real-time operation. Careful attention is needed to identify and
5.4 Calculation of Anti-surge Parameters 143
Fig. 5.11 Compressor performance curve
transform the data. Ensuring that the engineering units used during design match the
engineering units during operation for all the variables is crucial.
The minimal data set is collected and shown in Table 5.6. The second column
is the minimum flowmeter data extracted from the flowmeter datasheet. The surge
points are shown in the second section of the table, corresponding to the 8 points
marked as 1 to 8 in Fig. 5.11. The actual measurement data from incipient surge tests
(see Sect. 8.3.1) are included in the table to show the difference between the offline
data during design and online data during operation.
Table 5.6 Example: compressor performance data

The objective is to determine the coefficients C0 , C1 , C2 , . . . in the ASP formula

in Eq. 5.28, which requires the pressure ratio (Pd /Ps ) and the equivalent flow (Fe ).
The pressure ratio is the ratio between the discharge and suction pressure (abso-
lute), and the calculation is thus trivial. Calculating the equivalent flow, Fe , requires
the flowmeter’s DP value (P).
The flowmeter data provides the vital link between offline design data and online
measurement through the flow coefficient C, which can be calculated as follows:
Ps M 5.11 × 100 × 21.81

ρs = = = 4.06 kg/m3 (5.35)
Z s R Ts 0.99 × 8.314 × (60.00 + 273.15)
Fm 23.50
C=√ =√ = 4.41. (5.36)
Ps ρs 7 × 4.06
Since the flow coefficient C remains unchanged for any fluid once the orifice is
built, the P value can be calculated from the given mass flow rate Fm and the gas
density ρ in Table 5.6 for all the operating points, even though they have different
operating conditions or gas properties:
1 Fm2
Fm = C ρ P → P = . (5.37)
C2 ρ
Once all the surge points are identified, the gas density ρ and the differential
pressure (P) for each point are then calculated. These values are used to calculate
the pressure ratio and equivalent flow, which are presented in the final two rows of
Table 5.7:
Table 5.7 Example: compressor performance data, with compression ratio and equivalent flow
⎧
⎪
⎪ Pd
⎨ = Pd /Ps
Ps
P (5.38)
⎪
⎪ Ps,r
⎩ Fe = Fm,r .
Pr Ps
With the pressure ratio (Pd /Ps ) and equivalent flow (Fe ) available, the coefficients
of the ASP formula are calculated by fitting the data to the following equations, which
is rearranged from Eq. 5.28:
2 3
Pd Pd Pd
Fe2 ⇐ C0 + C1 + C2 + C3 . (5.39)
Ps Ps Ps
The ASP calculation becomes a curve-fitting problem with a polynomial equation:

2 3 2
Pd Pd Pd
J = min Fe2 − C0 + C1 + C2 + C3 . (5.40)
C0 ,C1 ,... Ps Ps Ps
There are numerous tools for solving the problem. A handy approach is using the
“Trendline” feature in Excel’s scatter plot, as shown in Fig. 5.12. The results are as
follows, with two to four coefficients:
⎧
⎪ Pd
⎪
⎪ F 2
= −1594.44 + 1292.11
⎪
⎪
e
⎪
⎨ Ps 2
Pd Pd
Fe = 920.78 − 936.44
2
+ 476.89
⎪
⎪ Ps Ps
⎪
⎪ 3
⎪
⎪ Pd Pd 2 Pd
⎩ Fe = −3004.11 + 4299.22
2
− 1803.44 + 324.44 .
Ps Ps Ps
(5.41)
We can verify that the ASP values at the surge points are all close to 30%.
In actual implementation, the surge reference line should be chosen so that all
the surge points fall on or to the left of the line. The unconstrained curve-fitting
solution shown in Fig. 5.41 has half the surge points on the right-hand side of the
surge reference line, which is usually unacceptable. The proper solution is by solving
a curve-fitting problem as follows:
2 3 2

Pd Pd Pd
J = min − C0 + C1
Fe2 + C2 + C3 (5.42)
C0 ,C1 ,... Ps Ps Ps
2 3
Pd Pd Pd
s.t. Fe − C0 + C1
2
+ C2 + C3 ≤ 0.
Ps Ps Ps
Fig. 5.12 Surge reference lines calculated with Excel
Dedicated tools are needed to solve this constrained curve-fitting problem. Opti-
mization algorithms such as Excel Solver11 can be used to find the constrained solu-
tion. A non-negative least-squares (NNLS) method (Lawson and Hanson 1974) is
used in the companion software CPACS, which is a fast and reliable analytic solution
for solving this curve-fitting problem. See Appendix A for more information on the
software tool.
The surge reference lines calculated with CPACS are shown in Fig. 5.13. Note
that all the surge points are now on or to the left of the surge reference line. The
corresponding ASP formulas are given by
⎧
⎪ Pd
⎪
⎪ F 2
= −1616.28 + 1377.04
⎪
⎪
e
⎪
⎨ Ps 2
Pd Pd
Fe = 840.81 − 862.54
2
+ 473.66
⎪
⎪ Ps Ps
⎪
⎪ 3
⎪
⎪ Pd Pd 2 Pd
⎩ Fe = −3204.93 + 4562.07
2
− 1900.58 + 338.55 .
Ps Ps Ps
(5.43)
11 Microsoft Excel® is a trademark of Microsoft Corporation. Solver is part of Excel.

Fig. 5.13 Surge reference lines calculated with CPACS tool
The ASP values for all the surge points are listed in Table 5.8. All ASPs have a value
of 30% or smaller, indicating that they are all on (ASP = 30%) or to the left (ASP
< 30%) of the SRL.
The surge reference line with three coefficients is selected as the best trade-
off between complexity and accuracy for this example, as shown in Fig. 5.43. The
corresponding ASP formula is thus given by
Fe
ASP = 2 . (5.44)
Pd Pd
840.81 − 862.64 + 473.66
Ps Ps
The final performance map with the surge reference line (SRL), surge control line
(SCL), capacity control line (CCL), and surge trip line (STL) is given in Fig. 5.14
(comparing with Fig. 4.17).
Validating the performance curve and surge reference line via incipient surge test
is a critical step in commissioning, which is discussed in detail in Chap. 8. The
three surge points from the incipient surge tests are shown in Fig. 5.14. Their values
are close to the surge reference line, as shown in Table 5.8, which confirms that
the manufacturer-provided performance curves match well enough with the actual
compressor in the field.
Table 5.8 Example: compressor performance data, with anti-surge parameters
Fig. 5.14 Example of performance map
5.5 Summary
Theoretical analysis of pump and compressor behavior is customarily based on poly-

tropic head and volumetric flow since a fixed relationship exists between the two. The
performance map under this head and flow coordinate is not affected by inlet condi-
tion changes, allowing the surge lines to be visually and mathematically described.
The head versus flow coordinate is thus called an invariant coordinate.
In practical applications, however, the head and flow are neither directly mea-
surable nor can they be readily calculated in real time. The relationship based on
alternative variables is thus needed as the basis for real-time applications. Reduced
head and reduced flow are popular among commercial applications because the coor-
dinate system based on the “reduced” variables preserves the invariant relationship
References 149
between the head and flow. These coordinate systems are called equivalent coordi-
nates.
Anti-surge control requires a timely and reliable indicator of the location of the
operating point relative to the surge point. The SRL defines the minimum stable
flow for all inlet conditions. A surge indicator based on this SRL offers a simple and
convenient real-time indication of the surge condition to support real-time decisions
for operation and control. One of the surge indicators, the anti-surge parameter (ASP),
is based on equivalent flow, which offers a good compromise between flexibility and
simplicity and is demonstrated with real-world examples.
References
API (2014) API standard 670 – machinery protection systems. Technical report, American
Petroleum Institute
Batson BW (1996) Invariant coordinate systems for compressor control. In: Proceedings of the inter-
national gas turbine and aeroengine congress and exhibition. The American Society of Mechanical
Engineers, Birmingham
Bloch HP (2006) A practical guide to compressor technology, 2nd edn. Wiley Interscience
Giampaob T (2010) Compressor handbook: principles and practice. The Fairmont Press
Lawson CL, Hanson RJ (1974) Solving least squares problem. Prentice Hall
forty-second turbomachinery symposium, Houston, Texas
Chapter 6
Basic Control Schemes
The overall control strategy presented in Chap. 4 explains the basic principle of con-
trolling the process and equipment. However, the process flow configuration varies
drastically across different applications, making it impossible to have a universally
applicable control design. Instead, the more practical and effective approach is to
fully understand the process dynamics and equipment characteristics, as discussed
in Chap. 2, apply the overall control strategy discussed in Chap. 4, and create fit-
for-purpose solutions for each specific application. The surge indicators discussed in
Chap. 5 provide the quantitative indications of the operating point in relation to the
operating envelope and facilitate the detailed control design. This chapter presents
several basic control designs that help understand the objectives and requirements.
These basic control schemes can serve as the basis for scaling up to more complex
process control solutions discussed in Chap. 7.
6.1 Centrifugal Pumps
Figure 6.1 shows a simple pumping process in a complex process setting. The cen-
trifugal pump P-101 transfers heavy crude oil from a processing tank T-101 to
the downstream operation for further processing. The operating objective is to main-
tain the material balance by pumping out as much fluid as it enters the processing
tank T-101 while maintaining the processing tank (static equipment) and the pump
(rotating equipment) in safe and efficient operation.
The overall control strategy is laid out in Chap. 4 as follows:
1. Capacity control. Capacity control is the normal regulatory control that keeps the
supply and demand balanced to maintain the desired level in the tank and head
and flow in the pump.
https://doi.org/10.1007/978-3-031-43122-7_6
152 6 Basic Control Schemes
Fig. 6.1 A typical process control solution with a centrifugal pump (reproduced, with permission,
from Niu and Xiao 2022)
Fig. 6.2 Performance curve for a centrifugal pump
2. Protective control. The minimum recycle flow control is a protective overriding

control against the pump surge. The pump flow must be maintained above the
minimum flow limit for stable operation.
The available control targets and control handles are illustrated in Fig. 6.1.
Figure 6.2 shows the manufacturer-provided performance curves.
6.1.1 Capacity Control
Capacity control aims to balance the supply and demand through the manipulation
of the pump flow, regulated by either throttling the control valve or adjusting the
pump speed.
6.1 Centrifugal Pumps 153
Fig. 6.3 A typical process control solution for a centrifugal pump with a fixed flow setpoint
Capacity Control with Fixed-Speed Pumps
Most centrifugal pumps operate at a fixed speed, with a typical capacity control
design as illustrated in Fig. 6.3.
The level controller LC-102 serves as the capacity controller. A decrease in inlet
flow FI-101 will cause the tank level LC102.PV to go down. The level controller
LC-102 “senses” the level change and adjusts the valve opening LCV-102 to reduce
the outlet flow FI-102. The reduced outlet flow causes the pump discharge pressure
PI-102 to increase, resulting in less pump flow FI-103. As a result, the tank
level LC-102.PV is brought back to its setpoint LC-102.SP and the supply-and-
demand balance is restored.
Conversely, when the inlet flow FI-101 increases, the tank level LC102.PV
will go up. The level controller LC-102 opens the control valve LCV-102 to let
out more flow, resulting in a lower discharge pressure PI-102 at the pump. The
lower pressure results in a higher pump flow FI-103 to bring the tank level back
on target.
The pump curves can intuitively explain this change in the pump pressure PI-102
and flow FI-103. For fixed-speed pump operation, the performance map in Fig. 6.2
is reduced to a single fixed-speed line at 100% pump speed (2,897 rpm), as shown
in Fig. 6.4.
The increase in pump pressure PI-102 (thus the pump head) causes the operating
point to move to the left along the constant-speed pump curve, resulting in less flow
FI-103 from F1 to F2.
The pump performance curve in Fig. 6.4 is rather flat, especially near the minimum
flow (MIN FLOW) point. A flat line indicates a large process gain between head
and flow, making the flow excessively sensitive to pressure changes, which can be
challenging for operation and control. An improved control design is with a cascade
control scheme shown in Fig. 6.5.
A flow controller FC-102 is added at the discharge line. The level controller
LC-102 and flow controller FC-102 work together to form a standard cascade
Fig. 6.4 Performance curve for a centrifugal fixed-speed pump
Fig. 6.5 A cascade control scheme for a centrifugal pump
control loop, resulting in better control performance from more reliable control of
the outlet flow FC-102.PV with the pump. See Niu and Xiao (2022), Smith (2010)
for the benefits of cascade control.
The control schemes in Figs. 6.3 and 6.5 are for supply-driven operations, where
the fluctuations in the inlet flow FI-101 are propagated through the tank level
LC-102 to the downstream operation FI-102.
The operation becomes demand driven if the process flow is dictated by down-
stream demand, i.e., FC-102. The capacity control design for the demand-driven
operation would change to Fig. 6.6. Higher demand from downstream operation
(FC-102) results in more flow (FI-103) from the pump, lowering the discharge
pressure PI-102. A lower discharge pressure (and lower head) results in a higher
pump flow FI-103 that causes the tank level LC-102 to decrease. The level con-
troller LC-101 detects the level change and opens the inlet valve to draw more inlet
flow FI-101 into the tank.
Fig. 6.6 Demand-driven capacity control for a fixed-speed pump
Fig. 6.7 Supply-driven capacity control with demand override
In practice, the supply-and-demand model can change unpredictably. For instance,

a supply-driven operation can turn into a demand-driven operation due to reduced
demand from downstream caused by a partial shutdown. A supply-driven control
typically requires a demand override to protect the supply-and-demand balance fully.
Similarly, a demand-driven control scheme should include a supply override.
Figure 6.7 shows a complete supply-driven control scheme with demand override,
where the overriding flow controller LC-101 is spawned off the same level trans-
mitter LT-101. As an overriding controller, LC-101 remains dormant (output at
100%) during normal supply-driven operation when the normal tank level is below
the controller setpoint LC101.SP. It becomes active if the tank level becomes exces-
sively high (higher than the controller setpoint) due to reduced demand flow. The
reverse-acting configuration will allow the controller to reduce the inlet flow to bring
the tank level back on target.
Fig. 6.9 Performance curve for a centrifugal variable-speed pump
Capacity Control with Variable-Speed Pumps
Although most pumps operate at a fixed speed, some large-capacity pumps operate
at variable speeds for energy efficiency. Figure 6.8 shows a typical capacity control
design for a variable-speed pump, which is identical to the design in Fig. 6.5, except
that the control handle is the pump speed instead of the control valve.
On the performance map, the change in pump flow rate is achieved via speed
variation, causing the operating point to move roughly along the constant-resistance
curve, as shown in Fig. 6.9. Changes in system resistance (e.g., back pressure) can
cause the operating point to move along the constant-speed curves. The combined
effect allows the operating point to move within a wide area of the performance map.
The complete capacity control scheme for a supply-driven variable-speed cen-

trifugal pump with demand override is given in Fig. 6.10. The demand override with
controller LC-101 propagates the supply-and-demand imbalance upstream, where
adequate swing control must exist to accommodate the change in demand.
As explained in Sect. 4.2, adequate swing streams and swing-control mechanisms
must exist in all the above scenarios to ensure uninterrupted supply-and-demand
propagation along the entire process flow path.
6.1.2 Minimum and Maximum Flow Control
The process operation requires that the process throughput (capacity) be able to vary
between zero and the desired maximum flow range. However, centrifugal pumps all
have a stable flow range that must not be exceeded. For instance, the pump in Fig. 6.4
can only operate stably between a minimum flow limit of 114 m3 /h and a maximum
flow limit of 315 m3 /h at 100% speed (2,897 rpm). This mismatch in the flow range
required and the flow range available must be addressed by protective controllers.
A minimum flow controller is required to ensure the pump flow is always above
the minimum flow limit. As shown in Fig. 6.11, the minimum flow controller
FC-103 is designed and configured as a reverse-acting PID controller with its set-
point FC103.SP equal to the minimum flow limit. During normal operation, the
pump flow FC103.PV is higher than the setpoint FC103.SP; the reverse-acting
PID controller FC-103 would drive the recycle control valve FCV-103 to the fully
closed position. That is, the minimum flow controller FC-103 remains dormant
with zero recycle flow.
If the pump flow FC103.PV falls below the minimum flow limit, e.g., due to
insufficient process flow from the tank inlet FI-101, the minimum recycle flow
controller FC-103 opens the recycle valve to allow a portion of the pump flow to
Fig. 6.11 Minimum recycle flow control for centrifugal pump
be recirculated back to the tank to make up the difference.1 The amount of recycle
flow is given by

0, If FC103.PV ≥ FC103.SP
Recycle Flow = (6.1)
FC103.PV − FC103.SP, If FC103.PV < FC103.SP.
There are many variations in process configuration corresponding to the choices in

fixed- versus variable-speed operation, supply-driven versus demand-driven control,
and fixed versus variable setpoint for minimum flow control.
Minimum Flow Control with Fixed-Speed Pump and Volumetric Flow
The minimum flow controller FC-103 in Fig. 6.11 is a standard PID controller
that takes the actual flow measurement as the process value (FC103.PV). The
desired flow rate is the controller setpoint (FC103.SP), and the controller output
(FC103.OP) determines the valve opening FCV-103, thus the outlet flow rate.
A characteristic of centrifugal machines is that the relationship between the fluid
head and volumetric flow does not change with the fluid property (e.g., density)
and the operating condition (pressure and temperature). Therefore, for this pump
operating at a 100% fixed speed, the minimum flow limit of 114 m3 /h remains the
same for all liquids and under all operating conditions. This minimum flow limit
can thus be directly taken as the flow setpoint (FC103.SP) for the minimum flow
controller FC-103. In other words, the process value (FC103.PV) for the flow
controller FC-103 is the actual pump flow measurement in m3 /h, and the controller
1Note that the recycle line takes off between the pump flow measurement and process flow mea-
surement locations, which implies that pump flow PI-103 is not always the same as the process
outlet flow FI-102.
setpoint (FC103.SP) has a constant value of 114 × (1 + 10%) = 125 m3 /h, which
includes a 10% control margin.
If the measured pump flow rate FC103.PV falls below 125 m3 /h, the recycle
valve would be opened to maintain the flow at 125 m3 /h.
Minimum Flow Control with Fixed-Speed Pump and Mass Flow
Although the pump curve is provided as head versus volumetric flow, as shown in
Fig. 6.2, the actual online measurement may be the mass flow rate instead of the
volumetric flow rate.2 If the fluid density is assumed to be constant, the mass flow
limit can be calculated from the minimum volumetric flow and treated as a constant
value.
In this example, assume that the oil density is 960 kg/m3 . With a 10% safety
margin, the minimum mass flow limit is calculated as3 :
Fm = ρ Fv = (960 ÷ 3600) × 114 × (1 + 10%) = 33.4 kg/s. (6.2)

(kg/m3 ) (s/h) (m3 /h)
The control design in Fig. 6.11 remains unchanged, except that the engineering
units of FC103.PV and FC103.SP are now all in mass flow (kg/s), and the flow
setpoint value is set to 33.4 kg/s instead of 125 m3 /h.
Minimum Flow Control with Variable-Speed Pump and Variable

Setpoint
Now, let us extend the minimum flow control to variable-speed operation where the
control handle is the pump speed instead of the discharge control valve.
With the fixed-speed operation, the operating point can only move along a sin-
gle fixed-speed performance curve (Fig. 6.4). In contrast, with the variable-speed
operation, the operating point can move across multiple speed lines, as shown in
Fig. 6.9. The minimum flow limit has a different value at different pump speeds,
approximately proportional to the machine speed (according to the affinity laws).
For example, as the pump speed varies from 100% speed (2897 rpm) to 50%
speed (1450 rpm), the minimum flow limit decreases from 114 m3 /h to 57 m3 /h.
The maximum flow limit changes from 315 m3 /h to 158 m3 /h. If fixed flow setpoints
2 It is also possible that the pump performance curve is provided in mass flow, but the actual flow
measurement is in volumetric flow rate.
3 The flow conversion from m3 /h to kg/s also needs to multiply by 3600 s/h to be correct in engi-
neering units.
Fig. 6.12 Minimum and maximum flow limits for a centrifugal pump
are used for the minimum and maximum flow control, the feasible operating region
would be limited to the shaded area in Fig. 6.12, between the two vertical lines ➀
and ➁, while the feasible operating region by design is between lines ➂ and ➃. This
reduced feasible area does not even include the best operating point (BEP) and is
thus clearly unacceptable.
The flow controller FC-103 can use a variable setpoint to accommodate the
variable minimum flow limits. The minimum flow limits at different pump speeds are
given from the pump performance curves in Fig. 6.12. As the pump speed changes, the
setpoint of the minimum flow controller FC-103 is re-calculated and automatically
adjusted. For example, if the flow measurement is provided as mass flow Fm in kg/s,
and assuming that the actual fluid density ρ is the same as that in the datasheet, the
flow can be calculated from the pressure rise as4

Fm = C Pd − Ps . (6.3)
The coefficient C can be calculated from the head and flow data with a selected
minimum flow point on the performance curve, e.g., h = 1, 220 m and Fv = 114
m3 /h, after appropriate unit conversion, as follows:
4 If the flow measurement is in volumetric flow (m3 /h), the setpoint calculation must be changed
to volumetric flow, which is straightforward. Obviously, from the pump curve, the flow setpoint
can be calculated from the pump speed. Based on Affinity Laws, the pump is a linear function of
the machine speed, and the calculation is trivial. However, speed measurement is not as fast and
reliable as pressure and is not preferred.
Table 6.1 Minimum flow calculation for pump
Pd − Ps = ρ g h
= 960 × 9.8 × 1, 220
(kg/m3 ) (m/s2 ) (m)
= 11, 477, 760 = 11, 478 (kPa) (6.4)

(Pascal)
Fm = ρ Fv
= (114 ÷ 3600) × 960
(m3 /h) (s/h) (kg/m3 )
= 30.4 (kg/s) (6.5)

Fm
C=√
Pd − Ps
30.4
=√ = 0.284. (6.6)
11478
The flow setpoint can then be calculated from the pressure rise as

Fm = 0.284 Pd − Ps . (6.7)
In practical applications, the minimum flow values on the performance map do

not fall precisely on the same line due to errors and approximations. The most
conservative calculation among all scenarios should be used. For this purpose, we
start with digitizing the performance curve and collect all the minimum flow values
for all speeds. The results are listed in Table 6.1. The pressure rise Pd − Ps and the
mass flow Fm are calculated with Eqs. 6.4 and 6.5 for every pump speed at the surge
and choke points. The coefficient C is then calculated for each minimum flow point,
and the results are listed at the bottom of the table.
The most conservative value is the largest at all given speeds, C = 0.286 at 2,610
or 1,450 rpm. The calculation formula of the surge reference line for this pump is
thus given as

Fm = 0.286 Pd − Ps . (6.8)
Fig. 6.13 A typical process control solution with a centrifugal pump with variable flow setpoint
It is customary to have a safety margin for control, typically 10% extra flow, added
to the minimum flow limit:

Minimum Flow Limit: Fm = 0.286 Pd − Ps × (1 + 10%)

= 0.315 Pd − Ps . (6.9)
The modified control design will include pressure measurements on the suction
and discharge side, as shown in Fig. 6.13. An online calculation block FY-103
is added to compute the minimum flow limit from the suction Ps and discharge
pressures Pd . The result is sent down to the flow controller FC-103 as its remote
setpoint (FC103.RSP).5 The process value (FC103.PV) for the minimum flow
controller FC-103 is the actual flow measurement in kg/s.
Minimum Flow Control with Variable-Speed Pump and Fixed Setpoint
In practice, operations often prefer a fixed setpoint for the minimum flow controller
FC-103 instead of a variable setpoint at different speeds. For this purpose, a surge
indicator, the anti-surge parameter (ASP) in Eq. 5.33, can be used as the new process
variable (FC103.PV) to indicate the distance between the current operating point
and the minimum flow limit. The anti-surge parameter is defined from Eq. 6.8 based
on Eq. 5.33:
5 The PID controller FC-103 needs to be in RCAS mode to receive remote setpoint (FC103.RSP)
from external calculation block FY-103; see Niu and Xiao (2022) for PID controller modes.
Table 6.2 Minimum flow calculation for pump, with anti-surge parameters
Fig. 6.14 Minimum flow controller with a fixed ASP setpoint
Fm
ASP = C √ × 30%
Pd − Ps
Fm
= √ × 30%
0.286 Pd − Ps
Fm
= 1.049 √ . (6.10)
Pd − Ps
The ASP values for all the minimum flow limit points are calculated with Eq. 6.10
using the minimum flow data in Table 6.1 and appended to the end of the table.
The minimum flow line has a reference value of 30% by definition. All the ASP
values for the minimum flow points are 30% or smaller, indicating that the ASP
calculation is based on the worst-case scenario, as shown in Table 6.2. In other
words, any operating point with an ASP value greater than 30% is operating above the
minimum flow limit. Therefore, a PID controller XC-103 with the calculated ASP
value as the XC103.PV and a fixed setpoint (XC103.SP) of 30 × (1 + 10%) =
33% will achieve the same control result as with a variable setpoint in Eq. 6.9. The
control design is illustrated in Fig. 6.14.
The ASP value is calculated with XY-103 from the suction pressure PI-101,
discharge pressure PI-102, and pump flow FI-103. As a protective overriding
controller, if the process value XC103.PV is higher than the setpoint XC103.SP,
the reverse-acting controller will continue closing the valve until it is fully closed.
Therefore, the recycle valve remains fully closed during normal operation when
the process flow exceeds the minimum flow limit. Only when XC103.PV is below
XC103.SP will the controller open the recycle valve.
End-of-the-Curve Control with Fixed Setpoint
If the system resistance at the downstream operation becomes very low, the operating
point may move to the far right of the performance curve. Operating over the curve
leads to choked flow and reduced efficiency.
An anti-choke controller XC-102 can be added to prevent the operating point
from going too far to the right. Anti-choke control is also called the end-of-curve
control or maximum flow control. The same ASP value calculated by XY-103 is
fed to the anti-choke controller XC-102 as its process value XC102.PV. From
Table 6.2, the ASP setpoint (XC102.SP) for the anti-choke controller can be set to
92.7% × (1 − 10%) = 83.4%, based on the smallest (thus most conservative) values
of all the maximum flow points. The protective control design with both anti-surge
and anti-choke controls is shown in Fig. 6.15, with an operating range between
ASP=33% and 83%.
End-of-curve control can be readily added with fixed-speed pump control since
a discharge valve is installed. However, it is rarely implemented for variable-speed
pump control because of the need for discharge throttling valve. √
An orifice-based flowmeter has a turndown ratio of 10:1 in DP, or 3.16:1 ( 10 =
3.16) in terms of flow. From Table 6.2, the ratio between maximum flow and minimum
flow is about 3.1, which is no surprise since the diffuser of a centrifugal machine
behaves like a flow restriction component similar to an orifice plate.
In summary, the minimum flow control design can be based on a fixed flow
setpoint for a fixed-speed pump or a variable flow setpoint for a variable-speed
pump. However, the design with an anti-surge parameter is more generic. It operates
with a fixed setpoint and is thus preferred. It is also consistent with anti-surge control
for centrifugal compressors, as we move on to compressor control in the next section.
6.1.3 A Complete Control Design
A complete control design, including capacity control and minimum flow control, is
depicted in Fig. 6.16. The capacity control is through variable pump speed, and the
minimum flow control is through the recycle valve.
6.2 Centrifugal Compressors 165
Fig. 6.15 Minimum flow control and end-of-curve control for a centrifugal pump
Fig. 6.16 Capacity control and end-of-curve control
If end-of-curve control is also desired, the design in Fig. 6.16 becomes Fig. 6.17
and becomes fairly complex. However, with the basic designs discussed above, and
the incremental increase in complexity, the design is still relatively straightforward
to understand.
6.2 Centrifugal Compressors
Centrifugal compressors work by the same principle as centrifugal pumps. However,

the control design is considerably more complex due to the variable density of the
compressible gases that a compressor handles. Figure 6.18 shows a generic process
configuration with a single-stage centrifugal compressor and the potential process
measurements and control handles. This section discusses this simple compression
Fig. 6.17 A complete pump control solution
Fig. 6.18 A simple compression process with a single-stage compressor
process’s basic capacity control and anti-surge control design. The discussion of com-
plex compression processes involving multiple compressors is deferred to Chap. 7.
6.2.1 Capacity Control
Capacity control follows the overall control strategy discussed in Chap. 4. It is typi-
cally achieved by throttling the control valves if the machine is configured as fixed
speed or manipulating the machine speed if a variable-speed gearbox or variable-
frequency drive (VFD) motor is installed.
A critical decision is between supply-driven and demand-driven control. The suc-
tion header pressure is typically controlled for supply-driven operation, with the
supply/demand imbalance propagated from suction to discharge and from upstream
to downstream. On the other hand, in demand-driven operations, the discharge
Fig. 6.19 Supply-driven capacity control with throttling valve
Fig. 6.20 Supply-driven capacity control with speed
pressure is used for capacity control, and the material imbalance is propagated back-
ward from downstream to upstream.
A supply-driven capacity control with fixed-speed operation is illustrated in
Fig. 6.19. The dashed (blue) line indicates the direction of propagation of the sup-
ply/demand imbalance, with the suction header pressure P1 and discharge pressure
P2 being the relay points. The suction pressure controller PC-101 serves as the
capacity controller and manipulates the suction flow control valve PCV-101 to
adjust the process flow, which forces the compressor flow to change. The change
in compressor flow leads to changes in discharge pressure. The discharge header
pressure controller PC-102 detects the discharge pressure change and adjusts the
flow to the downstream operation.
If the compressor operates at variable speed, the supply-driven capacity control
will be through the same suction pressure controller PC-101 but via adjustment of
the compressor speed, as shown in Fig. 6.20.
If the capacity, or process flow, is dictated by downstream demand, the operation
becomes demand driven, as shown in Fig. 6.21. Demand change is propagated back-
ward, from downstream to upstream, via the discharge pressure controller PC-102
and the machine speed to the suction pressure controller PC-101.
Fig. 6.21 Demand-driven capacity control with speed
Fig. 6.22 Supply-driven capacity control with recycle
Capacity control can also be achieved with recycling, as illustrated in Fig. 6.22.
Recycling is the easiest but also the most inefficient way of capacity control. It is
rarely used as the primary capacity control for centrifugal compressors.
Similar to the pump controls discussed in Sect. 6.1, the supply-and-demand model
may reverse due to operational needs or plant upsets. For example, the supply-driven
operation in Fig. 6.20 changes to demand driven if the downstream consumption is
suddenly limited due to, e.g., a partial shutdown. Protective overriding control may
become necessary for reliable operation to handle the reversal of the supply-and-
demand model. A supply-driven operation with demand override is achieved through
suction pressure control with discharge pressure override, as shown in Fig. 6.23. The
dashed (blue) line designates the propagation path of the supply-driven control. In
contrast, the dotted (red) line indicates the propagation path if the supply-driven
operation becomes demand driven.6
6The capacity control becomes compressor driven if the throughput is limited due to the compres-
sor hitting the maximum speed or motor power limit. In this case, the operation upstream of the
compressor becomes demand driven, while the operation downstream turns supply driven. Conse-
quently, the critical requirement is that both upstream and downstream operations must have the
means to adjust the supply and demand to align with compressor capacity (swing control).
Fig. 6.23 Supply-driven capacity control with demand override
Since capacity control is concerned with the supply-and-demand relationship

along the entire process flow path, including the compressor, the design must be
based on a holistic view of the supply-and-demand model. The basic control design
discussed here provides the foundation for addressing more complex control needs.
For more in-depth discussions on capacity control, see Niu and Xiao (2022).
6.2.2 Anti-surge Control
An anti-surge controller is a standard protective controller that monitors and regulates

the compressor flow in real time to prevent the compressor from going into a surge,
similar to a minimum flow controller for a centrifugal pump. The control scheme
is typically a standard PID control loop, and is straightforward. The challenge is
calculating the anti-surge parameters (ASP), which is the process value (PV) of the
anti-surge controller that indicates the location of the operating point in relation to
the minimum flow limit.
Figure 6.24 shows a typical anti-surge control scheme for a single-stage centrifugal
compressor.
The anti-surge controller UC-111 is a standard PID controller, with the process
variable (UC111.PV) being the anti-surge parameter (ASP), calculated with the
function block UY-111 from suction pressure PI-111, discharge pressure PI-112,
and the suction flow FI-111. The measurement range of the ASP is from 0 to 100%,
with the surge reference line (SRL) at 30% and the stonewall at approximately 100%
(see Sect. 5.3).
The anti-surge control setpoint UC111.SP has a fixed value of 33%, which is
10% to the right of the surge reference line (30%×(1+10%)=33%). The controller
output (UC111.OP) is sent to the anti-surge control valve UCV-111. When the ASP
stays above the setpoint value of 33%, which indicates that the compressor flow is
above the minimum flow limit, the reverse-acting anti-surge controller UC-111 will
keep the recycle valve fully closed. If the ASP valve falls below 33%, e.g., due to
Fig. 6.24 Anti-surge control scheme, with flow measurement on suction
low process flow, the anti-surge controller UC-111 will open the recycle valve by a
calculated amount to keep the compressor flow above the minimum flow limit.
The orifice-based flowmeter FI-111 is typically located on the suction side, as
highlighted in Fig. 6.24, which is the preferred design for energy efficiency and cal-
culation simplicity. However, the orifice plate inevitably causes permanent pressure
losses. This pressure loss at the suction side may be unacceptable for certain low-
pressure operations. The alternative design is to place the flowmeter at the discharge
side where the pressure has been elevated, as highlighted in Fig. 6.25. The control
design is identical to that with the flowmeter on the suction, except that the ASP
calculation is slightly different, involving further approximations.
For compressors where simplicity outweighs efficiency, the control design can
be simplified to using two differential pressure measurements, one across the orifice
for flow measurement and another across the compressor as pressure rise, as shown
by Eq. 5.31. This primitive control scheme is shown in Fig. 6.26, which is almost
identical to Fig. 6.24, except that the ASP calculation in UY-111 is based on two
differential pressures PDI-111 and PDI-112 with a different calculation formula
(see Eq. 5.31). It was used in many early compressor control design (Boyce et al.
1983).
Fig. 6.25 Anti-surge control scheme, with flowmeter at discharge
Fig. 6.26 Anti-surge control scheme, with differential pressure DP for flow
6.2.3 Anti-surge Parameter Calculation
The primary challenge for anti-surge control is to know how far the current operating
point is from the surge limit. Like pump control, the surge indicator is the anti-surge
parameter (ASP), calculated from online pressure and flow measurements.
The recommended anti-surge parameter (ASP) is given by Eq. 5.28 and is repro-
duced here:
Fe
ASP = 2 3 , (6.11)
Pd Pd Pd
C0 + C1 + C2 + C3 + ···
Ps Ps Ps
where Fe is the equivalent flow through the flow-measuring orifice (FT-111).

Ps and Pd are the absolute suction pressure (PI-111) and discharge pressure
(PI-112), respectively. The number of terms inside the square root in the denomina-
tor (C0 , C1 , C2 , · · · ) depends on the complexity of the surge line. It will be determined
based on compressor performance data and the flow transmitter data.
The equivalent flow Fe of the compressor is derived from the DP-based flow
measurement and is given as

Ps Ps,r
Fe = Fm,r . (6.12)
Pr Ps
As an equivalent flow, Fe has the same measurement range and engineering unit
as the reference mass flow Fm,r that it is derived from.
The primary challenge to anti-surge control is the availability of the performance
data required by Eqs. 6.11 and 6.12 for anti-surge parameter calculation. A green-
field project starts from scratch. Therefore, the data’s availability and integrity are
usually not a concern since the data suppliers, such as manufacturers and contractors,
are all readily accessible. For a brown-field project, however, it is usually the contrary.
The availability and integrity of the data are the biggest challenge.
For a green-field project, before the plant is built and operational, the data for
design are primarily based on best engineering knowledge and offline calculations
dictated by the basis of design. They may include the predicted gas property, desired
operating pressure and temperature, and required capacity (mass flow rate). The
different operating scenarios the machine is expected to operate, such as winter,
summer, heavier gas, lighter gas, and turndown conditions, are typically provided.
Table 6.3 lists the typical data types and their engineering units for describing the
characteristics. These variables are illustrated on the control schematic in Fig. 6.27.
It is critical to have a coherent view of the different types of data and their roles in
control design.
Table 6.3 Typical process data for compressor control

Measurements Suction Side Discharge Side Base Units Other Units Imperial
Pressure Ps Pd kPa(a) bar(a) psi(a)
Temperature Ts Td K ◦C ◦F
Compressibility Zs Zd - - -
Adiabatic Index κs κd - - -
Volumetric Flow Fv,s Fv,d m3 /hr m3 /s gpm
Differential Pressure P s P d kPa mbar psi
Density ρs ρd kg/m3 lb/ft3
Mass Flow Rate Fm kg/s lb/s
Speed N % rpm
Molecular Weight M kg/kmol lb/kmol
Polytropic Head Hp kJ/kg
Polytropic Efficiency η % %
Fig. 6.27 Control target and handles for centrifugal compressor control
The data requirements can be classified into five categories, as listed in Table 6.4,
corresponding to the five essential components in a standard feedback control loop in
Fig. 6.28. For example, the compressor API datasheet specifies the desired operating
points, serving as the basis of design or control objectives. The flowmeter datasheet
defines the requirements for flow measurement, and the valve datasheet provides the
characteristics of the control valve.
The amount of data required can be overwhelming to the unprepared. There is
no standard or even consensus on what data must be provided and in what format.
Therefore, the end user must define their data requirements at the earliest opportunity.
Of the various data suppliers, the compressor manufacturer is the crucial one. The
process control engineer may need to insist on the preferred data type and format the
equipment manufacturers must provide for control design. For example:
Table 6.4 Data requirements for anti-surge control design

Role in Control Data Type Preferred Format Primary Supplier
➀ Control Objective API Design Datasheet Spreadsheet+Table Process Engineer
➁ Equipment Dynamics Performance Curves Spreadsheet+Chart Manufacturer
➂ Flow/Pressure Meters Flow Meter Datasheet Tabular Manufacturer
➃ Recycle Valves Valve Datasheet Tabular Manufacturer
➄ Control Scheme ASP formula + Setpoints Equation Process Control
Fig. 6.28 Essential components of a control loop
1. Curves versus spreadsheet. Traditionally, the performance data are provided as

“curves.” These curves must be “read out” as sampled values before use, which is
tedious and error prone. For modern applications, the equipment manufacturers
should provide all relevant data in digital format (e.g., spreadsheet) to avoid this
digitization process.
2. Head versus pressure. The equipment manufacturer typically supplies the perfor-
mance data as head/efficiency versus flow or pressure/temperature versus flow.
Since the invariant coordinates for control design (see Chap. 5) are based on pres-
sure ratio, the latter is always preferred. However, having the head and efficiency
data in addition to pressure and temperature is very beneficial for cross-validation
purposes.
Example 6.1 ASP calculation with software tools. The ASP calculation is usu-
ally facilitated by dedicated software tools, except for simple ones involving lim-
ited performance data. However, software tools are not a replacement for technical
know-how. A solid understanding of process control technologies and compressor
characteristics is required to use the tools correctly.
Here we demonstrate the concept and procedure of ASP calculation with a reduced
set of actual data. The companion CPACS software tool7 can be used to enter and
organize the compressor data, visualize the performance map, and assist the control
design.
An example of the compressor data is shown in Fig. 6.29. The first row of the
spreadsheet shows the type of data that can be entered, along with the engineering
units in the second row. Some data may be redundant, but having as much data as
possible is always beneficial for cross validation.
The compressor data may be provided with different engineering units. The unit
conversion is usually tedious and error prone. Most software tools should have the
unit conversion automatically taken care of internally.
The ASP coefficients for the desired ASP formula are calculated with the chosen
coordinate system and the selected operating scenarios. The ASP formula using
pressure ratio versus equivalent flow and with three coefficients is shown in Fig. 6.30,
along with the surge reference line (SRL) and surge control line (SCL):
Fe
ASP = 2 . (6.13)
Pd Pd
840.81 − 862.64 + 473.66
Ps Ps
The measurement range of the ASP is 0 ∼ 100%, with the SRL at 30% and the
stonewall roughly at 100%.
Appendix A provides more details on using software tools to facilitate data man-
agement, surge analysis, and control design.
6.2.4 † Flowmeter Design
Capacity control and anti-surge control rely on fast and reliable flow measurements.
However, measuring the actual flow inside the machine is not practical. Instead, the
flow is usually measured at the suction or discharge line using a DP-based flowmeter
such as an orifice or Venturi tube. The flowmeter design must align with control
requirements, including the flow measurement range, maximum pressure drop, and
proper compensation for pressure and temperature.
The flowmeter datasheet provides the connection between the differential pressure
P (measured) and the inferred flow. The process control engineers should be capable
of validating flowmeter designs for process control needs.
7 CPACS stands for Compressor Performance Analysis and Control Support. The CPACS tool is
an Excel add-in and is freely available at http://github.com/niucontrol/CPACS. Download the user

manual for a full description of the functionalities.
Fig. 6.29 The compressor API design data
Fig. 6.30 Performance map for a compressor
With a DP-based flowmeter, such as a restrictive orifice in Fig. 3.13a, the actual
measurement is the pressure drop P across the orifice. The flow through the orifice
(Fv or Fm ) is then inferred from the pressure drop P with Bernoulli’s equation.8
Between any two points ➀ and ➁ on the flow line, we have
1 1
P1 + ρ1 v12 = P2 + ρ2 v22 , (6.14)
2 2
8 Head, flow, and ASP are all inferred from other online measurements and, thus, they are all called
inferential properties.
where P1 and P2 are the pressure measurements at the two points, respectively.
ρ1 and ρ2 are the densities. v1 and v2 are the flow velocities.
Assume the change in fluid density across the orifice is negligible (ρ1 = ρ2 ). By
continuity of mass, A1 v1 = Av2 = Fv , where A1 and A2 are the cross-sectional areas
of the orifice and the vena contracta, and Fv is the volumetric flow. We then have:
1 2
P1 − P2 = ρ v2 − v12
2
2
1 Fv Fv2
= ρ − 2
2 A22 A1

1 1 1
= ρ Fv
2
− 2 . (6.15)
2 A22 A1
Define P = P1 − P2 , we have

A2 P
Fv = 2
1 − (A2 /A1 ) 2 ρ

π d2 2 P
=
4 1 − β4 ρ

P
=C , (6.16)
ρ
where d is the diameter of the orifice and D is the diameter of pipe, and with:
2
1 1 A2 d d
A1 = π D2, A2 = π d 2, = = β2, β =
4 4 A1 D D
C is the flow coefficient.

The more generic equation for mass flow measurement is given by ISO5167
(ISO-5167 2003) standard9 as

Cd π P P
Fv = ε d2 2 =C (6.17)
1−β 4 4 ρ ρ
Cd π
Fm = ε d 2 2 P ρ = C ρ P, (6.18)
1 − β4 4
where Cd is the discharge coefficient and ε is the expansion factor. They are exper-
imentally determined coefficients that account for friction losses and other discrep-
ancies between theory and practice.
9 Note that ISO 5167 (all parts) applies only to flow that remains subsonic throughout the measuring
section and where the fluid can be considered single phase. It does not apply to the measurement
of pulsating flow.
The critical information is the flow coefficient C, a relative measure of the effi-
ciency in allowing fluid to pass through. The flow coefficient connects the pressure
drop P across an orifice valve with the corresponding flow rate Fm or Fv .
One of the crucial design parameters of the flowmeter for compressor control is
the measurement range. The design requirement is to ensure that the flowmeter’s
effective measurement covers compressor’s flow range, required by both capacity
control and anti-surge control. A DP-based flowmeter has an inherent turndown
limit, below which the measurement is no more trustworthy. The turndown ratio is
typically 10:1 by DP or 3:1 by flow,10 see Fig. 3.13b. In other words, the effective
flow measurement range is approximately 30% to 100% of the maximum range.
As shown in Sect. 3.3.2, compressor flow has an approximate turndown ratio of
about 3:1 as well. If the stonewall is defined as 100%, the surge line is approximately
30%. The effective range of the flowmeter roughly matches the compressor range.
However, if the two ranges do not match, engineering knowledge is required to strike
a trade-off.
Example 6.2 Flowmeter data and flowmeter sizing. The flowmeter datasheet
describes the relationship between the flow rate and pressure drop under the design
or calibration condition. An example of the orifice flowmeter datasheet is given in
Fig. 6.31.
The flowmeter’s maximum and turndown flow rate can be visualized on the same
performance map as shown in Fig. 6.32. The measurement range provided by the
flowmeter and the measurement range required by the process control design can be
visually compared and intuitively verified from the map.
For validation purposes, the maximum flow is achieved at the maximum pressure
drop of P = 25, 000 Pa (250 mbar), with the given orifice size of 0.240 m (240
mm):
Cd π 2
Fm = ε d 2 P ρ
1 − β4 4
0.59789 3.14159 √
=√ × 0.99744 × × 0.2402 × 2 × 25000 × 46.26
1− 0.713174 4
= 47.65 kg/s. (6.19)
The sizing of the equipment and devices is primarily the responsibility of the
process, equipment, and instrumentation personnel. However, it is crucial for process
control to cross validate the sizing calculation and ensure it is acceptable for both
capacity control and anti-surge control.
The pressure drop is expected to be sufficiently large to achieve the required
measurement accuracy for process control, which demands a smaller orifice opening.
However, operations also have other requirements, such as a limit on the maximum
pressure loss at the flowmeter. Therefore, the design must consider many constraints,
√
10 Flow is proportional to the square root of P; thus, the flow turndown is 10:1=3.16:1.
Fig. 6.31 Orifice-based flowmeter datasheet
Fig. 6.32 Compressor curves with maximum and minimum flow lines
sometimes conflicting, to finalize. Many rounds of interactions between multiple

stakeholders are usually necessary.
The flowmeter design is to determine the orifice size based on the maximum flow
rate required by process design within the limit of maximum allowable pressure
loss. For the orifice meter in Fig. 6.31, with the requirement of maximum flow
Fm = 47.75537 kg/s at DP P = 25, 000 Pa (250 mbar), the orifice diameter can
be calculated from Eq. 6.18 as
Fm2
d= 4 (6.20)
Fm2 1
+ (π ε Cd )2 ρP
D4 8
47.755372
= 4 2
47.75537 1
4
+ (3.14159 × 0.59789 × 0.99744)2 × 46.26 × 25, 000
0.3366 8
= 0.2399 m
= 239.9 mm (6.21)
which agrees with the 240 mm specification in the datasheet.

If we want to achieve a 60.0 kg/s flow at P = 25 kPa, the orifice size would
need to increase to
60.02
d= 4
60.02 1
+ (3.14159 × 0.59789 × 0.99744)2 × 46.26 × 25, 000
0.33664 8
= 260.0 mm. (6.22)
6.2.5 Instrumented Safeguarding
If the anti-surge control fails to stop the pump or compressor from entering a surge,
then the anti-surge trip logic must detect the surge condition and proactively trip the
compressor to avoid damage. This machine safeguarding function provides the third
line of defense against surge, as shown in Fig. 4.1.
Tripping a compressor introduces its own risks to the machine (Botros et al. 2015).
The compressor should trip only when deemed absolutely necessary. For instance, a
single short surge is not a significant risk, but multiple surges within a short period
are. It is critical to strike a reasonable balance between keeping the operation online
and protecting the machine from damage.
The trip logic is typically based on either a long surge cycle (e.g., longer than
3 s) or repeated short surge cycles within a short period. See Elliott and Bloch
Fig. 6.33 Anti-surge trip conditions
(2021). For example, API 670 recommends the criterion of 3 surges within 10 s (API
2014).
• Deep surge: A surge cycle lasting longer than, e.g., 3 s.
• Repeated surge: Three repeated surge cycles within, e.g., 10 s.11
The trip conditions are summarized and illustrated in Fig. 6.33.
The anti-surge trip of the compressor is in addition to other normal SIS trip
protections, such as those on pressure, temperature, flow, or vibration.12
Anti-surge trip logic is based on the same anti-surge formula (ASP) value calcu-
lated in Eq. 6.11, with the same range from 0 to 100%, corresponding to zero and the
choked flow. With the surge reference line at 30%, the anti-surge trip line is typically
set to half the surge reference value, i.e., 15%. If the ASP value falls below this limit,
the anti-surge trip logic generates a trip request to the existing safety instrument
system (SIS), which processes the request following the standard protocols. One of
the actions is to de-energize the anti-surge control valve, which is always fail-open,
to cause it to open at full speed.
In a typical implementation, all control functions are implemented in control
systems such as standard DCS. The anti-surge trip logic is implemented in the safety
instrument system (SIS), independent of the anti-surge control in DCS, to avoid
common mode failures.
One generic implementation of the anti-surge trip logic based on three surges is
shown in Fig. 6.34. Tripping the compressor upon detecting the first or second surge
is also common, which is a straightforward simplification of the trip logic.
11 If two surges occur within 3 s, they are counted as one surge cycle. A surge may produce noisy
transmitter signals and the possibility of multiple surge counts from a single surge cycle.
12 The surge trip signal can be discerned from the first-out event in the sequence of event recording.
Fig. 6.34 Segregated safeguarding and control in implementation
The convention used in the SIS logic is that a signal of ONE (=1) indicates a healthy
condition, while a signal of ZERO (=0) indicates a trip condition. The different
function blocks are explained as follows:
1. Surge cycle. The anti-surge parameter (ASP) is calculated in the SIS independent
of the ASP calculated in DCS. The ASP value is compared with the surge trip
limit (SP, typically set to 15%). If the ASP value is higher than SP, a ONE signal,
indicating healthy condition, is produced by the “>” gate as the input to the first
OR gate. This ONE input locks the OR gate output to ONE (healthy).
If the ASP drops below SP, a ZERO signal, indicating surge condition, is produced
by the “>” gate and propagates to the first OR gate.
The OR gate output depends on another input from the startup delay block. If
both inputs are ZERO, the OR gate will produce a ZERO value, indicating a surge
condition to trigger the surge counting.
2. Block (D): Startup override. When the motor is started, the compressor needs
some time to reach the minimum stable condition. Since the ASP value may start
from zero during startup, the trip signal must be masked until the ASP value
reaches its normal stable value to avoid nuisance alarms and trips.
Block (D) is a DELAY ON block and provides a T0-second delay on the propa-
gation from its input to its output. That is, when the input of Block (D) (i.e., the
motor running status) changes from ZERO to ONE (ON LOAD), its output will
remain at ZERO for T0 seconds before changing to ONE. The ONE signal passes
through the negation gate and produces a ZERO signal as input to the first OR
gate.
3. Block (A): Start of surge counting. A surge longer than T1 seconds is counted
as one surge cycle. Block (A) is a DELAY OFF block. When its input changes
from ONE to ZERO, it starts a T1-second countdown. During the T1 seconds, the
DELAY OFF block blocks the input signal, and the output signal remains ONE.
When a surge occurs, the output of the first OR gate changes from ONE to ZERO,
and the DELAY OFF block (A) blocks the signal for T1 seconds. At the end of
the T1 seconds, if the OR gate output is still ZERO (i.e., surge condition still on),
the ZERO signal will pass through the final AND gate and generates an SIS trip
signal.
4. Block (B) and Block (C): Surge counting toward the three-surge cycles. Block
(B) is a DELAY ON block. When its input changes from ONE to ZERO, and its
output will immediately flip from ONE to ZERO and hold the ZERO output for T2
seconds. Block (C) is a DELAY OFF block, identical in functionality to Block
(A), but a different delay time can be assigned if required. Upon receiving the
first surge signal (a dip in ASP value), block (B) output changes to ZERO and
holds the ZERO for T2 seconds. The output of Block (C) changes to ZERO after
T1 seconds and thus opens the second OR gate to receive the second surge dip.
The T1-second delay imposed by Block (C) ensures that if two surges occur in
less than T1 seconds, it will be treated only as one. If the second ASP dip comes
in within T2 seconds of the first surge dip, the second Block (B)/Block(C) will
start counting. The Block (B) output will remain at ZERO for T2 seconds, while
the Block (C) output will change to ZERO after T1 seconds, and the last OR gate
is ready to take the third surge dip. If a third surge comes in before the first T2
seconds countdown expires, the surge dip (ZERO) will pass through the first, third,
and fourth OR gate and produces a trip signal.
If there are no more surges within the T2 seconds count, the Block (B) countdown
expires, and its output changes from ZERO to ONE and blocks the corresponding
OR gate. If any surge dips are longer than T1 seconds, a trip signal will be produced
immediately since the first trip condition is met.
6.2.6 A Complete Control Design
A complete control scheme, including both capacity control and anti-surge control, is
shown in Fig. 6.35 for fixed-speed supply-driven operation and Fig. 6.36 for variable-
speed supply-driven operation. The capacity is controlled with the suction pressure
controller PC-101 by manipulating either the suction valve PCV-101 or machine
speed SC-101. At the discharge side, pressure controller PC-102 is the relay point
to propagate the supply-and-demand fluctuations to downstream operation. Adequate
swing capacity and control mechanism must exist to absorb the fluctuations in the
gas flow from the compressor.
The discharge pressure controller PC-103 is an overriding controller to prevent
the discharge pressure from going excessively high. Similarly, pressure controller
PC-101B is provided to protect the suction header if the capacity controller PC-101
fails to maintain the suction pressure. The controller UC-111 provides the anti-surge
control function through the anti-surge control valves UCV-111.
Fig. 6.35 Basic capacity and anti-surge control with fixed-speed drive
Fig. 6.36 Basic capacity and anti-surge control with variable-speed drive
The control designs shown in Figs. 6.35 and 6.36 are acceptable but with inher-
ent flaws. One of the flaws is that the interaction between the two manipulated
variables (suction value and recycle valve in Fig. 6.35; machine speed and recycle
valve in Fig. 6.36) and the two controlled variables (capacity and surge margin) is
not adequately addressed, which may result in inefficient operation or sub-optimal
performance. Proper decoupling with a multivariable control scheme is preferred,
which is the topic of advanced control design in Chap. 7.
6.3 Reciprocating Pumps and Compressors 185
6.3 Reciprocating Pumps and Compressors
Reciprocating pumps and compressors work on different principles than centrifugal

machines. One significant difference is that reciprocating machines do not suffer
from the surge and choke phenomena. As a result, the control solution for processes
with reciprocating machines is much more straightforward.
6.3.1 Capacity Control for Reciprocating Machines
The capacity control of reciprocating machines is similar to that of centrifugal

machines in concept. The system resistance does not impact the pump flow; there-
fore, the typical control handles are the recycle flow and machine speed. The basis for
capacity control is the supply-and-demand relationship of the overall process config-
uration, with the pump or compressor as an integral part. For example, the capacity
control of a supply-driven process is achieved with a pressure controller PC-101 at
the suction header, as shown in Fig. 6.37a. The pressure controller PC-102 prop-
agates the supply-and-demand imbalance forward to downstream. The protective
controller PC-103 provides overriding control if the supply-driven control reverses
to demand driven and resulting in excessively high discharge pressure.
A demand-driven control has the capacity controller PC-102 on the discharge
side, as shown in Fig. 6.37b. The supply-and-demand relationship is propagated
backward via the two normal regulatory controllers, PC-102 and PC-101, with the
overriding controller PC-103 to protect the pump against changes in the supply-and-
demand relationship that can potentially cause an excessively low suction pressure.
Other means for reciprocating compressor capacity control include inlet valve
uploaders and clearance pockets. Valve uploaders are mechanisms that open or
close one or more of a cylinders’ inlet valves to provide unloading. For instance,
valve uploaders with three-step uploading can provide 0, 50%, and 100% of the
nominal cylinder capacity. On the other hand, the clearance pocket is a volume
chamber separated from the normal cylinder-end clearance volume by a valve or
Fig. 6.37 Reciprocating machine with capacity and protective control

plug. Opening the clearance pocket reduces the cylinder inlet volumetric flow by
trapping the additional gas in the enlarged clearance volume at the end of the piston
stroke.
6.3.2 Protective Control for Reciprocating Machines
Since surge and choke are not concerns in reciprocating machines, protective control
is mainly about protection against extreme pressures. The compression ratio of a
reciprocating machine is limited only by the driver’s power rating. The overriding
controller PC-103 provides the protective control function against excessively high
discharge pressure (Fig. 6.37a) for supply-driven control or low pressure on the
suction side (Fig. 6.37b) for demand-driven control.
6.4 Practical Considerations
A successful control scheme must properly address many practical details and poten-
tial abnormal operating scenarios. Some important considerations are listed here.
6.4.1 Speed of Response
Pumps and compressors operate at very high speeds with extremely fast responses
in pressure and flow. The control loop has five essential components. The speed of
response of the control loop depends on all the components in the loop and is dictated
by the worst-performing component.
Process measurements such as flows and pressures with hard-wired 4 ∼ 20 mA
signals have a typical response time of tens of milliseconds. Fieldbus (FF) signals,
which have a latent transmission delay, are usually not adequate for anti-surge control,
although sufficient for capacity control.
The final control element (FCE), with a typical response time in seconds, is the
bottleneck for control. The typical requirement is that the recycle valve should be
able to move from a fully closed to a fully open position in less than two seconds.
On the process side, the speed of response is measured by the response time from
the time the recycle valve is opened to the time the desired recycle flow is attained. The
surge volume, measured by the amount of gas entrapped at the machine’s discharge,
is a critical factor affecting the speed of response. In case of a surge, it is critical to
allow the trapped gas between the compressor discharge and the recycle to quickly
recirculate back to the suction, as shown in Fig. 6.38. The recycle line is preferably
taken off before the discharge cooler to minimize the surge volume. However, the
recycled gas requires a suction cooler after the recycle tie-in point to cool it down.
6.4 Practical Considerations 187
Fig. 6.38 Minimizing surge volume to increase speed of response
Fig. 6.39 Surge volume with hot recycle
For multi-stage compressors or multiple compressors operating in series, the rec-

ommended piping is illustrated in Fig. 6.38, where the take-off of the recycle lines is
from upstream of the coolers and is sent back upstream of the suction coolers. With
this arrangement, the surge volume for each stage is minimized.
If for process reasons, e.g., in E&P operation where a suction cooler may not be
installed, the recycle flow must be drawn from downstream the cooler (cold recycle),
the surge volume will include the discharge cooler, which can significantly increase
the response time. See Fig. 6.39. A bypass line (the red line) may need to be provided
if the speed of response becomes a concern, which is determined by vigorous dynamic
simulation.
A cold bypass line is taken downstream of the discharge cooler. In extreme cases,
such as startup and trip, a hot bypass line upstream of the discharge cooler may also
be required to protect the compressor.
The speed of control calculation, if implemented in standard DCS, is determined
by the scan frequency of the DCS. With modern DCS systems, it is common to see
execution speed at 100 ms or faster. Some proprietary implementations may offer

control systems that execute at 10 to 20 ms. However, considering that the bottleneck
is the control valve at a speed of 2,000 ms (2 s) (Elliott and Bloch 2021), it is hard
to justify raising the control calculation speed to 10 ms for most applications.
Unlike most control solutions from commercial vendors where the regulatory
control, protective control, and safeguarding function are all lumped in a single
proprietary software and hardware system, the control strategy discussed in this
book has a clear separation between the three levels of the functions as three lines
of defenses against a surge. Capacity control does not require the same speed of
execution as anti-surge control does. During turndown, the capacity control starts to
open the recycle valve as soon as the operating point reaches the capacity control line
(CCL), which is ahead of the surge control line (SCL). Most of the time, the capacity
control is sufficient to stabilize the operating point at the CCL without triggering
the anti-surge control. If capacity control fails to hold the line and anti-surge control
must respond, the valve stem is already lifted up from its seat by the capacity control
action, thus reducing the response time of the anti-surge control valve. Lifting the
valve from its seat can take up to half a second if starting from a fully closed position.
6.4.2 † Sensitivity Analysis
Compressor control, including capacity control, anti-surge control, and anti-choke

control, all depend on knowing the location of the operating point in relation to
the surge line. Although an accurate formula is desired to represent the SRL, many
practical factors affect the accuracy of the SRL calculation, with the following being
a few:
1. Surge points. The surge points on the performance curves provided by the pump
and compressor manufacturers are estimated based on the mechanical construc-
tion of the machine and the equation of states (EOS) of the fluid property. Although
the performance data are usually verified by surge tests at manufacturer’s facility
and incipient surge tests at the site, the accuracy of these surge points can still
vary noticeably. A 1% or 2% error is not uncommon.
2. Surge reference line. The surge reference line (SRL) is, at best, an approximation
of the last known location of the most conservative surge points and surge lines.
The SRL may shift unpredictably with time. The changes in the gas property (e.g.,
molecular weight) and operating condition (e.g., temperature and pressure) also
impact the ASP calculation results. In addition, the assumption of Z s /Z d being
constant does not exactly hold and can be a source of error.
3. Accuracy of curve-fitting. The surge reference line is derived from the surge points
via curve fitting. The ASP formula for representing the SRL is not a perfect fit to
the surge points.
4. Field instrumentation. Errors and uncertainties in measurements are inevitable.
These errors will directly affect the accuracy of the ASP calculation.
6.5 Summary 189
5. Recycle valve. Valve stiction or slip is common as the valve experiences wear
and tear over time. As a result, even if the anti-surge control action is adequate, it
may not achieve the desired response if the valve performance has deteriorated.
An extra margin needs to be added to maintain the desired safety margin.
6. Control margin. The surge reference line serves as an indication of when the surge
phenomenon may start. A safety margin must be included in practical operation
and control, usually 10% of the flow. The choice of 10% is purely based on
experience and may not be adequate for all applications. As most people tend to
err on the safe side, this control margin may have a large built-in margin.
For the above reasons, the SRL calculation is not rocket science requiring high
precision. It is merely an approximation affected by many practical factors. It is
critical to understand the key factors contributing to the impreciseness of the SRL
and budget the effort accordingly when calculating the ASP coefficient. For instance,
the SRL can be approximated with a straight line, a parabolic function, or as complex
as a neural-network model. However, the incremental improvement gained by using
a more complex formula is easily offset by other uncertainties or by adjusting other
tuning parameters, e.g., the 10% safe margin. A balanced decision based on a holistic
view of all the uncertainties is always recommended.
6.5 Summary
The capacity control design of pumps and compressors is dictated by the overall
process control needs, determined by the supply-and-demand relationship of the
process flow configuration. The pump and compressors are controlled to respond
to the overall process capacity control. The primary control handles can be one or
more of the machine speed, control valve, and recycle valve. The control design thus
varies with the control targets to meet, and the control handles available. However,
the governing factor is the cause-and-effect relationship between the controlled and
manipulated variables.
The minimum recycle flow control of centrifugal pumps and anti-surge control
of centrifugal compressors work on the same principle, except that a pump deals
with an incompressible liquid, and a compressor handles compressible gases. The
objective is to keep the machine flow above the minimum flow limit. However, the
challenge is knowing the current flow rate in relation to the minimum flow limit. An
incompressible liquid assumes a constant density, and calculating the minimum flow
limit is relatively straightforward. It can be a fixed flow value for a fixed-speed pump
or a variable flow value that can be inferred from the pressure or speed measurement.
A more general approach that applies to all centrifugal machines is to use the anti-
surge parameter (ASP). The anti-surge control or minimum flow control becomes
a standard PID control scheme, with the ASP value being the process value (PV),
while the setpoint assumes a constant value at 33% with the SRL at 30%. With the
same ASP calculation, the choke reference line would have approximately a value
of 100%. Therefore, anti-choke control can also be easily implemented if desired.
Calculating the anti-surge parameters is the major effort in the control design
involving centrifugal machines. The ASP calculation is based on several types of
data from different sources and can thus seem overwhelming. Understanding the
control objectives and data requirements can reduce the frustration.
The control design for pumping processes is relatively straightforward compared
to the control design for compression processes. However, many pumps in operation
do not have adequate pump control, resulting in inefficient and unsafe operations
and significant energy waste.
References
API (2014) API standard 670 – machinery protection systems. Technical report, American
Petroleum Institute
Botros K, Hill S, Grose J (2015) A new approach to designing centrifugal compressor surge control
systems. In: 44th turbomachinery and 31st pump symposia, Houston
12th turbomachinery symposium, College Station, Texas
ISO-5167 (2003) Measurement of fluid flow by means of pressure differential devices inserted in
circular cross-section conduits running full, 2nd edn. ISO
Smith CL (2010) Advanced process control – beyond single-loop control. Wiley Inc
Chapter 7
Advanced Control Solutions
Chapter 6 presents some basic control designs for simple pumping and compression
processes. In practical applications, however, the process configuration can be signif-
icantly more complex in order to meet advanced operational requirements. Examples
of this complexity include multi-stage compressors with a single driver; side streams
joining or leaving the flow path; multiple machines operating in series or parallel;
and wide variations in flow capacity, operating pressure, and fluid properties. The
ever-increasing complexity requires a better understanding of the process flow, the
machine characteristics, and the detrimental effects of a surge in order to reach the
desired level of optimality in operation and control. Additionally, when multiple
machines are operating together, they expose many optimization opportunities to
improve reliability or efficiency. This chapter presents several design examples to
show how the basic schemes presented in Chap. 6 can be used to achieve advanced
control solutions for complex pumping or compression processes.
7.1 Integration Between Capacity and Anti-surge Control
The different components of a complex compression system interact and even inter-
fere with each other if not properly integrated. An integrated solution that may include
capacity control, anti-surge control, load-balancing control, and other protective con-
trols must be based on a coherent design to ensure overall optimality (Golden et al.
2002).
7.1.1 Capacity Control Revisited
Capacity refers to the throughput flow. There are machine capacity and
process capacity, corresponding to the machine flow and process flow. At the process
level, the capacity refers to the process capacity, i.e., the process flow. Adjusting the
machine capacity is an essential part of process capacity control.
https://doi.org/10.1007/978-3-031-43122-7_7
192 7 Advanced Control Solutions
Fig. 7.1 Capacity turndown for centrifugal machines
Centrifugal machines have three primary control handles for capacity control:
machine speed, system resistance, and recycling. While for reciprocating machines,
the control handles are usually the machine speed and recycle since the flow is
independent of the system resistance. The cause-and-effect relationship is illustrated
in Fig. 7.1.
• Machine speed. Reducing the machine speed causes the operating point to move
along the constant-resistance curve until it reaches the minimum allowable speed
at point C . Further reducing the process flow to C requires recycling.
• System resistance. Increasing system resistance causes the pressure ratio to
increase and flow to decrease. As the resistance increases, the operating point
moves along the constant-speed curve from A to B until the surge line is reached.
Turning down the capacity further from B to B will require recycling.
• Recycle. The desired process flow can also be achieved by recycling from A to
D without changing the machine speed or resistance. A fixed-speed reciprocating
machine is typically based on direct recycling, which is simple and reliable but
highly inefficient. It is typically the last resort for capacity control.
The control handles affect the machine capacity in different ways. In most appli-
cations, capacity control is achieved with a combination of two or more control
handles. For instance, a typical control design for a variable-speed machine uses
the machine speed and throttling valves to maintain the desired flow and pressure.
If the operating point drops below the surge limit, recycle flow must be introduced
for further turndown. The trajectory of the operating points is illustrated in Fig. 7.2,
where the capacity control is under a constant-pressure ratio. The operating point
can reach any point within the operating envelope by using multiple control handles.
7.1 Integration Between Capacity and Anti-surge Control 193
Fig. 7.2 Capacity turndown for reciprocating machines
Fig. 7.3 A generic 2 × 2 compression process
We will demonstrate the control design with a supply-driven process consisting of

two variable-speed compressors, each with two stages. See Fig. 7.3 for an illustration.
This 2 × 2 process is generic enough to be scaled up to more trains or stages or
scaled down to a single train or stage. The pressure ratio and throughput flow must
be controlled, requiring all three control handles. The interaction among the multi-
ple control handles and control targets constitutes a multivariable control problem,
requiring an advanced control solution.
Fig. 7.4 A minimal control design for capacity control
7.1.2 An Unintegrated Design
A direct scale-up from the control design of a single compressor train in Fig. 6.36
to the 2 × 2 compression process in Fig. 7.3 is shown in Fig. 7.4. The capacity
control is supply driven, with the suction pressure controller PC-001 manipulating
the rotating speed of both compressor trains to control the process capacity. The
discharge header pressure controller PC-002 propagates the supply variation further
downstream. The pressure controllers at the suction and discharge headers (PC-001
and PC-002) maintain a constant compression ratio.1
With the pressure ratio under control, the trajectory of the operating points of all
stages is approximately a horizontal line, as illustrated in Fig. 7.5.
During severe capacity turndown, the operating point will move to the left along a
horizontal line from point A to E , under the combined effect of speed reduction and
pressure control (throttling). When the operating point reaches the SCL, about 10%
of the flow to the right of the surge reference line (SRL), the anti-surge controller
kicks in and quickly opens the recycle valve to increase the surge margin. The process
flow continues turndown and reaches the desired point of E. The compressor flow,
as the sum of process and recycle flow, remains at point E . The anti-surge controller
(e.g., UC-111 for Stage 1) determines the required amount of recycle flow.
1 The two pressure controllers, PC-001 and PC-003, are installed at the suction and discharge
headers, which differs from the pressure at the compressor suction and discharge. Strictly speaking,
the pressure ratio P2 /P1 , not Pd /Ps , is kept constant by the two capacity controllers.
Fig. 7.5 Trajectory of operating point for a primitive capacity control design
This minimal design works but is very primitive. The control result is not smooth
or reliable because the anti-surge controller (UC-xxx) serves two conflicting roles:
capacity control and anti-surge control during a severe turndown. The high priority
of anti-surge control demands aggressive action, but capacity control requires smooth
and accurate control actions. As a result, the turndown from point A to E (see Fig. 7.5)
under pressure controller PC-001 is slow and smooth, but the further turndown from
E to E under anti-surge controller UC-111 may be overly aggressive.
The capacity control (PC-001, SC-101, SC-201) and anti-surge control
(UC-111, UC-121, UC-211, and UC-221) functions are completely indepen-
dent, although their control handles (speed and recycle valves) and control targets
(capacity and surge margins) are heavily coupled. Without coordination, the machine
can run into an inefficient operating scenario with a high recycle flow well before
reaching the minimum flow limit. The capacity control must be decoupled from
the anti-surge control to improve the control performance, requiring an integrated
solution.
7.1.3 Integrated Design of Capacity and Anti-surge Control
The control design in Fig. 7.4 relies on the anti-surge controllers (UC-xxx) to
perform both capacity and anti-surge control duties with conflicting performance
requirements. The result is thus not ideal when the operating point is near the surge
line.
Fig. 7.6 An improved control design for capacity control
An improved design is to enhance the capacity control with a split-range control

so that the capacity controller and anti-surge controller are separated in function.
The capacity control is responsible for maintaining the capacity, while the anti-
surge controller is dedicated to surge prevention only. This separation is achieved by
introducing a capacity control line (CCL), typically set to 36%, which is 20% (of
flow) away from the surge reference line (SRL). The capacity turndown is treated as
a normal operating condition. The capacity controller uses the machine speed and
the recycle in a split-range arrangement to adjust the capacity smoothly. The more
aggressive anti-surge controller will be activated only when the capacity control fails
to hold the capacity line, and the compressor flow falls below the surge control line
(SCL). The schematic for this improved design is shown in Fig. 7.6.
The capacity controller (e.g., UC-112) and the anti-surge controller (e.g.,
UC-111) are two separate controllers sharing the same recycle valve as the con-
trol handle. They can thus be independently tuned to suit the different dynamics.
For example, the capacity controller can have a normal tuning for smooth control as
the first line of defense against surge, while the anti-surge controller can be tuned
much more aggressively to serve as the next line of defense. Under normal operating
conditions, the operating point would remain at the right side of the CCL. If both
Fig. 7.7 Trajectory of operating point with improved capacity control design
Table 7.1 Capacity control with a split-range scheme

Capacity Controller Output Process Flow Compressor Flow Recycle Valve
0% - 36% 0% - 36% 36% 100%-0%
36% - 100% 36% - 100% 36% - 100% 0%
UC-111 and UC-112 are active, the larger controller output will be sent down, via
the high selector, to the recycle valve UCV-111.
The trajectory of the operating point is illustrated in Fig. 7.7. Compared with
Fig. 7.5, the recycle valve will be opened at the CCL instead of the SCL. The capacity
turndown from Point A to E is via speed reduction under pressure control PC-001.
Once the CCL is reached, the further turndown from E to E is via the recycle valve
under the secondary capacity controller UC-112. The compressor flow will remain
at E while the process flow can be reduced to zero by introducing sufficient recycle
flow.
The relationship can be illustrated in Fig. 7.8, where the process flow, compressor
flow, and recycle flow constitute an interesting split-range control scenario: the turn-
down of process flow is firstly by reducing compressor flow. When the compressor
flow reaches the CCL, the recycle flow is introduced to further reduce the process
flow. The turndown of the process flow via the capacity controller is thus smooth and
stable, as shown in Fig. 7.8a and illustrated by Table 7.1.
Suppose the capacity controller UC-112 fails to hold the compressor flow at the
capacity control limit, and the operating point reaches the SCL. In that case, the
anti-surge controller UC-111 aggressively opens the recycle valve, which is by the
Fig. 7.8 The split-range control in compressor capacity control
necessity of surge prevention, but is a significant disturbance to capacity control by

itself. See Fig. 7.8b.
Under normal operating conditions, the operating point would unlikely reach the
surge control limit when the capacity control scheme performs well. In this sense, the
improved capacity control helps the anti-surge control. This improved design scarifies
a small margin of the operating capacity (between the CCL and SCL) but offers
improved reliability and performance. This sacrifice is worthwhile for applications
where the operation near the surge line is sporadic and temporary.
A properly sized compressor would have the normal operating point near the best
efficiency point (BEP), which is well away from the surge reference line and even
the capacity control line. If the machine is constantly running at or near the surge
line, it implies that the machine is oversized for the operation. The operation will be
inefficient no matter how the control solution is designed.
7.1.4 † Integrated Design with Feedforward Compensation
The design in Fig. 7.6 is an improvement over Fig. 7.4. However, the reactive nature
of the capacity controller (e.g., UC-112) remains a drawback since it is based on the
feedback principle and has to wait for the surge margin to drop below the capacity
control limit before it reacts.
A further improvement in control design is introducing feedforward actions to
react to capacity turndown requests proactively, as illustrated in Fig. 7.9. The output
(OP) of the capacity controller PC-001 represents the desired surge margin. This
output can thus serve as the feedforward input to force the compressor to respond
7.2 Load-Balancing Control 199
proactively via recycle valve instead of waiting for the compressor to respond pas-
sively through machine speed change.
One modification to the design is to take the desired surge margin (i.e., the output of
the capacity controller PC-001) as the process value (PV) for the capacity controllers
UC-112, UC-122, UC-212, and UC-222. The controller compares the requested
surge margin from capacity controller PC-001 with the capacity control limit (36%
in this case) and reduces the recycle valve opening if the desired surge margin
exceeds 36%. Conversely, if the surge margin requested by capacity control is below
36%, the controller UC-112 will open the recycle valve to allow more recycle flow.
See Fig. 7.8.
The second modification is introducing two load controllers XC-101 and
XC-201, where the load is defined as the surge margin between 0% and 100%.
The output of the capacity controller PC-001, the desired surge margin, is sent
down to the load controllers. If the requested load is higher than the capacity control
limit of 36%, the load controllers will adjust the machine speeds to achieve this surge
margin. If the requested surge margin is less than 36%, the load controller would not
further reduce the machine speed but instead relies on the recycle flow to achieve
further capacity turndown.
The trajectory of the operating point is the same as in Fig. 7.7, except that the
controller UC-112 does not wait for the operating point to reach the capacity con-
trol line to introduce recycle flow. The capacity controller PC-001 triggers this
action. Therefore, it is a feedforward signal to the capacity controller UC-112.
Due to the feedforward nature, the control action is quick and proactive, leading to
smooth capacity control performance and reducing the fluctuations that may drive
the machine close to the SCL.
7.2 Load-Balancing Control
Multiple compressors are often used in parallel to increase the capacity or in series to
achieve a higher head. Distributing the load among the different compressor stages
or trains is an optimization problem of significant economic value.2
Surge and choke are two critical conditions limiting a compressor’s operating
range. Avoiding surge has a higher priority than achieving the best efficiency. Thus,
it makes more sense to use the surge margin as the basis for load balancing (Hafaifa
et al. 2014). In other words, load-balancing control should keep all the compressors
and all the stages at the same surge margin instead of the same flow or speed. In
addition, load balancing based on surge margin (ASP value) also makes mixing and
matching different designs and capacities simple and reliable in complex networks
(Jacobson et al. 2016).
2 Pumping process with multiple pumps follows a similar concept but is more straightforward.
Fig. 7.9 An integrated capacity control solution
7.2.1 † Load Balancing Among Compressor Trains
For a complex compression process with multiple compressor trains operating in

parallel, the flow distribution among the multiple trains can significantly impact the
achievable capacity range and the operating efficiency.
As discussed in Sect. 7.4, when multiple compressors of identical design operate
in parallel, the best efficiency is obtained when all compressors have the same speed
and flow (and thus head). However, when compressors of different designs and
capacities are put together, running them all at the same flow rate or speed is no
longer optimal. For example, Fig. 7.10 illustrates the flow and pressure range of two
different compressors.
Compressors operating in parallel share the same pressure ratio because of com-
mon suction and discharge pressure headers. It would be inefficient for the compres-
sor with higher capacity to recycle simultaneously as the low-capacity compressor
does. Instead, it is more efficient for the two parallel compressors to have different
recycle flows.
Fig. 7.10 Load of different compressors
A “lazy” approach designates one machine as the base load machine, operating
at a fixed load (speed or flow), and the other as the swing machine to regulate the
overall capacity. However, this “base loading” approach is inefficient (see Sect. 7.4).
It also increases the surge risk because the swing machine is solely responsible for
taking the capacity fluctuations.
The control design in Figs. 7.4 and 7.6 balances the compressor load by forcing
all the compressors to run at the same speed, which is acceptable if all the com-
pressors have similar performance (capacity and head). In actual applications where
compressors are of distinct sizes,3 the same speed and pressure ratio will cause the
compressors to operate at different flow rates, with different efficiency and surge
margin, as shown in Fig. 7.10. Instead of balancing the multiple compressor trains
on speed or flow, it is safer and more efficient to balance the operation based on the
surge margin.
The integrated design introduces two load controllers, XC-101 and XC-201
in Fig. 7.9, which balance the two machines on the same surge margin. The two
load controllers receive the same setpoint from the capacity controller PC-101
in the form of the desired ASP value and manipulate the respective compres-
sor speed to drive the minimum ASP value for all stages on the same train (i.e.,
ASP1=min(ASP11,ASP12)) to the same setpoint value.
Under the same compression ratio (head), the two compressors will operate with
the same surge margin but at different speeds and flow rates (capacity). The per-
3 It is almost impossible to find two compressors with identical performance characteristics due
to the difference in design, age, or wear. Even identically designed machines will exhibit different
resistance to flow in a real-world setting, resulting in unbalanced loads if uncontrolled.
Fig. 7.11 Load balancing for parallel compressors
formance curves are illustrated in Fig. 7.11, where the operating points for the two
compressors are at A1 and B1, respectively.
Note that the setpoint is set to the capacity control limit of 36%, which is the
split point for the split-range control between machine speed and recycling. Above
36%, capacity control is achieved by machine speed. Below 36%, the recycle valve
is opened to reduce the process flow further.
7.2.2 ‡ Load Balancing Among Stages
For compressors with multiple stages on a common shaft, all stages will operate at
the same speed. Since not all stages are built with perfect capacity balance, each stage
will not approach the surge line simultaneously during severe turndown operations.
The worst-performing stage will reach the surge line while others still remain safe
from the surge limit.
The conventional control design is a simple fan-out control that sends the same
capacity control output to all the load controllers (e.g., XC-101) and capacity con-
trollers (e.g., UC-112). The recycling at all stages is simultaneous, based on the
worst-performing stage, as shown in Fig. 7.9, regardless of the actual surge margin
of each stage.
When the two compressors operate in series, they will be subject to the same mass
flow rate, but the pressure ratio or speed is different. Recycling at the same time and
Fig. 7.12 Load balancing among compressor stages
by the same amount by all stages based on the worst-case scenario is operationally
inefficient, although simple in design.
The more efficient design is to open the recycle line only when necessary. This
just-in-time recycling is achieved with additional calculation blocks at each stage,
e.g., XY-101 and UY-113, for the first stage of the first train. Other stages are
configured similarly. See Fig. 7.12 for the complete design.
The desired surge margin is calculated by the capacity controller PC-001 based
on the suction pressure. An increase in pressure demands the machine to speed up
and increase the throughput to bring the pressure back on target. The output of the
capacity controller is the desired surge margin in ASP, which would be sent down to
the capacity controllers at every stage.
The calculation block XY-101 computes the difference between the desired ASP
value for this train and the minimum ASP value of all the stages on this train. This
value is the feedforward action each stage is expected to take. However, each stage
does not have the same surge margin and does not need to recycle by the same
Fig. 7.13 Load balancing among compressor stages (example)
amount. The difference between the ASP value of the current stage and the worst-
case stage is calculated by UY-111 and extracted from the change request by the
capacity controller. As a result, each stage will adjust the demanded value with its
own surge margin and recycle only when the surge margin is exhausted.
Figure 7.13 provides a sample scenario to illustrate the load-balancing concept
among compressor stages.
Assume that the compressor is in stable operation. Stage 1 of the first train has
a surge margin of ASP = 42%, and Stage 2 with ASP = 34%. Suppose a sudden
decrease in inlet gas flow causes the suction pressure to drop. The direct-acting
capacity controller PC-001 responds with a decrease in its output to 30%, aiming
to reduce the compressor throughput. This 30% is thus the new load target, in the
form of surge margin ASP = 30%.
The new load target ASP = 30% is sent down to the two load controllers, XC101
and XC-201, via the two high selectors, which clamp the value to 36%. In other
words, the two load controllers are only required to bring the surge margin down
7.3 A Fully Integrated Control Design 205
to 36%. The new load target ASP = 30% is sent to the recycle valves of all four
stages simultaneously. The four secondary capacity controllers UC112 to UC-222
are responsible for the flow turndown below 36%.
An offset is added to the desired surge margin for each capacity controller
UC-xx2 as calculated in Fig. 7.13. The result is that Stage 2 will start recycling
since UC122.PV at 30% is below its setpoint of 33%. Stage 1 with ASP = 42%
has a sufficient surge margin and does not need to recycle immediately. The calcu-
lation is repeated at each execution scan, and eventually, all stages will be running
at approximately the same surge margin.
For stability, the difference in valve opening among the stages should be limited,
e.g., to LB<10%. In addition, a 30-minute filter is applied to the feedforward request
LB to reduce the effect of the measurement noises and avoid fluctuations, and it can
be fine-tuned during commissioning.
7.3 A Fully Integrated Control Design
The process configuration can differ considerably from application to application.

The control solution can thus be drastically different as well. With a good understand-
ing of the overall control strategy and intimate familiarity with the basic designs, more
complex control solutions can then be designed to address specific needs.
7.3.1 A Complete Solution for Compressor Control
A fully implemented process control solution can include many components, as

summarized in Fig. 7.14.
With a good understanding of the process dynamics and compressor characteris-
tics and following the basic design examples discussed here, it is straightforward to
tailor the generic design in Fig. 7.14 to suit the specific applications. For example,
additional control loops, such as discharge pressure overriding control for each stage
and power protection for each motor, can easily be added to the integrated control
scheme.
A fully integrated control design comprising capacity control, anti-surge control,
load-balancing control, and protective control is illustrated in Fig. 7.15, with the
following features:
1. Capacity control. As a supply-driven operation, the top-level capacity control is
achieved by suction pressure controller PC-001, cascaded to the two load con-
trollers, XC-101 and XC-201, to manipulate the speed of the two compressors.
The output is also fanned-out to the four stages and cascaded to the recycle con-
trollers UC-112, UC-122, UC-212, and UC-222.
The discharge pressure controller PC-002 is the relay controller to propagate

the supply/demand fluctuation further downstream.
2. Load balancing. Load-balancing control is provided among the compressor trains
and stages.
• Load balancing by train is achieved by the load controllers (XC-101 and

XC-201). The load of the two trains is balanced on the same surge margin
as both controllers receive the same setpoint from the master capacity control
controller PC-001.
• Load balancing by stage is achieved by a feedforward term from the master
capacity controller PC-001, adjusted with the actual surge margin of each stage
through the recycle controllers UC-112, UC-122, UC-212, and UC-222.
3. Anti-surge controller. Each stage has an independent anti-surge controller

(UC-111, UC-121, UC-211, and UC-221). The anti-surge controllers share
the same recycle valve with the capacity controllers to provide surge protection
in case the capacity controllers (UC-112, UC-122, UC-212, and UC-222) fail
to hold the surge margin above the capacity control limit of 36%.
4. Protective controllers. Several typical protective controllers are included in this
design, while more protective controllers can be added as necessary.
• Protective controllers PC-003 for capacity control. The discharge pressure
controller PC-003 is implemented to protect the discharge header pressure.
Suppose the discharge header pressure becomes excessively high due to reduced
demand from downstream. In that case, the discharge pressure controller will
override the capacity controller PC-001 through a low selector to slow down
the machines and bring the discharge pressure back to normal range. Simi-
larly, if the suction header pressure becomes excessively high, the protective
Fig. 7.14 Components in a fully integrated compressor control solution

Fig. 7.15 A complete control design for the 2 × 2 compression process
controller PC-001B will open the flaring valve and send excessive gas to the
flare.
• Protective controller for an electric motor. The driving motor is often damaged
due to over-current. The power limit often constrains the throughput since
the motor operation must be within the power limit. A protective controller
IC-101 on the motor power helps protect the motor and allows operations to
push for maximum capacity if the motor power is the bottleneck.
The 2 × 2 compressor system discussed here provides a general example of
advanced control design. The system can be simplified to a single stage or scaled up
to multiple stages and trains.
In many applications, the motor power is the capacity limit. The regulatory con-
troller PC-101 and the protective controllers (e.g., IC-101) can work together
to push for the maximum capacity subject to the power limit. Setting the capacity
control setpoint (PC-101.SP) at a lower value will force more throughput up to

the motor power limit.
7.3.2 † Integration Between Control and Safeguarding
As shown in the safety onion in Fig. 1.20, process control is the first line of defense
against undesired operating conditions. If the control layer fails, the instrumented
safeguarding layer proactively trips the equipment or unit to avoid more severe
consequences, such as loss of primary containment or equipment damage.
Although safeguarding logic is essential, they typically do not appear on the
control diagrams.
7.3.3 A Real-World Example
We present a plant-wide control example to conclude the advanced control design

discussions. An integrated oil-and-gas production facility was built to handle oil-
and-gas production via miscible gas injection (MGI)—an enhanced oil recovery
mechanism. This multi-billion dollar green-field plant has a design capacity of 6
MMSCMD of sweet gas, 60,000 BPD of oil production, and 16 MMSCMD of high-
pressure sour injection gas. A process overview diagram (POD) is developed (and
simplified) from hundreds of pages of PFD and P&ID and is shown in Fig. 7.16 for
the plant-level overview (Niu and Xiao 2022).
This example features a complex supply-and-demand model, and thus the primary
consideration of our discussion is in capacity control, i.e., dynamic material and
energy balancing. The operation is characterized by produced gas of extraordinarily
high-pressure and high-sour content (H2 S and CO2 ). For this reason, safe operation
is paramount in the process and control design.
The operating objective is to produce oil and gas for sale and sour gas for injection.
The gas processing area, which is the focus of this discussion, comprises gas produc-
tion, separation, cleaning and conditioning, and gas compression. Many packaged
units by various vendors are implemented to clean and condition the gas, condensate,
and the produced water. Each unit is a complex operation, and all have specifications
to meet, such as purity, dew point, vapor pressure, and delivery pressure. The three
consumers of processed gas include, in order of priority, export to sales, injection
wells in Area B, and injection wells in Area A.
There are 14 compressors with a total of 18 stages. Therefore, compressor control
and protection are at the center of the control design. What is noteworthy in this
control design includes:
1. The complex supply-and-demand model. The export gas compressor is demand
driven with the discharge pressure controller regulating the capacity. On the other
hand, the GSU off-gas compressor and the condensate flash gas compressors are
supply driven, and therefore the capacity control is performed by the suction
pressure controllers.
2. The two header pressures, PC-01 and PC-02, are the critical nodes for over-
all capacity control. The fluctuations in these two pressures reflect the dynamic
supply-and-demand swings. The normal operating pressure for header A is main-
tained by adjusting the wellhead pressure. However, the dynamics from the well-
head to the pressure header are relatively slow; thus, overriding controllers are
provided to protect this pressure header against high and low pressures. The low
protective controller PC-01A is provided to cut back the gas to the injection
wells if the pressure is below the normal header pressure by 1.0 bar. During
normal operation, since the PC-01A setpoint of 74 bar is lower than the normal
header pressure of 75 bar, the direct-acting pressure controller PC-01A will drive
the valve MV-17 to fully open. However, if the header pressure falls below 74
bar, the protective controller PC-01A will close the flow to area A injection first,
Fig. 7.16 Process overview diagram of an E&P surface facility (reproduced, with permission, from
Niu and Xiao 2022)
Fig. 7.17 Plant-wide control: normal regulatory control (reproduced, with permission, from Niu
and Xiao 2022)
then the flow to area B through the split-range calculation PY-01A.

If a severe plant upset drives the pressure to above 78 bar during an abnormal
situation, the excess gas will be sent to flare via a direct-acting protective pressure
controller PC-01B.
Header B pressure is maintained by manipulating the gas flow from area B. This
stream from area B is demand driven. Sufficient swing capacity must be provided
inside the area B control solution to meet the demand change.
The normal regulatory control at the plant level is primarily for the dynamic
supply-and-demand balance. Following appropriate cause-and-effect analysis and
dynamic response analysis, as described in Niu and Xiao (2022), the control overview
diagram (COD) with the normal regulatory control loops is shown in Fig. 7.17.
There are many protective controls in the design. For example, the supply-driven
streams and the compressors need to have demand override. Each compressor needs
multiple protective control loops to protect operational integrity.
The actual process is much more complicated than presented here since this is
a multi-billion dollar project involving many unit operations. However, the overall
7.4 Multi-machine Optimization 211
control strategy as the top-level design is the foundation for a sound control solution,
with the equipment control being a central piece.
7.4 Multi-machine Optimization
The operation of multiple machines can be considerably complex and challenging,

with unlimited scenarios corresponding to different process flow configurations. Here
we discuss several typical configurations to illustrate the opportunity to improve
the overall operating efficiency. The control objectives can be minimizing power
consumption, maximizing capacity, or minimizing running hours.
7.4.1 † Pumps and Compressors in Parallel
Multiple pumps or compressors can operate in parallel to increase capacity or provide

redundancy. However, they may run into the scenario that some trains operate at
full capacity while others are starved and have to recycle. Recycling is notoriously
inefficient because the machine wastes energy to unnecessarily re-pressurize the
recycled fluid.
The affinity laws for centrifugal machines state that the power required is a cubic
function of the flow or speed. As a result, the load disturbance significantly impacts
power consumption. The following simple examples illustrate the basic concept.
Example 7.1 Compressors operating in parallel. Figure 7.18 shows two centrifugal
compressors or pumps running in parallel with a total flow of F = F1 + F2 , where
F1 and F2 are the individual flows through the two trains.
Now consider the following three scenarios:
1. One machine is shut down, and the running machine takes the entire load of F.
Based on the fan laws, the following power consumption would result:
Fig. 7.18 Load distribution with two compressors in parallel

W1 = C F 3 , (7.1)
where C is a constant representing the relationship between power consumption

and flow.
2. Both machines are running, with equal flow split between the two trains, i.e.,
F1 = F2 = F/2. The power consumption becomes
W2 = C F13 + C F23
3 3
1 1
=C F +C F
2 2
1
= C F3
4
1
= W1 .
4
3. Both machines are running, and the flow split is 1:2. That is, one-third of the flow
goes through the first machine and two-third of the flow is sent to the second
machine. In this case, the power consumption changes to
W3 = C F13 + C F23
3 3
1 2
=C F +C F
3 3
1
= C F3
3
1
= W1 .
3
For the same flow F, the optimized scenario (Scenario 2) consumes only a quarter
of the power of the worst-case scenario (Scenario 1). A flow imbalance of 1:2 between
the two machines can increase the power consumption by one-third.4 Therefore, bal-
ancing the loads among the machines in an extensive network has many optimization
opportunities for significant energy savings.
The above calculation shows the general concept of operation optimization with
a focus on power efficiency only. Many other factors affect operating costs and
must be considered simultaneously. For example, safety and reliability always have
a high priority than optimization. For parallel machines with different designs and
capacities, or even for machines with identical designs but different maintenance
conditions, operating at the same flow or speed do not necessarily ensure that they
operate within the same distance from the surge line. If any machine prematurely
goes into recycling due to surge while others operate safely away from the surge line,
it would be less efficient and less stable than keeping all the machines at the same
4From scenario 2 to scenario 3, the power consumption increase is given by (1/3 − 1/4)/(1/4) =
1/3.
7.4 Multi-machine Optimization 213
Fig. 7.19 Pressure profile with three compressors in series
distance from the surge line. Advanced load-balancing control is needed, which is
discussed in Sect. 7.2.
7.4.2 † Pumps and Compressors in Series
Multiple pumps or compressors may also operate in series to achieve a higher pressure
ratio. With multiple machines working in series, how should the pressure ratio be
distributed among each machine? The simplistic arrangement is to have an equal
pressure ratio across each machine based on the best efficiency.
Example 7.2 Balancing pressure points in serial compressor operation. The com-
plex compression network in Fig. 1.10 is an example of compressors operating in
series. For the convenience of presentation, we simplify the network to a three-
compressor network operating in series, as illustrated in Fig. 7.19. The requirement
is to raise the pressure from P1 at the suction header of compressor 1 to P4 at the
discharge header of compressor 3. The total compression ratio is given by

P4 P2 P3 P4
Pressure Ratio = = · · . (7.2)
P1 P1 P2 P3
The fan laws state that the power consumption is a cubic function of the flow and
the pressure (head) is a quadratic function of flow; therefore, the power required is
the sum of the three machines as follows:
3/2 3/2
P2 3/2 P3 P4
W =C + + , (7.3)
P1 P2 P3
where C is the coefficient that relates the pressure ratio to power, which is oversimpli-
fied5 to show the general concept. It is not difficult to prove that the minimum power
5 Each stage is different in build and efficiency, so is the coefficient C. Assuming the same coefficient
is a gross simplification.
consumption W is achieved when all three machines have the same compression
ratio. That is,
13
P2 P3 P4 P4
= = = . (7.4)
P1 P2 P3 P1
Example 7.3 Balancing pressure setpoints. Assume that the discharge pressure P4
in Fig. 7.19 is at 96 bar(a), determined by export pressure. The suction pressure of
the first compressor, P1 , is manipulated by the capacity controller at 16 bar(a). The
optimal pressures P2 and P3 for the discharge header of compressors 1 and 2 are then
given by6
13 13
P2 P4 96
= = = 1.8 (7.5)
P1 P1 16
P2 = P1 × PR = 16 × 1.8 = 28.8 (bara) (7.6)
P3 = P2 × PR = 29.0 × 1.8 = 51.8 (bara), (7.7)
where P1 is the wellhead pressure in an oil-and-gas process facility. It decreases from

16 bar(a) to 10 bar(a) as the reservoir pressure decreases and the production wells
deplete. Theoretically, the optimal pressure setpoints for P2 and P3 should change as
P1 changes. That is, with the pressure P1 at 10 bar(a), the compression ratio should
now be
13 13
P2 P4 96
= = = 2.1 (7.8)
P1 P1 10
P2 = P1 × PR = 10 × 2.1 = 21.0 (bar(a)) (7.9)
P3 = P2 × PR = 21.0 × 2.1 = 44.1 (bar(a)). (7.10)
The discharge pressure of stage 1 should be decreased from 28.8 bar(a) to 21.0
bar(a) to maintain optimal energy efficiency. Table 7.2 lists the power consumption
in the three scenarios:
1. With suction pressure at 16 bar(a) and with an equal split of pressure ratio among
the three compressors, the power consumption is at 7.25C.
2. When suction pressure is reduced to 10 bar(a), if the pressures P2 and P3 remain
the same, the power consumption increases to 9.77C.
3. With suction pressure at 10 bar(a), if the pressures P2 and P3 are adjusted to main-
tain an equal pressure ratio among the three compressors, the power consumption
can be reduced to 9.13C from 9.77C.
6 An inter-stage cooler typically introduces a pressure drop between 35 and 70 kPa. The best practice
is to split the pressure drop between the discharge pressure of the stage upstream of the cooler
and the suction pressure downstream (Campbell 2004). The pressure should be at approximately
P2 = 21.2 + 0.7/2 = 21.55 bar(a) if the pressure drop at the cooler is considered.
7.5 Implementation Considerations 215
Table 7.2 Power efficiency at different pressure ratios
After the suction pressure is reduced from 16 to 10 bar(a), if the pressure profile
of the compressors does not change accordingly, the power consumption increases
by 35%. However, if the pressure profile is adjusted to maintain an equal split in
pressure ratio, the power consumption will increase by 26% instead of 35%. For
mega-watt compressors, this is a considerable saving in energy consumption.
There are many other practical factors to consider when determining the operating
pressures in actual operation.
The above analysis is oversimplified to demonstrate that there are many opportu-
nities for improving the overall efficiency of the operation. The actual situation can
be much more complex, affected by many factors. For instance, for two parallel com-
pressors with significantly different capacities, the larger one may run at base load
and the smaller as swing. The small one starts and stops as frequently as necessary.
7.5 Implementation Considerations
An integrated control solution like that in Fig. 7.15 for a complex compression process
is one of the most complex control schemes achieved with standard PID controllers
on a standard control system platform with standard hardware and software. Like
any control solution, the design balances many constraints and is mostly a trade-off
between desire and practicality.
7.5.1 Naming Convention
As the process flow configuration becomes more complex with multiple compressor
trains and multiple stages, the naming of the equipment, process tags, and control
valves can become overwhelmingly confusing. A systematic and consistent naming
convention helps improve the clarity of the presentation, significantly benefiting the
design, operation, and maintenance. For example, the control overview diagram in
Fig. 7.15 has followed the below naming conventions:
1. All tag names have omitted the default prefix unless different. For example, the tag
name of XY-101 should read as 23-XY-101, where the prefix is the sequence
number of the processing unit that the controllers are installed.
2. The three digits in the tag name have the following meanings:
• First digit: 0 for the unit level, 1, 2, 3, . . ., indicates the compressor train.
• Second digit: 0 for compressor level, 1, 2, 3, . . ., indicates the stage in the
current compressor train.
• Third digit: A sequence number.
Such a convention makes it easy to locate the tag by the tag name along. For
example:
• PC-001 means a pressure control at the unit level.
• XC-101 means an XC tag for compressor train #1 and at the compressor
level.
• UCV-211 means the recycle valve for stage #1 of compressor #2.
A consistent naming convention is possible only for green-field project designs.
For existing plants, it might be unrealistic to change the tag names.
7.5.2 Manual Versus Automatic Operation
The anti-surge controller can always be left in automatic mode except for maintenance
and crippled operation. The justification is as follows.
An anti-surge controller is a reverse-acting PID controller, implying that when the
anti-surge parameter is higher than its setpoint (e.g., 33%), the controller will drive
the surge control valve to the fully closed position. Conversely, if the ASP value
drops below the surge control setpoint of 33%, the valve will be opened. During
a shutdown, the solenoid valve is de-energized, and the surge control valve is thus
in the fully open position under the pressure of the spring for the fail-open valve.
During startup, the ASP value is typically higher than the control setpoint due to high
recycle flow and low compression ratio. The PID controller automatically starts to
close the valve upon energizing the solenoid valve.
On the other hand, the load controller should stay in manual mode during startup
until the operation is stabilized.
7.5.3 † Controller Configurations
Incorrect controller configuration can be detrimental to the performance of the con-

troller. Controller configuration is primarily the responsibility of the control system
engineers (or DCS engineers). However, some aspects require a good understanding
of the process and equipment and an intimate familiarity with the control design,
7.5 Implementation Considerations 217
which process control engineers should specify. Some of them are discussed here,
with the help of the control diagram in Fig. 7.15.
1. Control action. All the anti-surge controllers (e.g., UC111) shall be configured as
REVERSE acting. The controller output shall be configured as 0% = fully closed
and 100% = fully open.
For improved stability, a rate limiter is implemented for the anti-surge control
valve (ASCV), with the following functionalities:
a. The output to the ASCV shall be limited to between 0% and 100%, where
0% means fully closed and 100% is fully open.7
b. During normal operation, the closing of the ASCV is limited to a pre-defined
rate of 5% per minute (adjustable). No limit is imposed on the rate of opening.
The ASCV is allowed to open fast for anti-surge control but shall close slowly
to avoid chattering.
c. In case of transmitter failure (any of the inputs to the high selector becomes
BAD PV), the ASCV will be slowly ramped up to 100%. The ramping rate
is set to 7% per minute.
d. In case of emergency shutdown (PSD or ESD), the ASCV is set to a fully
open position in one shot.
2. Control equation. Anti-surge control is a protective control scheme; the I-PD
variation of the PID algorithm shall be used. In addition, Gap PID control is
recommended to allow faster action when the ASC control error exceeds a pre-
defined band. When the ASP value is below the setpoint, the normal controller
gain is restored to allow quick opening action. When the ASP value is above the
setpoint, a smaller gain (factor of 0.25) shall be used for slower action on valve
closing.
3. Anti-reset windup. The anti-surge controller (e.g., UC111) output can potentially
be overridden via the high selector by other controllers or function blocks, such as
the capacity controller (e.g., UC-112) or the high discharge pressure constraint
controller (e.g., PC-112), which may cause the controller output to wind up
due to the reset (integral) action. The anti-reset windup (ARWU) function must
be configured to prevent reset windup. See Niu and Xiao (2022) for a detailed
discussion on PID reset windup and how to prevent it.
4. Bumpless transfer. For smooth control, the bumpless transfer function may need
to be configured to ensure that no bump is created when the control mode of load
controllers (e.g., XC101) is switched between CAS and other modes. A bumpless
function block may need to be implemented to ensure bumpless transfer.
7 Since the ASCV is fail-open (air-to-close), a signal reversal should be performed in the AO block
to ensure negative feedback for the PID control loop.
7.6 Summary
Equipment control is an integral part of process control. As the process configuration

becomes more complex with many streams and equipment, a structured analysis and
design approach should be followed to ensure reliable and efficient operation. The
machine characteristics discussed in Chap. 2 are the basis for analysis, the overall
control strategy in Chap. 4 provides the general guideline, and the basic design
in Chap. 6 offers the basic building blocks for the complex control solutions. The
application scenarios always have all kinds of uncertainties, and the control solution
should be fit-for-purpose and avoid over-simplification or over-design.
References
Campbell J (2004) Gas conditioning and processing - the equipment modules, vol 2, 8th edn. John
M, Campbell and Company
Golden S, Fulton SA, Hanson DW (2002) Understanding centrifugal compressor performance in a
connected process system. Petroleum Technology Quarterly
Jacobson W, Staroselsky S, Zaghloul M, McWirter J, Tolmatsky M (2016) Compressor load-sharing
control and surge detection techniques. In: Proceedings of the 45th turbomachinery and 32nd
pump symposia, Houston
Chapter 8
Commissioning, Startup, and Monitoring
A good design is only half the work completed. It is followed by a painstaking

commissioning effort to implement the control solution in the target control system
and tie all the loose ends together to prepare the plant for the startup. A successful
startup marks the beginning of the operating phase, during which process control is a
core part of the support team to keep the control solution in operation. Performance
monitoring is essential to daily operation, with a focus on the pumps and compressors
and, more importantly, on the overall process performance. Once performance issues
are detected, timely troubleshooting and correction are necessary to keep the machine
in sound condition. All these activities are very practical oriented, with no standard to
follow. However, this chapter provides some practical methodologies and guidelines
to help with field practice.
8.1 Application Life Cycle
Process control is an essential and integrated part of a project team, whether a green-
field new project or a brown-field revamping project. Process control is a “super
glue,” interacting and collaborating with multiple specialties, from process engi-
neering, rotating equipment, instrumentation, and operations to project engineering
and procurement. For this reason, a process control engineer in the project team must
understand the work process, requirements, responsibility, authority, and deliverables
at each project phase, especially for a project with a complex process configuration
involving multiple pumps and compressors.
Large projects are typically divided into multiple phases: appraise, select, define,
execute, and operate. Engineering design, performed in the select and define phases,
normally goes through conceptual design, front-end engineering design (FEED), and
detailed design (DD).
https://doi.org/10.1007/978-3-031-43122-7_8
220 8 Commissioning, Startup, and Monitoring
Table 8.1 Process control deliverables
Once the detailed design is finalized, the project enters the commissioning phase
to assemble all components in the field. Acceptance tests ensure that the designed
solution is implemented and functioning per specification.
The project enters the operate phase after the successful plant startup. The opera-
tion phase typically lasts for several decades with various improvements and upgrades
to the design and configuration.
Table 8.1 is a simplified list of the project phases, the process control deliverables,
and quality assurance checkpoints at each phase (Niu and Xiao 2022).
Project engineering is a dedicated specialty by itself and is beyond the scope of
this book to discuss in detail. However, some special requirements and expectations
related to compressor process control are worth mentioning. The following sections
discuss several key process control activities in different project phases.
8.2 Commissioning
Implementation of the control solution is typically outsourced to third-party contrac-

tors, such as the control system (DCS, PLC) vendors, with the detailed specifications
laid out in the process control narrative (PCN) and piping and instrument diagrams
(P&ID).
During the commissioning phase, the implementer often needs to frequently inter-
act with process control to clarify and validate the control design. The milestone
activities that require process control’s active involvement include the factory accep-
tance test (FAT), site-acceptance test (SAT), and startup to ensure that the installed
solution will perform as per design.
8.2 Commissioning 221
8.2.1 Factory Acceptance Test (FAT)
The different subsystems or components, such as the piping, equipment, and control
solutions, are usually designed, manufactured, or implemented by different vendors
and contractors. A factory acceptance test (FAT) is required for each subsystem to
ensure they meet their design specifications.
The name of the factory acceptance test (FAT) indicates that the acceptance tests
are performed in the manufacturer’s facility. For example, FAT for the process control
solution is often performed at the control system vendor’s facility, where the control
system such as DCS or PLC, is temporarily hosted before being shipped to the site.
At the time of FAT, the control scheme has been implemented in the target control
system but is not yet connected to the field devices or instruments.
A process control solution is typically built with PID control loops. As a feedback
control loop, a PID consists of five essential components (see Fig. 6.28), including the
process and equipment, measurements, final control elements, control objectives, and
control algorithm. The final control elements and live measurements are unavailable
yet at the FAT stage. For the acceptance test purpose, all I/O points are connected
to, and terminated at, the junction boxes from where the actual field instruments and
devices are simulated with on/off switches, signal generators, and simulators.
Figure 8.1 shows the data flow in a PID feedback control loop, with the discon-
nection between the control system and field equipment (compared with Fig. 1.13).
The closed-loop performance is tested with all the process measurements simu-
lated at the analog input (AI) modules. The controller output is sent to the analog
output (AO) block, where the signal is measured and validated. Therefore, FAT is a
functional test on the control solution focusing on the correctness of the control loop
configuration inside the control systems (e.g., DCS). The dynamic response behavior
cannot be tested until the controller is connected to the process and equipment.
The process control engineer is responsible for establishing the FAT protocol and
reviewing/approving the FAT test plan for the control scheme.
FAT is typically conducted in two steps: white-box testing and black-box testing.
White-box testing focuses on internal configurations. It typically involves the veri-
fication against the control overview diagram (COD) and the piping and instrument
Fig. 8.1 Data flow in open control loop for factory acceptance test
diagram (P&ID) to ensure that the process control specifications are correctly fol-
lowed, and all the controller and supporting elements are properly configured. Some
of the key elements to test on the controller configurations include the following:
1. Loop check. For every control loop, follow the control schematic and ensure the
control loops are complete from the analog input AI module (for measurements),
controller, and to the analog output (AO) module (for the final control elements),
as per the control narratives.
2. Anti-surge parameter calculation. Verify that the input tags are properly connected
and with the correct engineering units, especially the flow transmitter, which shall
be the square root of the raw pressure differential signal. Also, verify that the anti-
surge parameter formula and coefficients are entered correctly and are consistent
with the pressure measurement, e.g., gage versus absolute pressure, engineering
units.
3. Anti-reset windup. All PID controllers whose output may be disconnected or
overridden by other control blocks must be adequately configured against reset
windup. The PID anti-reset windup (ARWU) configuration should be verified by
checking the output values of the controller under different scenarios. See Niu
and Xiao (2022) for the need and configuration of the anti-reset windup.
4. Valve move speed. The anti-surge control valve must open fast to respond to
surge protection requirements and close slowly to avoid chattering in the control
action. The PID or valve configuration needs a rate of change limit to enforce this
requirement.
5. Operator interface. Anti-surge control and anti-surge trip status should be provided
to the operators in real time. Since the operator can do little to influence the anti-
surge performance, the mode and setpoint for the anti-surge PID controller are
typically locked from operators.
Black-box testing focuses on external behavior, including the control loops’ func-
tionality, the startup sequences’ correctness, and the trip logic against the cause-and-
effect diagram (CED), as a few examples. As a result, black-box testing primarily
consists of systematic perturbations of selected variables and observing the response
of the output of the control components under test.
For example, to validate the anti-surge parameter calculation, we can “force” the
suction pressure, discharge pressure, and flow values at the AI module to some values
that we know will produce an ASP value of 30%, then verify that the output value
of the ASP calculation block indeed produces a 30%.
The direction of control action must be correctly configured and verified. Assume
the anti-surge controller (ASC) has a default setpoint of 33%. By “forcing” the PV
value to 30%, we expect the controller to ramp up its output and open the anti-
surge valve. However, since the signal is not sent to the valve in the field, and thus no
feedback comes back to the controller, the control error will remain at 33% − 30% =
3%. Due to the integral action, the controller will continue increasing the output to
100%. It is crucial to note that although this behavior is correct in a FAT environment
since the control loop is not fully closed, the loop response is different in actual
operation with closed-loop control.
Fig. 8.2 Data flow in closed control loop for site-acceptance test
The surge test of the compressor is a mandatory part of the FAT to demonstrate
that the compressor’s performance complies with the requirements. The surge per-
formance at the maximum speed, minimum speed, and the speed corresponding to
the certified case (see Sect. 5.1) must be tested at a minimum. Process control may
not be required to witness the surge test at the manufacturer’s shop, but the surge test
data must be made available to validate the performance curves and, thus, the surge
indicators.
Another vital aspect of FAT is testing the operability and user-friendliness of the
human–machine interface (HMI). The HMI’s hierarchy and their compliance with
the graphics standard should be reviewed, and the navigation links be tested. Layered
HMI graphics is a standard requirement for HMI design. The functionality of each
control loop should be verified from their respective controller faceplate.
The deliverable from FAT is typically a punch list that logs all the issues discov-
ered, including those fixed in situ and those pending resolutions.
8.2.2 Site-Acceptance Test (SAT)
Site-acceptance test (SAT) is another crucial checkpoint involving process control

to ensure the correctness and integrity of the control loops to get ready for startup.
SAT is conducted at the site when all the I/O points are connected with the actual
field instruments and devices, but before the plant is started with the process gas.
The control loop in Fig. 8.1 is “closed” during SAT to become Fig. 8.2.
At the SAT stage, the loop dynamics shown in the diagram still do not reflect the
true response of the process since the plant has not yet gone “live.” The test is on
the integrity of the information flow from the controller to the field and back to the
controller, including the final control elements and process measurements. In other
words, a change in controller output will be sent to the valve (actuator) in the field,1
but the process variable will not change since the plant is still in shutdown condition.
1 The instrument air should have been available at this time.

Nevertheless, the SAT is a necessary step in acceptance testing but insufficient to

ensure the correct response of all the system components.
The scope of SAT for process control covers the entire plant or unit operation,
with the pump and compressor as integrated components. The goal is to test and
verify that the control implementation complies with the detailed design and meets
the control performance expectation at all levels.
The site-acceptance test typically starts with a follow-up on the punch list from
the FAT to verify that all the issues on the punch list from the factory acceptance
test are properly addressed. This step may also involve spot-checking the controllers
against the loop configuration specifications, especially the anti-surge controllers, to
ensure that the configurations are not intentionally or inadvertently altered after FAT.
The equipment manufacturer is typically required to be at the site to demonstrate
the equipment performance to the client and establish the performance baseline. This
baseline can be used later to detect performance deviations during normal operation.
For instance, for a compressor, the head, capacity, power, and efficiency will be
tested and evaluated (Brun and Nored 2006). The manufacturer-certified point (see
Table 5.1) typically must be verified with the inlet condition matching the datasheet.2
The first step of the SAT usually is on the sensors and transmitters, which typically
include the suction pressure, discharge pressure, and flow at all compressor stages.
At the time of SAT, instrument engineers should have installed and tested all the
sensors and transmitters. However, some control loops, such as the anti-surge control,
have more stringent performance requirements on the sensors and transmitters that
necessities additional testing. Some typical test items are discussed as follows.
The location and connection of all required transmitters for control and safeguard-
ing should be checked against the specification. The impulse lines of all flow and
pressure transmitters should be reasonably short not to increase the signal response
time unnecessarily. All the impulse lines are self-draining toward process tapping
and with no low pockets.
All transmitter tags shall have the same range configured as in the field. For DP-
based flow transmitters, ensure that the transmitter range in pressure (mBar or kPa)
matches the range in mass flow (kg/s). See flowmeter design in Sect. 6.2.4.
A five-point check (0, 25%, 50%, 75%, and 100% of the flow measurement range)
can be performed using a pressure calibrator with the help of an instrumentation
engineer. The instrumentation engineer injects pressure signals into the DP transducer
at the transmitter in the field. Inside the control room, the process control engineer
verifies that the flow indications in the DCS reading match the expected values. For
instance, assume the raw DP value from the sensor in the field is 0 ∼ 40 kPa, and the
measurement range of the flow indicator in DCS is between 0 and 25 kg/s. The flow
signal in DCS should match the DP signal at the transducer, as shown in Table 8.2.
2The relevant standards governing field tests include ASME PTC 10-1997, “Performance Test Code
on Compressors and Exhausters,” and ISO 5389, “Turbo-compressors-Performance Test Code.”
Table 8.2 Differential pressure from DP sensor and flow rate in DCS
Output of DP sensor (=P) kPa 0.0 10 20 30 40
DP Signal (Percent-in-Range: 0~100) % (0%) (25%) (50%) (75%) (100%)
Input to AI if square-rooting at transmitter mA 4 12 15.3 17.86 20
Input to AI if square-rooting at AI mA 4 8 12 16 20
Output of AI (=mass flow) kg/s 0 12.5 17.68 21.65 25
Indicated Flow (Percent-in-Range: 0~100) % (0%) (50%) (70.7%) (86.6%) (100%)
Table 8.3 Flowmeter testing and calibration

Pressure Injected 0.0 25% 50% 75% 100%
Field Signal 4 mA 8 mA 12 mA 16 mA 20 mA
Pressure Signal Received 0.0 25% 50% 75% 100%
For DP-based flow meters, the raw measurement is the differential pressure3 (P,
in kPa or mbar), and the expected measurement indication is flow (Fm , in kg/s). From
DP to flow, a square-root transformation must be performed either in the transmitter
or inside DCS (e.g., in the AI block). It is crucial to verify that this square-rooting
is performed once, and only once, at the specified location (transmitter or DCS).
If the square-rooting is performed at the transmitter, the analog input value to the
AI module in DCS should match the third row in Table 8.2. On the other hand, if the
square-rooting is performed on the DCS, the input to the AI module should match
the fourth row. If square-rooting is not performed at all or done more than once (e.g.,
at both the transmitter and AI), the flow readings will not match the last two rows,
except at zero and full range.
All stages’ suction and discharge pressure transmitters should be checked with
either a pressure calibrator (if possible, depending on pressure range) or a 4–20 mA
signal generator from the field. The resulting pressure indication in the DCS shall
match the values in Table 8.3 within a reasonable error range, e.g., less than 0.5%.
The performance of final control elements, especially the anti-surge control valve,
can significantly impact the control system’s performance. Stroke testing is a manda-
tory step to verify the adequacy of the surge control valve during SAT. Stroke tests
on the anti-surge control valve can be conducted with a 4 ∼ 20 mA loop calibrator
for the following valve, focusing on the below aspects:
• Fully open. As a fail-open valve, the anti-surge control valve shall be fully open
when a 4 mA input is injected.
• Fully close. The valve shall be fully closed if a 20 mA signal is supplied.
3One reason to prefer square-rooting in DCS is that both the raw DP and inferred flow values are
accessible from DCS.
Table 8.4 Valve stroking

from field Input Signal 4mA 8mA 12mA 16mA 20mA
Valve Position 100% 75% 50% 25% 0
• Speed of opening. The valve opening speed is critical for anti-surge control. Inject
a 20 mA signal to cause the valve to close fully and then quickly disconnect the
signal. The valve should reach a fully open position in less than two seconds.
• Speed of closing. The closing speed is not as critical as the opening speed but
typically shall be no more than 10 seconds from a fully open position to close fully.
The recycle valve is shared by capacity and anti-surge control, with the former
requiring a fast-open characteristics and the latter demanding a linear behavior.
The closing speed can be verified by starting with a fully open position (input ≤4
mA), then quickly increase the signal to above 20 mA, and observe the time the
valve takes to go to the fully closed position.
• Valve positioning and non-linearity. The recycle valve shall be able to position itself
correctly as required. Excessive offset or overshoot (e.g., >5%) would degrade
the control performance and is unacceptable.
The test on the positioning performance can be conducted as follows: inject input
signals in the order of 4 mA → 8 mA → 12 mA → 16 mA → 20 mA and observe
whether the valve can be positioned quickly and to a reasonable accuracy (~2%)
as indicated in Table 8.4.
Then repeat the test in the opposite direction, i.e., 20 mA → 16 mA → 12 mA
→ 8 mA → 4 mA, and verify that the valve openings match the table from fully
closed to fully open.
• Deadband or dead angle. The anti-surge control valve is typically fully closed
during normal operating conditions. It takes time to lift the valve stem from its
seat before the valve starts to open. This delay is called the deadband, and excessive
deadband slows down the valve opening. In case a surge condition is detected, the
valve must open quickly, and the deadband is expected to be minimum.
To test the deadband, put the valve in the fully closed position (input ≥20 mA),
gradually reduce the signal, and record the signal value that causes the valve to
start to open. Usually, this deadband should be less than 5%, i.e., the valve should
start to open before the signal goes below 19 mA. In terms of time, the valve should
not take more than 0.5 seconds to move from the full thrust on the seat with a step
down of the signal from 20 mA.
• Step resolution and hysteresis. For capacity control, a sticky valve degrades the
control performance. The stiction is expected to be under 5%. The step resolution
can be tested as follows:
– Put the valve at 50% open (input = 12 mA).
– Slowly increase the signal input, and observe how much signal change is needed
before the valve move can be observed.
– Do the same by slowly decreasing the signal.
8.3 Plant Startup 227
Similarly, to test the amount of hysteresis, start with the valve at 50% opening
(input = 12 mA), slowly increase the signal input to 75% (input = 8 mA), and
wait for the valve to settle. Then slowly decrease the signal input and observe how
much signal change is needed to cause the valve to change direction.
• Overshoot. Overshoot is detrimental to valve performance. Typically the overshoot
in the valve position should be less than 3% of the steady-state change when
responding to a step change in the input signal.
8.3 Plant Startup
The initial plant startup is critical and has a high risk of equipment damage if
improperly operated. The plant startup requires the participation of multiple par-
ties, including the machine manufacturer, operations, rotating equipment engineer,
process engineering, and process control.
8.3.1 Incipient Surge Test
At the commissioning phase, the compressor has been shipped to the site, installed,
and connected to the rest of the plant. The compressor performance data and the
anti-surge control, which are based on simulated data from the manufacturer, must
be validated with actual operating data from the field. The incipient surge test serves
this purpose by pushing the operating point as close as possible to the surge limit to
validate and improve the previously estimated surge points and the surge reference
line.
From a process control perspective, the objective of the surge test is to evaluate
the accuracy of the anti-surge parameter (ASP), which by far has been exclusively
based on predicted performance data. The incipient surge test is the first opportunity
to validate the ASP with actual measurements and is thus very important.
The surge test is almost always required as part of the commissioning activities for
a new compressor installation. The surge test is typically performed with the machine
in total recycle mode. The three performance lines: surge control line (SCL), surge
reference line (SRL), and surge trip line (STL), in that order, are verified with the
tests.
Since the process gas may not be available, the surge tests are usually conducted
with nitrogen or fuel gas. Although these gases differ from the actual gas, the test
results are unaffected when viewed under invariant coordinates (see Chap. 5).
1. Sanity check of the SCL. The surge control line is typically 10% to the right of
the SRL on the performance map. The SCL sanity check is to prove that when
the compressor operates at SCL, no evidence of surge incipience is observed.
The test can be performed as follows. With the anti-surge control in automatic
mode, slowly reducing the process to cause the actual ASP value to decrease,
approaching the surge control setpoint (typically at 33%). Once the ASP value
goes below the surge control setpoint value, the output of the anti-surge controller
should start to open the anti-surge control valve. The compressor must maintain
the ASP value at 33% as the process flow decreases.
A sanity check on the surge control line is necessary but not sufficient since the
distance to surge cannot be verified based on this test alone.
2. Sanity check of the SRL. The test intends to validate and improve the accuracy
of the actual SRL, which is instrumental to the success of both capacity control
and anti-surge control.
To verify the surge reference line, continue from Step 1. Keep the anti-surge
control in automatic mode and maintain a sufficient level of recycle flow, e.g.,
30% of valve opening. Slowly reduce the surge control setpoint to force the
operating point to move further left to approach the SRL.
During the surge test, closely monitor any sign of surge incipience. Stop the map
exploration immediately to avoid entering an actual surge if an incipient surge is
observed. Typical signs of surge incipience include vibrations in the machine and
oscillation in the flow and pressure measurements. Once observed, the machine
flow needs to be quickly increased to prevent the machine from experiencing an
actual surge. If the ASP value at the point of an incipient surge is in the vicinity
(e.g., 2% of SRL, or ASP between 30% × (1 ± 2%) = 30 ± 0.6%) of the SRL,
it can be concluded that the ASP calculation is adequate.
Suppose surge incipience is noticed much earlier or later than expected (relative
to the SRL). In that case, it indicates that the ASP calculation deviates too much
from the actual performance. The vendor-provided performance curves should
be reviewed and validated. If necessary, the ASP calculation must also be re-
examined and improved.
This sanity check on the surge reference line is typically both necessary and
sufficient. However, note that the field environment typically cannot be controlled
as ideally as in the factory acceptance test environment due to difference in field
installation and gas property; the test results have more uncertainties and should
be interpreted with discretion.
3. Testing the STL. Testing the surge trip logic is part of the acceptance test. This test
is intense since the compressor is intentionally moved beyond the manufacturer-
provided limit to cause the machine to trip on an actual surge.
The surge trip test is an excellent opportunity to further confirm the SRL via an
actual trip event. Discuss with the commissioning engineers to agree on using the
anti-surge controller to initiate the trip event.
After the compressor is started successfully and running stably, gradually reduce
compressor speed. The capacity control and the anti-surge control should be able
to open the recycle lines to protect the compressor from a surge.
The surge trip line is usually set at half the SRL, i.e., 15%. If not, lower the
setpoint even more, and try again. The compressor flow can be reduced by slowly
decreasing the anti-surge control setpoint of a selected stage to below 30% and
testing whether the compressor would go to a surge trip. If the compressor load
is high and the anti-surge valve remains closed, it may be necessary to reduce the
process flow to force the compressor to surge.
8.3.2 Pre-startup
The pump and compressor are usually started from the DCS console in the central
control room via carefully designed logic and sequences. Testing on the startup
sequence involves several parties’ joint efforts, including process control. Some
required preparation for machine startup relevant to process control includes the
following:
• Verify that all solenoid valves on the anti-surge recycle line are de-energized, and
cross check that all anti-surge control valves are fully open.4
• Check and ensure all controllers and calculation blocks are in the correct operating
mode. For example, all anti-surge controllers, pressure protection controllers, and
power constraint controllers are in AUTO mode, and the load controllers are in
MAN mode.
• Check and ensure that all control setpoints are set at the correct values. For first-time
startups, set all anti-surge control setpoints to higher, more conservative values,
e.g., 40% instead of 33%. This higher setpoint will cause the machine to recycle
earlier, but it is safer for the initial test.
• From the control overview screen on DCS, check and make sure that the process
values of the suction/discharge pressures and flows are consistent and make sense.
For example, before the compressor is started, all the suction and discharge pres-
sures shall be approximately the same because of pressure equalization. When the
motor is started, the suction and discharge pressures shall increase in ascending
order from the first stage to the last.
• Check and ensure that the values of the corresponding suction/discharge pressures
and flows (PZ and FZ) for safeguarding match their corresponding readings (PI
and FI) for control.
8.3.3 During Startup
The automated startup procedure is typically initiated by pushing a button from the
DCS screen. The compressor speed is ramped up quickly to reach the minimum
stable speed, then slowly increases to reach the desired operating range.
• During initial speed ramping, closely watch the pressures and flows of all stages
as they increase. The ASP values calculated independently in DCS and SIS should
4Because of de-energized solenoids, the fail-open recycle vales should be all in the fully open
position.
decrease in tandem, starting from 100% and slowly decreasing as the recycle valve
is closed.
• Watch and verify that all the anti-surge control valves (ASCV) close at the rate set
by the rate of change limit (e.g., 5%/min). The ASCV will continue closing until
the ASP values reach their setpoints. Depending on the load, the anti-surge control
valves may go fully closed (when the load is high) or partially closed (when the
load is low).
• Manually and slowly increase the speed to the desired load (max 100%). Closely
watch all the pressures, flows, and ASP values.
Figure 8.3 shows the startup sequence of a variable-speed compressor in terms of
the machine speed, recycle valve, and surge parameter. The speed change is divided
into multiple steps as follows:
1. At a certain point of the startup sequence (T=0), a compressor LOAD signal is
switched to active (=1).
2. Upon receiving the LOAD signal, compressor startup logic quickly ramps up the
machine speed to above the minimum stable speed, e.g., from 0% to 75% within
45 seconds.
3. When the speed exceeds the minimum stable speed (e.g., 75%), the solenoid
valves on the recycle lines for all stages are energized. The startup logic ensures
that all anti-surge controllers are in AUTO mode.
4. Anti-surge controller automatically starts closing the anti-surge control valves
because the ASP value is above the control setpoint.
5. The startup logic continues to ramp the machine speed to normal operating speed,
e.g., 95%, but at a slower rate, e.g., 1%/min increase.
6. After the machine speed reaches 90% and anti-surge controllers reach the steady
state, switch all load controllers to cascade modes for automatic capacity control.
The typical startup sequence for a fixed-speed centrifugal compressor is more
straightforward, as illustrated in Fig. 8.4.
8.3.4 Post-startup
The control scheme, especially the PID configurations, should be fine-tuned to

improve the performance after the compressor successfully starts and reaches a stable
operational condition. These changes may include the following:
• Controller response. For each anti-surge controller, increase the controller setpoint
above its process value (PV) to force the controller to open the recycle valve.
Monitor the transient response in the controller PV.
• Controller fine tunings. Based on the transient behavior of the controllers, adjust
the PID tuning parameters to improve the control performance.
• Interaction between stages. When the anti-surge control valve opens in one stage,
watch the surge parameters of other stages for inter-stage interactions. Watch
Fig. 8.3 Startup sequence of a variable-speed compressor
Fig. 8.4 Startup sequence of a fixed-speed compressor

how fast the anti-surge controller opens the recycle valves to maintain the surge
parameter on the setpoint. Re-tune the controllers as necessary.
• Discharge pressure constraint controllers. Temporarily lower the controller SP to
below its PV and watch the response of the controller output and the control valve
opening. Fine-tune the controller to achieve the desired speed of response.
• Power constraint controller. Similarly, if the motor power constraint controller is
implemented, temporarily decrease the setpoint to watch the decrease in machine
speed.
Step tests may be needed to introduce dynamic changes in the interested process
variable to understand the process dynamics better.
The process control engineer must update the process control narratives (PCN)
document by incorporating any significant changes made during commissioning and
startup. The status of the PCN document will change to “as-built” accordingly.
8.4 Performance Monitoring
Health and performance monitoring is through various process variables and perfor-
mance indicators. Many operating issues can be detected and avoided before they
become costly. A thermodynamic analysis is the most straightforward and practical
approach to determining performance issues based on key control variables and key
performance indicators.5
The key performance indicator (KPI) is typically a derived variable that provides
a reliable indication of the performance in the concerned area. The KPI must be
intuitive, reliable, and comprehensive to help higher level decisions.
8.4.1 Performance Indicators
There are many process measurements around the pumps and compressors for dif-
ferent purposes. Online control is based on fast and reliable measurements. Online
monitoring may require additional measurements. For each process flow configu-
ration, a group of critical variables for each compressor can provide a reasonably
accurate description of the status and condition of the machine and should be con-
tinuously monitored.
1. Pressure, flow, temperature, and speed. The directly measurable process variables
usually include pressure, flow, temperature, and speed. The changes in these
measurements are governed by physical laws and are expected to be smooth
under normal operating conditions. Any significant abrupt changes in the signal
5 Mechanical and electrical issues are primary concerns for online monitoring but are out of the
scope of process control discussed in this book.
8.4 Performance Monitoring 233
Fig. 8.5 Control handles for centrifugal compressor control
profile may indicate an abnormal condition. For example, oscillation in flow and
pressure may indicate an incipient surge. Persistently high discharge temperature
may indicate decreased efficiency.
The type and location of the measurements required by online monitoring are
shown in Fig. 8.5.
2. Head. The polytropic head is one of the three basic variables describing pump
and compressor performance (see Sect. 2.2.1). If the fluid head shows a signif-
icant decrease for the same machine speed and flow rate, it may indicate that
the machine is experiencing increased energy losses. The machine’s efficiency
should be evaluated for more insights.
As discussed in Sect. 5.1, the head cannot be directly measured but can be indi-
rectly calculated from the pressure ratio as follows:
n−1
n Z s R Ts Pd n
Hp = −1 (8.1)
n−1 M Ps

Pd Pd Ts
n = ln ln · . (8.2)
Ps Ps Td
The gas molecular weight or density is needed to calculate the fluid head. They
are available only offline; thus, monitoring the fluid head is typically based on
both online measurements and offline gas property information. The machine
should be inspected if the calculated head is significantly lower (e.g., by 10%)
than expected from the manufacturer-provided performance curve.
3. Efficiency. The overall efficiency, including the mechanical and hydraulic effi-
ciencies, is affected by many factors and is complicated to calculate accurately.
Polytropic efficiency, on the other hand, concerns the thermodynamic behavior

inside the machine and can be calculated from operating data as

Hp κ −1 n−1
ηp = = (8.3)
Hactual κ n

Pd Pd Ts
n = ln ln · , (8.4)
Ps Ps Td
where the ratio of specific heat κ is known from offline gas analysis, and other
variables are from online measurements. A persistent drop of more than 2% in
polytropic efficiency η p is usually an indicator of compressor performance issues,
e.g., fouling.
4. Motor power and gearbox torque. The motor and gearbox’s condition must also
be monitored as part of the process configuration. If available, motor power and
gearbox torque are important variables to monitor online. Torque6 can be an
essential variable for monitoring and protective control. For example, multiple
incidents have been reported where the required torque exceeds the gearbox limit
and causes severe damage to the gearbox.
5. Anti-surge parameter (ASP). Process measurements provide an isolated view
into the status or health of a specific variable. However, most sophisticated per-
formance indicators, such as surge indicators, are based on multiple variables and
backed with physical insights. Surge is the most critical abnormal operating con-
dition for a centrifugal machine that needs to be monitored. When a compressor
approaches the surge condition, one or more symptoms will be observed, such as
a sudden decrease in flow, excessive vibrations, and loud noises. However, relying
on these symptoms or individual variables to monitor and detect surges is inaccu-
rate and untimely. The ASP is derived from multiple field measurements based on
engineering insight and is more effective for monitoring the surge phenomenon.
6. Power consumption. Another KPI of interest is power consumption. The actual
power consumption (apparent power) can be directly measured with a watt-meter.
The gas power, the useful power transferred to the fluid, can be calculated from
the polytropic head and flow rate.
⎧
⎨Power = [Polytropic Head] × [MassFlow
⎪ Rate]/[Polytropic Efficiency]
n−1
H p · Fm Fm n Ps Pd n
⎪
⎩W = = −1 .
ηp η P n − 1 ρs Ps
The power consumption against throughput provides another overall performance

indicator of operating efficiency.
6 A torque meter can provide additional information for verifying the compressor performance but
is not widely installed for various reasons.
8.4 Performance Monitoring 235
8.4.2 Performance Visualization
Visual presentation of the compressor status and health is an intuitive and powerful
tool for performance monitoring, especially for the operators. Performance moni-
toring should include the actual measurement variables, the inferred performance
indicators, and the control scheme. The focus can be on the real-time status, the
machine’s health, or process unit’s operating efficiency.
1. Control overview diagram. An overview diagram on one screen provides an at-
a-glance summary of the critical variables and KPIs, offering the operators and
engineers a holistic view of the status and performance for startup, operation,
and troubleshooting. The overview page can be based on the control overview
diagram (COD) from design, such as Fig. 7.15, with an appropriate color scheme
to indicate the different status and condition, e.g., red color or flickering numbers
to indicate an abnormal condition.
2. Live performance map. Modern DCS natively supports schematics, trending, and
other HMI functionalities for general monitoring. For pumps and compressors,
a live performance map is becoming a standard requirement for compressor per-
formance monitoring. Live performance maps show the compressor’s operating
point in relation to the surge line. Figure 8.6 is a real-life example of a compres-
sor circuit with four parallel compressors, each with four stages. The four slanted
lines are, from left to right, the surge trip line, the surge reference line, the surge
control line, and the capacity control line (see Sect. 4.3.5 for definitions).
The performance map is used as a backdrop image on the monitoring screen. The
real-time pressure ratio and equivalent flow are mapped from the engineering
units to screen pixels. The anti-surge parameter is calculated online and superim-
posed on the performance map. In this example, since all four compressors have
identical designs, the four stages of each compressor share the same performance
map. For instance, the first chart shows the performance map of the first stage of
all four compressors, with the four real-time operating points superimposed on
the performance map.
The control solution is designed with load balancing among all the stages and
compressor trains (see Chap. 7), all stages are balanced with the same surge
margin.
3. Controller performance monitoring. Controller performance monitoring (CPM) is
becoming indispensable in a modern operating plant. CPM continuously monitors
and assesses the control loop’s status and performance and promptly alerts the
concerned parties for abnormal conditions. The control scheme for pumps and
compressors is part of the overall process control solution. All the controllers
should be included in the CPM for continuous monitoring. The primary concerns
include whether the controller operates in the intended modes (such as automatic
and cascade) and whether the controlled variables track the desired setpoints
satisfactorily.
The status and performance of a control loop are calculated based on controller
measurements or properties: controller mode (Mode), setpoint (SP), process value
(PV), and controller output (OP).
Fig. 8.6 Real-time monitoring of the operating point
Through CPM, the up-time and performance of the controllers, excess moves of
the control valves, and frequency of operator interventions on control setpoints
or outputs can be closely monitored to detect potential problems.
8.5 Troubleshooting and Problem-Solving
Troubleshooting is always a challenging task and demands the most knowledge and
experience. Since operation and control problems are highly unpredictable, there is no
fixed recipe for troubleshooting. Nevertheless, there are some general methodologies
and best practice procedures to follow (King 2016; Niu and Xiao 2022). A few
particular aspects are given in this section.
8.5 Troubleshooting and Problem-Solving 237
8.5.1 † Surge Detection
Compressor surge control is one of the most critical and challenging tasks for trou-
bleshooting. With a sound surge prevention scheme, the machine is not expected
to enter surge conditions during normal operation. If a surge condition is detected
or reported due to abnormal conditions, the critical first step is determining if it is
genuinely a surge and what has caused it. Typical symptoms of surge may include
the following:
• Radial vibrations and axial thrust displacements.
• “whooshing” noises or loud “bangs.” The noise can originate from the collapse of
the gas voids within the compressor, the flexing of the suction filter walls, or the
slamming shut of the non-return valves (McMillan 1983).
• Flow fluctuations and oscillations. A deep surge starts with a sharp drop in the
flow rate that is physically impossible for any other process reason. The flow can
precipitously drop from the current operating point to zero (or even negative) in
tens of milliseconds. Fluctuating pressures at suction and discharge can also be
observed.
• High gas temperatures. During a surge, the flow reversal can occur multiple times
per second and not be cooled, the temperature increase can be rapid and thermal
expansion can cause significant damage.
The surge phenomenon is generally the result of one or more of the following
deficiencies or events:
• Inappropriate compressor design.
• Inadvertent loss of machine speed.
• Malfunction of the instrument, control valve, or guide vane.
• Poor matching of the compressor to the process requirements in capacity, type of
gas, operating pressure, or temperature.
• Restrictions in the gas flow, such as blockage, internal plugging, and fouling, either
at suction or discharge.
• Abnormal operating conditions, such as severe capacity turndown, startup, shut-
down, and crippled operation.
• Inappropriate distribution of load among multiple compressors.
• Changes in the operating conditions such as pressure, temperature, and gas com-
position (see fluid density).
• Unfavorable arrangement of piping and process components of the system that
potentially magnify surge.
• Inadequate anti-surge control system.
It is thus crucial to identify the cause of the surge and rule out non-control issues
before investing time in troubleshooting the control scheme.
8.5.2 † Troubleshooting of Surge Control Scheme
Troubleshooting a control scheme requires the same knowledge as designing a control

solution (Perez and Conkey 2016). As compressor control is based on feedback
control, the general troubleshooting procedure is the same as with any other feedback
control schemes, i.e., focusing on the five components of the control loop: process
dynamics and machine characteristics, the process measurements, the final control
elements, the control objectives, and the control algorithm.
Process dynamics refers to both the machine characteristics and the process
dynamics. The first question to ask is whether there have been any recent modifica-
tions to the process or machine that may change in the cause-and-effect relationship
or the transient response behavior of the process variables.
Process measurement problems are relatively easy to detect and fix. The most
common instrumentation for machine control systems includes flow and pressure
measurements. Common problems include blocked impulse lines, condensation, and
processor failures in the instrument. For instance, the transmitters must be installed
above the process piping for gas applications. The impulse tubing length must be
short7 (shorter than the API RP 55 recommendation of 3 m) and with a rising slope
of at least 30◦ to allow for free drainage of any condensate that may form within
the tubing. Bends and elbows must be minimized to avoid leaks caused by tubing
bending.
Measurement provides a window of view into the process status and condition.
Incorrect measurement can result in erroneous control action. For instance, one com-
mon problem with flowmeters is with calibration. An orifice-based flowmeter mea-
sures the pressure differential (P) across the orifice and deduces the flow rate. To
infer the flow from the differential pressure requires an accurate representation of the
relationship between them, which is ensured via regular inspection and calibration.
Troubleshooting measurement problems can be based on spatial and causal infor-
mation. For instance, the value difference in redundant measurements of the same
variable should not exceed a certain threshold (spatial information). A flow through
the compressor cannot change without a corresponding change in the pressure ratio,
consumed power, or rotating speed (causal information) (Elliott and Bloch 2021; Niu
and Xiao 2022).
Control valves and machine speed are the final control elements (FCE) in compres-
sor control, with the anti-surge control valve being the most critical. The performance
of the anti-surge valve significantly impacts the control performance. Problems with
the anti-surge control valve typically include the following:
• Valve sizing. Too large or small a control valve is a problem for control perfor-
mance.
• Valve characteristics. The recycle valve serves the purpose of surge avoidance
and is expected to be fast-open for anti-surge control and linear in characteristics
7 Long impulse lines contribute to signal noise, as the gas resonance phenomena in the impulse
tubing can amplify pulsations in the pipe (Elliott and Bloch 2021).
8.6 Summary 239
for capacity control. A good compromise is required to meet the two conflicting
requirements.
• Response speed. The valve is expected to open quickly if a surge condition is
detected. Over time, the valve response may become slower or even stuck due to
increased friction. Regular inspection and maintenance are essential.
• Positioning accuracy. A valve is expected to move to a new position quickly and
accurately without much delay or overshoot. Inaccurate positioning can degrade
the control performance.
A recycle valve can become “stuck” in one position due, most likely, to increased
stiction or debris buildup in or around the valve seat. A “stuck” valve is a serious
problem for control. The longer the valve remains in one position, the higher the risk
of the valve becoming stuck. The risk can be mitigated by periodically sending a
test command signal to the valve (Elliott and Bloch 2021). The test signal may be a
single step, a series of steps, or a ramp. The perturbation should only be introduced
when the system is in steady-state condition.
The next component in the feedback control loop is the control setpoint. For
instance, the surge control setpoint is determined during design and rarely changes
during operation unless the control design is revamped. However, the controller
setpoint may have been tampered with, driving the controller to a different operating
condition.
The last component is the control algorithm. The typical issue is with PID tunings.
However, unless there are significant and permanent changes in other components
in the control loop, the control algorithm and PID tuning typically do not need to
change.
Mechanical troubles are outside the process control scope. However, there are
some occasions where process control can positively contribute to the troubleshooting
effort from a unique perspective.
Any changes to the control scheme must follow the proper management of change
(MOC) procedure and be captured in the process control narrative (PCN) document.
8.6 Summary
Commissioning, startup, operation, and maintenance are the critical steps to turn a
sound control design on paper into a functioning application in the field. These steps
require contributions from various disciplines to ensure the control implementation
aligns with the design specification, performs as expected, and is accepted by oper-
ation personnel. Process control is essential throughout these steps to guarantee the
control solution is satisfactory.
During normal operation, constant monitoring and continuous improvement are
essential for the control solution’s sustained performance, relying on a thorough
comprehension of the process, equipment, and the process control solution.
The most knowledge of the equipment and control solution is required to trou-
bleshoot and correct operational problems, and practical experience is crucial; there
is no substitute for it.
References
Brun K, Nored MG (2006) Guideline for field testing of gas turbine and centrifugal compressor
performance. Technical report, Gas Machinery Research Council, Southwest Research Institute
McMillan GK (1983) Centrifugal and axial compressor control. Instrument Society of America.
Perez RX, Conkey AP (2016) Troubleshooting rotating machinery — including centrifugal pumps
and compressors, reciprocating pumps and compressors, fans, steam turbines, electric motors,
and more. Scrivener Publishing and Wiley
Appendix A
Performance Analysis and Control Design
with Software Tool
Following the descriptions in this book, it is possible to analyze the compressor

performance and design the control solution with the traditional pencil-and-paper
approach, as shown in Sect. 5.4. However, the calculation becomes overwhelm-
ingly complex for modern compressors with more stringent operating requirements
and large amounts of data. Software tools are needed to facilitate all the activities,
from data entry and visualization to performance analysis and anti-surge parameter
calculation. This chapter provides an overview of the common activities related to
performance analysis and control design, using the companion CPACS software tool.
A.1 Introduction
Process control for pumps and compressors includes both capacity control and anti-
surge control (or minimum flow control for pumps). Anti-surge control is the most
crucial component due to its criticality to equipment safety. It is also because the
compressor characteristics is significantly different from that of static equipment
that we are familiar with (King 2016; Niu and Xiao 2022). A software tool can
help understand machines’ characteristics and behavior and facilitates the control
solutions’ analysis and design.
A.1.1 Data Requirements
A surge indicator like the anti-surge parameter (ASP) is developed with offline design
data at design time and then used with real-time measurements during online opera-
tion. Collecting and analyzing these data is the first step of control design, requiring
a good understanding of the compressor characteristics (Chap. 2).
© The Editor(s) (if applicable) and The Author(s), under exclusive license 241
to Springer Nature Switzerland AG 2023
https://doi.org/10.1007/978-3-031-43122-7
242 Appenix A: Performance Analysis and Control Design with Software Tool
The data required by anti-surge control design typically include the following, as
shown in Table 6.4:
1. Compressor API design data. The compressor API datasheet defines the desired
operating scenarios under which the compressor is expected to operate, such as
design case, summer case, winter case, light gas, and heavy gas. Finalizing the
design datasheet may require several interactions between the end user and the
machine manufacturer.
2. Compressor performance data. The compressor vendor designs the compressor to
meet the requirements specified in the API datasheet. The performance curves are
the characteristic behavior predicted by the compressor manufacturer, confirmed
by performance tests at the manufacturer’s facility and incipient surge tests at the
site. A set of performance curves are provided for each design case. Figure 2.14 is
an example of the performance curve in the traditional chart format. An example
of the API datasheet can be found in Table 5.8.
3. Flowmeter data. The flowmeter datasheet defines the requirements of flow mea-
surement. An example of the flowmeter datasheet is given in Fig. 6.31.
4. The valve data. The valve datasheet provides the characteristics of the control
valve. Of importance to the anti-surge parameter calculation is the valve constant.
5. Real-time operating data. The ASP calculation aims to provide the surge indicator
for online anti-surge control. Once online measurements are available, they can
be collected and visualized on the performance map.
It can be a painful experience to “excavate” the required data for an existing
compressor that has been in operation for many years with incomplete or inaccurate
documentation. Familiarity with the work process and data requirements is essential
to identify and validate the data.
A.1.2 The Work Process
The compressor control design is typically a part of a larger project, either green
field or revamping. The anti-surge parameter calculation is the key effort of anti-surge
control, lasting from the beginning to the end of the project. The work process is thus a
progressive advancement of data collection, validation, processing, and continuous
improvement. Consequently, one crucial interaction between process control and
other parties in the project team is the data exchange, either as design input or
deliverable.
For a green-field project, the data availability is progressive as the project pro-
gresses. It is also very interactive between the design team and the data providers.
The data requirement and inter-dependency are shown in Fig. A.1.
Appendix A: Performance Analysis and Control Design with Software Tool 243
Fig. A.1 Data requirements for anti-surge parameter calculation
Fig. A.2 CPACS main menu
A.1.3 The CPACS Software Tool
CPACS, for compressor performance analysis and control support, is an Excel1 add-
in under the Windows operating system. The CPACS tool is freely downloadable from
http://github.com/niucontrol/CPACS, along with the user manual. It aims to be a
simple yet versatile engineering tool for process control engineers to enter, manage,
and visualize compressor data; analyze the compressor control performance; and
calculate the anti-surge parameters (ASP) for anti-surge control.
Users are expected to have a basic understanding of thermodynamics, compressor
characteristics, and control theory, as discussed in Chaps. 1 to 4 of this book, to
effectively use this tool.
To start CPACS, open the Excel file containing the CPACS file, and the CPACS
menu should appear in the menu bar.
The current version of CPACS has the following functions (see Fig. A.2):
• Graph digitizer. For older compressors, performance data were typically provided
by compressor vendors as performance curves. The performance data need to be
“read out” from the curve. Graph digitizer facilitates this process.
• Data analysis. This function provides all the functionality related to data visual-
ization, validation, to anti-surge parameter calculation.
1 Microsoft Windows® and Microsoft Excel® are registered trademarks of Microsoft Corporation.
A.2 Data Preparation and Management
The first step in anti-surge control design is collecting the required data. Depending
on the data source, this step is often the primary challenge and error source for ASP
calculation with a significant impact on the online control performance.
A.2.1 Data Format
Five types of compressor data can be entered into CPACS. Some are required, while
some are optional.
1. Reference Condition (1REF). The reference condition is typically the same as

the flowmeter calibration condition, which is recommended to match the certified
condition (see Sect. 5.1) in the compressor API datasheet. If the flowmeter is
installed on the discharge side, the reference inlet condition should use the certified
condition from the compressor API datasheet as the suction condition.
The reference condition is entered as one row. The first column shall be filled as
1REF (for reference condition). The second column is not used and can be left
blank. The reference condition includes the following:
• SPEED. Compressor speed.
• MW. Gas molecular weight.
• P1. Suction pressure.
• T1. Suction temperature.
• Z1. Gas compressibility at suction.
• P2, T2, Z2. Discharge condition if the flowmeter is on discharge. Leave them
blank if the flowmeter is on the suction side.
• FM. Maximum mass flow in kg/s as the flow measurement range.
• DP. Maximum differential pressure (DP Measurement Range), corresponding
to (i.e., matching) the measurement range of the mass flow.
• K1, K2. The ratio of specific heat (κ = Cp /Cv ) for suction and discharge
(Optional).
• CV. Recycle valve flow coefficient in gpm. If the CV value is provided, the
maximum flow line for the surge valve can be plotted on the compressor map.
2. Operating Limits (2LIMIT, optional). If applicable, the design constraints and
operating limits can better define the operating envelope, allowing the anti-surge
calculation to approximate the surge line better. Some common operating limits
that may affect the operating envelope include the following:
• Compression ratio (P2/P1). For a centrifugal compressor, the compression
ratio is typically between 1.5 and 4.0. Although vendor data may cover a wider
range, a compressor can rarely operate outside this range.
• Discharge pressure (P2): In actual operation, there is almost always a high-

pressure trip setting on the discharge pressure to protect the equipment and
piping. This trip limit implies that the operating point can never reach this area.
That area can be excluded from the calculation to improve the design accuracy.
• Motor power (POWER): For compressors driven by an electric motor, a certain
region of the compressor operating envelope may never be reachable due to the
power limit of the motor and should be excluded from the ASP calculation.
The limits can also be specified on any other parameters. For example, to exclude
the performance data above 100% speed, a value of 1.0 can be specified.
3. Design operating conditions (3DESIGN). The design condition derives from the
compressor API Datasheet. They specify the different operating scenarios under
which the compressor is expected to operate.
In CPACS, each scenario is entered as a row. The first column in the row shall
be filled as 3DESIGN (for Design Point). The second column is the name of the
scenario or case (e.g., CASE A, SUMMER). Other required data for each scenario
include the following:
• MW, Molecular weight (kg/kmol).
• P1, Suction pressure (kPaa).
• T1, Suction temperature (Kelvin).
• Z1, Gas compressibility at suction (-). Typically gas compressibility has a value
less than but close to 1.0.
• P2, Discharge pressure (kPaa).
• T2, Discharge temperature (Kelvin).
• Z2, Gas compressibility at discharge. Normally between 0.9 and 1.0. At very
high pressure, the value can be higher than 1.
• FM, or FV1. Mass flow or volume flow at suction. Typically either mass flow
or volume flow should be specified. If both are provided, cross validation is
automatically performed.
The following data are optional. However, if available, they should be entered as
they provide redundant information for CPACS to cross validate the data.
• SPEED, compressor speed.
• FV1, volumetric flow at suction.
• K1, ratio of specific Heat: optional for CPACS.
• K2, ratio of specific heat at discharge.
• HEAD, polytropic head.
• EFF, polytropic efficiency.
• POWER, gas power.
4. Performance data (4PERF). Performance data is available from the vendor as
performance curves (charts) or spreadsheets. For ease of data entry, the spread-
sheet format shall be requested. If the performance data are given as performance
curves, the curves need to be digitized to extract the required performance data.
The data are organized as follows:
• Scenarios (Cases). The scenarios (or cases) are defined in the compressor API
datasheets.
• Speed lines. Each scenario can include multiple speed lines, typically 105%,
100%, 90%, 80%, and 70% of the normal speed.
• Points. Each line includes multiple operating points, the first being the surge
point and the last being the stonewall (choke) point. Between surge and choke
points, there should be at least one point. Typically five points are recom-
mended, and more points will give a more accurate representation of the per-
formance curve.
For each operating point, the following performance data are required:
• MW, molecular weight.
• SPEED, compressor speed.
• P1, suction pressure.
• T1, discharge pressure.
• Z1, compressibility at the suction.
• P2, discharge pressure.
• T2, discharge temperature.
• Z2, compressibility at discharge.
• FM, mass flow rate.
Optionally, the following can be included for cross-validation purposes.
• POWER, gas power.
• HEAD, polytropic head.
• EFF, polytropic efficiency.
• K2, ratio of specific heat at the discharge side.
The discharge condition can be specified either as discharge pressure and tem-
perature or as head and efficiency. The discharge pressure and temperature are
automatically calculated if the polytropic head and efficiency are given (common
for older/existing compressors).
5. 5RTOP: Real-time operating data. Real-time operating points, if supplied, are
superimposed onto the operating envelope. The following data can be provided:
• P1, suction pressure.
• P2, discharge pressure.
• FE or DP, equivalent flow, or differential pressure across the orifice.
A.2.2 Source of Design Data
ASP calculation starts with data entry. In CPACS, click Calculate ASP to bring
up the main interface, as shown in Fig. A.3.
Fig. A.3 ASP calculation menu
For data entry from scratch, run NEW first to create a template. Data is entered
by rows, with each row representing ONE operating point. The data are entered
either via copy and paste from a vendor-provided electronic spreadsheet (preferred)
or manually entered with digitized data from performance curves.
1. Manual data entry. Traditionally, the performance data are provided as “curves”
on paper. The data values must be digitized first.
The digitization process used to be based on pencils and rulers. The desired data
values are extracted from the selected data points on the curve. The corresponding
values are then “read out” and collected. For instance, Fig. A.4 shows the five
points for one performance curve and the corresponding head and flow values.
This manual digitization process is highly tedious and error prone, feasible only
for simple cases with limited number of performance curves.
CPACS has a digitization tool to facilitate this digitization process. The
graphics digitizer menu opens the user interface, as shown in Fig. A.5.
A scanned image (JPG or GIF) of the performance curve can be loaded into the
digitization tool (Step 1). The tool is calibrated by selecting three points on the
two coordinate axes (Step 2). After calibration, clicking on a selected point on
the performance curve will produce the values (Step 3). The values are captured
and saved to the spreadsheet for copy and paste to the CPACS main interface.
2. Batch and automated data entry. Data entry is a tedious and usually the most time-
consuming part of surge parameter calculation. For new projects/installations,
it is strongly recommended that the vendor supply the compressor data (API
datasheets + Performance curves) as electronic spreadsheets (e.g., Excel) to facil-
itate the data entry process.
For instance, the largest compressor application the author worked on was a
multi-billion dollar green-field project comprising 14 compressors with a total
of 18 stages. The API datasheet specifies 19 operating scenarios. As a result,
the performance curves include 19 cases for each compressor stage; each case
has 8 speed lines (105%, 100%, 95%, 90%, 85%, 80%, 75%, and 70%), and
each line has up to 20 operating points. Each operating point has more than 10
attributes (P1, T1, Z1, K1, P2, T2, K2, Z2, Mw, SPEED, FM, FV, POWER, HEAD,
Fig. A.4 Manual digitization of compressor performance curve

Fig. A.5 Digitization of compressor performance curve with CPACS
and EFF). The data entry can have over 4,000 operating points with over 40,000
individual values for each stage. It would require enormous effort to digitize and
enter the performance data into CPACS (or any other control design tool) if only
performance curves were provided. Fortunately, the compressor manufacturer
agreed upfront to supply all the performance data in both curves and spreadsheets.
The vendor data can be readily transferred to CPACS through copy and paste.
Excel macros can be developed/utilized to further simplify this automated data
entry process.
A.2.3 Engineering Units
Compressor data can be provided in a wide range of engineering units. The mixed
use of engineering units is another major challenge for most users. It is beneficial for
the end user to understand the various engineering units commonly used.
Most software tools can handle the unit conversion internally. CPACS uses a set
of base engineering units for internal calculation. Table A.1 is a partial list of the
acceptable engineering units by CPACS. User data can be supplied in units of their
choice.
Table A.1 Engineering units in CPACS

Variable Symbol Base unit Other units
Molecular weight mw kg/kmol
Speed N rpm %
Flow, mass FM kg/s kg/h, T/d
Flow, volume FV m3 /h m3 /s, m3 /d, gpm, cfm
Flow, standard FVS m3 /h m3 /s, m3 /d
Flow, equivalent FME kg/s
Pressure P kPaa kPag, bar(a), bar(g), psia, psig, atm
Pressure differential DP kPa mbar
Temperature T K C, F, R, DegC, DegF
Polytropic head HEAD kJ/kg J/Kg
Efficiency EFF %
Power W kw HP
Density RHO kg/m3 lbm/ft3
A.3 Data Visualization and Validation
Due to the large amount of data, errors in data entry are very common. Although
manual sanity check is essential and irreplaceable, software tools can help identify
many common errors based on some simple validation rules.
A.3.1 Automatic Error Detection
Clicking on LOAD will cause CPACS to read all the entered data into memory and
simultaneously check for common anomalies, such as
• Worksheet format error. A missing value in a specific cell, or a value with an
incorrect data type, will be flagged as an error for correction.
• Obvious data entry errors. For example, a zero or negative value for temperature
(in Kelvin), an excessively large (>1.5) or small (<0.5) value for compressibility,
and a very small value for gas molecular weight (e.g., less than 2 kg/kmol) are
automatically detected with alerts for attention.
• Data points outside the normal operating envelope. For example, a compression
ratio higher than 5.0 is very unlikely to be realistic. A polytropic exponent outside
the normal range (e.g., [1.0 ∼ 2.0]) may also indicate a bad point.
Fatal errors must be corrected before proceeding. The less severe issues are high-
lighted as warnings.
Fig. A.6 Performance curve: head versus flow
A.3.2 Visual Inspection
Supplying performance curves in chart format has been the standard practice by
compressor vendors. Even when the performance curves are provided in electronic
format, having the chart format (typically in PDF) for cross validation is still bene-
ficial.
CPACS provides the convenience of plotting one compressor property against
another. The plots can be used to visually inspect the compressor data for abnormal
values or outliers and cross validate vendor-supplied performance curves. Below are
a few examples of using the CPACS charting function to examine the compressor
performance.
1. Head versus flow plot (Fig. A.6). Theoretically, at a constant operating speed, the
polytropic head of a compressor is a function of the volumetric flow only. From
CPACS, choose the plotting option of HEAD as Y-Axis and FV1 (flow, volumetric,
suction) as X-Axis, and click PLOT MAP.
• The chart should match vendor-provided performance curves (chart format).
This visual comparison is the most straightforward way of cross checking the
correctness of data entry.
• It is expected to see that all the performance lines cluster around each speed
line. Any point that significantly deviates from the clusters may be incorrectly
entered.
• For variable-speed compressors, the design point for each scenario shall appear
on the head-flow map at the appropriate speed line (interpolated). The design
point should show up on the 100% speed line for a fixed-speed compressor.
Fig. A.7 Performance curve: pressure ratio versus DP/P1
Fig. A.8 Performance curve: ratio versus DP/P1 with limits
2. P2/P1 versus DP/P1 plot (Fig. A.7). Choose P2/P1 as the Y-Axis and DP/P1
as the X-Axis. It is expected to see that
• All the surge lines exhibit a shape of straight, forward bending, backward
bending, or S-shape line. No surge points should significantly deviate from this
pattern.
• The choke points (stonewall) should exhibit a pattern similar to the surge point
line. Any point deviating too much from this pattern may be a bad point.
3. P2/P1 versus DP/P1 Plot, with Limits Active (Fig. A.8). If operating limits
are specified (in the 2LIMIT row in CPACS), click on the “with limits” check
box, and redo PLOT MAP. The operating points outside the limits will be invisible
in the new plot.
Fig. A.9 Pressure ratio versus equivalent flow coordinates
A.4 Surge Parameter Calculation
With a software tool like CPACS, the ASP calculation typically becomes straight-
forward. The key is knowing which ASP formula design scenarios will be used for
the calculation.
A.4.1 Anti-surge Parameters Calculation
With CPACS, select the desired equivalent coordinate (see Chap. 5) by choosing the
appropriate Y-axis, X-axis, and the ASP formula (ASP1 to ASP4). Then select any
combination of design cases from the Case list, and click UPDATE ASP, the anti-
surge parameter (ASP) will be calculated with the selected case data. The various
control-related lines can be displayed on the chart by selection from the LINE list
box.
The default plot is the pressure ratio (P2/P1) versus the equivalent flow (FTE),
as shown in Fig. A.9.
The performance curves under the pressure ratio (P2/P1) versus the reduced
flow (DP/P1) coordinate are shown in Fig. A.10.
Note that on P2/P1 versus DP/P1 plots in Fig. A.10, the surge reference lines
are nearly straight. While on P2/P1 versus FTE plots in Fig. A.9, the surge reference
lines bend forward, since FTE is proportional to sqrt(DP/P1).
All equivalent coordinate systems are approximate. They remain invariant within
a limited range of inlet condition variations. The resulting curves will become less
invariant if the inlet conditions differ significantly. See Fig. A.11 for the nitrogen
case, which significantly differs from the natural gas.
Fig. A.10 Performance curves under pressure ratio versus reduced flow coordinate
Fig. A.11 Performance curve with N2 gas
A.4.2 Data Selection
The surge parameters are calculated based on the worst-case operating scenarios
provided. It may not be necessary to include all the vendor-supplied data for surge
parameter calculation. Including certain extreme scenarios may unnecessarily reduce
the available operating region.
The final data selection shall reflect the anticipated operating requirement when
the compressor goes online. Good engineering judgment is needed to decide which
data shall be used and which shall be excluded.
In CPACS, any combination of cases can be selected for surge parameter calcula-
tion. Besides, a different selection of cases may be used for plotting and thus provides
the flexibility for comparison and cross validation. For example, Case D can be
selected for surge parameter calculation. In contrast, CASE D, Case D+95%MW,

and Case D-5%MW can be selected to plot the performance map to check how these
two scenarios perform with the calculated surge parameters.
A.4.3 ASP Formula
The general anti-surge parameter (ASP) formula (see Eq. 5.28) is given as a third-
order polynomial:
X
ASP = , (A.1)
P1 ∗ (C1 + C2 ∗ Y + C3 ∗ Y 2 + C4 ∗ Y 3 )
where X is the equivalent flow from the X-axis list, and Y is the equivalent head
from the Y-Axis list. Not all the X- and Y-axes combinations are valid as equivalent
coordinates. CPACS automatically rejects those non-invariant coordinates if selected.
For example, Fig. A.12 shows the ASP results based on the selections on the right.
The ASP formula, with three coefficients (ASP3), is given by CPACS as
X
ASP = √ (A.2)
840.81 − 862.64 Y + 473.66 Y 2
with X being the equivalent flow FME in Eq. 5.29 and Y being the pressure ratio PR.
This three-coefficient ASP formula is chosen because the surge reference line is
bending forward. For S-shaped surge reference lines, four-coefficient ASP formula
ASP4 may provide better performance. For most other applications, a two-coefficient
ASP formula is usually sufficient.
Fig. A.12 Anti-surge parameter (ASP) formula

A.4.4 Real-Time Data
The online measurements are typically limited to pressure and flow only. If available,
the operating points corresponding to the online measurement can be entered into
CPACS and visualized, to show their location relative to the surge reference line or
surge control line.
A.5 Summary
CPACS is an engineering tool developed for personal use, and is available for free
download at http://www.github.com/niucontrol/cpacs. While it has been used suc-
cessfully in various compressor control projects, no explicit or implicit guarantee or
warranty is provided or implied, including its usefulness or accuracy. The current
version is provided “as is,” and its usage is solely at the discretion of the user. The
author reserves the rights to make any changes without prior notice.
To report bugs or provide feedback, please send an email to NiuControl@
gmail.com or leave a comment at http://niucontrol.blogspot.com.
References

Glossary
Absolute Pressure The pressure measured above a perfect vacuum.

Absolute Temperature The temperature above absolute zero, stated in degrees
Rankine or Kelvin. Kelvin is the Celsius temperature plus 273.15 degrees.
Affinity Laws The affinity laws (also known as the “Fan Laws” or “Pump Laws”)
for pumps/fans are used to express the relationship between performance vari-
ables such as head, volumetric flow rate, shaft speed, and power. They apply to
centrifugal and axial flows in pumps, fans, and hydraulic turbines.
Automation The use of automatic control devices in a process so that human
supervision is minimized.
Brake Horsepower (BHP) The amount of power required to drive the pump or
compressor.
Capacity The rate of flow, determined by delivered mass flow rate divided by inlet
gas density.
Compressibility An coefficient factor to account for the deviation of real gas from
ideal gas
Compressor Curve The plot of discharge pressure versus suction volumetric flow
for a typical compressor speed or vane position at a specified suction temperature,
pressure, and molecular weight.
Compressor Diffuser The stationary passage around the compressor impeller
where the velocity pressure is converted to static pressure.
Compressor Impeller The blades on the rotating compressor shaft that impart the
velocity to the entering gas.
Compressor Map The compressor characteristic curves and the surge curve at a
specified suction temperature, pressure, and molecular weight.
Compressor Rotor The rotating element in the computer that includes the com-
pressor impeller and shaft.
Compressor Surge Point The capacity below which the compressor operation
becomes aerodynamically unstable.
Constraint Variable Process output that must be maintained within an operating
range.
https://doi.org/10.1007/978-3-031-43122-7
258 Glossary
Control Element A component of the control system that relates the error between
the desired value and the manipulated variable.
Cycling Periodic changes in a process variable. Also known as oscillation.
Dead band The largest range of values of a process variable (e.g., input variable)
within which the value does not respond to changes or cause changes to other
variables.
Disturbance An input signal that affects the controlled variable but is not directly
manipulated. Sometimes also called load change.
Fail Safe Fail-safe allows a system to return to a safe state after a breakdown of
the control. The fail-safe design allows the process to avoid hazardous conditions
that may otherwise occur.
Density The mass of gas per unit volume, equal to the reciprocal of the specific
volume. The density is a thermodynamic property determined from the absolute
total pressure and temperature at a point in the fluid using an equation of state.
Differential Pressure (DP) The difference between any two pressures measured
with respect to a common reference (i.e., the difference between two gauge pres-
sures.)
Equation of State An equation or series of equations that functionally relate the
gas thermodynamic properties, such as pressure, temperature, density, compress-
ibility, and specific heats.
Final Control Element The control element that directly changes the manipu-
lated variable, such as a valve or motor.
Gas Constant Universal gas constant, R = 8.314 kJ/(kg · mol · K)
Gas Power The power transmitted to the gas in a compressor, equal to the product
of the mass flow rate compressed and the gas work.
Gage Pressure Also called gauge pressure. The pressure measured with the exist-
ing barometric pressure as the zero base reference.
Inlet Guide Vane Stationary blades at the inlet eye of the impeller that direct the
angle of the gas flow into the impeller. The angle of the blades can be adjustable
to impart varying amounts of rotation to the gas and thus varies the capacity and
the discharge pressure of the impeller.
Instrument Range The region in which the instrument is designed to operate.
Interlock An interlock is similar to a trip but is considered a “self-resetting” trip.
Interlocks are typically used for on/off control and are not usually classified as
safety-related. They are often initiated by CD’S or PALL systems.
Isentropic Compression A reversible, adiabatic compression process.
Jockey Pump Low capacity pressure maintenance pump working on the principle
of on/off control.
Life Cycle The time interval that commences with the initiation of the concept
and terminates with the disposal of the asset.
Life Cycle Cost The total cost of ownership of an item considering all the costs
of acquisition, personnel training, operation, maintenance, modification, and dis-
posal.
Manipulated Variable Process input that is adjusted to maintain the controlled
parameter at the setpoint.
Glossary 259
Machine Mach Number The ratio of the blade tip velocity at the first impeller
diameter to the acoustic velocity of the gas at the suction conditions.
Mechanical Losses The total power consumed by frictional losses in integral gear-
ing, bearings and seals.
Net Positive Suction Head Available (NPSHA) The pressure available at the pump
suction flange. See also NPSHR.
Net Positive Suction Head Required (NPSHR) The pressure required at the pump
suction flange to prevent cavitation.
Overshoot The difference between the final steady-state value and the value of
the first maximum (or minimum if the response is downward) in a step response
of an under-damped system, typically expressed as a fraction;
Performance Curve A graphical representation of the performance characteris-
tics of a pump or compressor.
Permissive A permissive is a specific type of interlock used to prevent actions
from taking place until certain criteria are met. Permissives are typically initiated
by DCS or PLC systems, but if they have safety implications and meet the required
SIL rating, they may be implemented in an SIS or hardwired system.
Polytropic Compression A reversible, non-adiabatic compression process between
the total suction pressure and temperature and the total discharge pressure and
temperature.
Pressure Ratio The ratio of absolute total discharge pressure to absolute total
suction pressure.
Pressure Rise The algebraic difference between the discharge and suction pres-
sures.
Pump Efficiency The percentage of brake horsepower converted into useful work.
Pump Priming A pump should be primed to meet the NPSH requirements before
the pump can be started. This is typically done by keeping the pump elevation
below the lowest level of the liquid supply so that the pump remains primed at all
times and the NPSH requirements are met.
Rangeability The ratio of maximum value to minimum controllable value in a
final control element.
Redundancy The existence of more than one means of performing the same
intended function.
Resistance The potential required to produce change such as against a back-
pressure.
Restriction Orifice (RO) A restriction orifice limits the flow through the pipe to
a set flow by choking the flow at its sonic velocity.
Safety Instrument System A function comprising one or more initiators, a logic
solver, and one or more final control elements whose purpose is to prevent or
mitigate hazardous situations.
Self Regulation The inherent characteristic of a system that produces a steady-
state after a disturbance.
Sequence A sequence refers to a prearranged set of actions carried out by the
control system. Sequences can be initiated by events or operator actions.
260 Glossary
Setpoint Value at which the controlled parameter is to be maintained by the control

system.
Shaft Power The power delivered to the compressor shaft by the gas turbine, also
known as brake power. Shaft power is equal to gas power plus mechanical losses.
Step Response The response of a process variable in response to a step change in
another process variable.
Specific Gravity The relative weight of a liquid in comparison with water (S.G.
of water = 1.0).
Specific Heat (C) The amount of heat to raise one mass of gas by one degree
Specific Heat at Constant Pressure (C p ) Specific heat at constant pressure
Specific Heat at Constant Temperature (Cv ) Specific heat at constant tempera-
ture
System Resistance The pressure on the machine discharge caused by the resis-
tance to flow created by friction between the fluid and the piping system.
System Resistance Curve A graphical representation of the hydraulic character-
istics of a piping system.
Transducer Any device that transmits, amplifies, or changes a signal.
Trip A trip is an action initiated by the control system that forces a device or
devices to a predetermined state. Trips are often related to safety functions and
are typically initiated by a Safety Instrument System (SIS) or a hardwired system.
In some cases, PLCs or DCS may also initiate trips if they meet the necessary
independence and safety integrity level (SIL) requirements.
Throttling The intentional restriction of flow by partially closing or opening a
valve.
Valve Positioner A device that precisely controls the control valve stem position
by adjusting the instrument air pressure to the control valve.
Valve Seat The part of the valve that the valve plug rests upon when the valve is
fully closed.
Valve Stem A connecting rod between the diaphragm in a valve actuator and the
valve plug that allows air pressure on the diaphragm to control the valve plug
position in the valve and hence flow through the valve.
Index
A Best efficiency point, 45, 63, 71, 93, 198

Abnormal condition, 12 Brown-field project, 172, 219
Acceptance test Bumpless transfer, 217
black-box testing, 221 Bypass
factory acceptance, 221 cold, 187
site acceptance, 223 hot, 187
white-box testing, 221
Add-in, 243
Adiabatic, 30 C
Advanced control solution, 14 Capacity, 3, 69
Affinity laws, 50 Capacity control, 92, 93, 151, 153, 164, 166,
Analog input (AI), 222 205
Analog output (AO), 221 Capacity control line, 147, 196
Anti-choke control, 164 Capacity turndown, 198
Anti-reset windup, 217, 222 Cascade control, 153
Anti-surge control, 169, 206 Cause and effect, 49, 69
Anti-surge control valve, 169, 225, 230 analysis, 210
Anti-surge controller, 196 relationship, 49, 69, 94, 192
Anti-surge parameter, 140, 164, 234, 241 table, 69, 95, 107, 124
Anti-surge recycle valve, 9 Central control room, 229
Anti-surge trip, 180 Certified point, 123, 223, 224
Anti-surge trip logic, 180 Check valve, 9
API datasheet, 122 Choked flow, 61, 81
As-built document, 232 Choke line, 64
Automated operation, 216 Choke point, 62
Automated startup, 229 Choke valve, 59
Cold bypass, 86
Commissioning, 220
B Common mode failure, 181
Backflow, 9 Complex PID control scheme, 14
Backlean angle, 55 Compressibility, 28
Back pressure, 9 Compressor flow, 169, 197
Base load, 201, 215 Compressor performance monitoring, 113
Basic slope, 55 Compressor surge, 9
Basis of design, 172, 173 Conceptual design, 219
https://doi.org/10.1007/978-3-031-43122-7
262 Index
Control design kinetic, 9

equipment-centric, 1 potential, 9
process-centric, 1 Equation of state, 28, 188
Control error, 18 Equivalent coordinate, 131
Control handle, 69, 152 Equivalent flow, 139, 144, 172
Control overview diagram, 210 Euler’s Equation, 54
Control system, 220 Expansion factor, 177
Control target, 69, 152
Control valve, 15, 41
Controlled variable, 94, 95 F
Cooler Factory acceptance test, 221
discharge cooler, 8 Fan laws, 6, 50, 213
inter-stage cooler, 8 Fast-open valve, 222
suction cooler, 8 Feedback control, 221
Crippled mode, 68, 86 Feedforward, 198, 203
Critical flow, 81 Final control element, 15, 96, 221, 225
Fire and gas system, 14
Flare, 207
D Flow
Datasheet choked flow, 61
API datasheet, 173 compressor flow, 169, 197
compressor, 122 equivalent flow, 130, 144
control valve, 173, 242 forward flow, 72, 84
flowmeter, 173 machine flow, 72
process flow, 72, 76, 102
Datasheet flowmeter, 242
recycle flow, 72, 76, 102
Deadband, 226
reduced flow, 129
Deep surge, 63
square-rooting, 225
Deep water disposal, 83
volumetric, 129
Deliverables, 220
Flowmeter, 175
Demand driven, 94
DP based, 129, 225
Demand-driven control, 167
Forward flow, 72, 84
Density, 28, 29, 124, 144
4–20 mA signal, 225
Dependent variables, 69
Front-end engineering design, 219
Derivative action, 18
Detailed design, 219
Differential pressure, 144, 225 G
Digitization, 161, 174, 247 Gas power, 45
Direct acting, 204, 209 Green-field project, 172, 219
Discharge coefficient, 177 Guaranteed point, 123
Distributed control system, 13, 21
Dynamic response analysis, 210
H
Hairdryer, 74
E Heat loss, 27
E&P facility, 8, 67 Hot bypass, 86
Efficiency, 45 Human–machine interface, 13, 223
flow, 59
isentropic, 46
polytropic, 45, 46 I
Efficiency ellipse, 46 Ideal gas, 28
Emulsion, 7 Ideal gas law, 28
End-of-curve control, 164 Incipient surge test, 143, 147, 188, 227
Energy Independent variable, 69
Index 263
Inferential property, 15, 106 O

Inlet condition, 123, 124 On/off valve, 15
Inlet guide vane, 41, 96 Operating envelope, 9, 90, 104
Instrumented safeguarding, 13, 19, 91 Operating point, 67, 90, 159, 227, 235
Integral action, 18, 222 Optimization, 211
Integrated design, 196, 201, 205 Orifice, 170
Invariant coordinate, 64, 127, 128, 130–133 Orifice plate, 129
Isentropic, 30 Overall control strategy, 151
Isobaric, 29 Overriding control, 92, 183
Isochoric, 29, 31, 47
P
J Performance curve, 37
Joule–Thomson effect, 107 Performance indicator, 235
Performance monitoring, 232, 235
PID
K
components, 221
Key performance indicator, 113
configuration, 216
equation, 217
Gap control, 217
L
output, 169
Life-cycle, 16, 21
process variable, 169
Life-cycle ownership, 17, 21
reverse action, 157
Live performance map, 235
tuning, 230
Load balancing, 117, 199
by compressor train, 116 PID control, 18, 221
by stage, 116 PID tuning parameters, 230, 239
stage, 204 Piping and instrumentation diagram, 220,
train, 201 222
Load-balancing control, 116, 119, 205, 206, Plant automation system, 13
213 Polytropic efficiency, 46
Load controller, 199, 201 Polytropic exponent, 29, 47
Load distribution, 116, 199, 211 Polytropic index, 48
Load sharing, 116 Power, 44, 212, 234
Loop calibrator, 225 brake horsepower, 44
drive horsepower, 44
fluid power, 44
M gas power, 44
Machine flow, 76, 105 hydraulic power, 44
Management of change, 239 motor, 234
Manipulated variable, 14, 17, 95, 98 motor power, 44
Manual operation, 216 shaft power, 44
Measurement range, 111, 175, 224 Power consumption, 211, 234
provided, 178 Pressure ratio, 128, 129, 144, 213, 233
required, 178 Pressure rise, 128, 129
Middle of three, 107 Process control narrative, 220, 232
Minimum flow control, 162, 164 Process dynamics, 15
Mode transition, 13, 68, 82, 91 Process flow, 76, 102
Mollier diagram, 37 Process variable, 64
Project
brown field, 219
N green field, 219
Net suction pressure head, 97 Proportional action, 18
Non-invariant coordinate, 128 Protective control, 92, 152, 169, 206
Normal operation, 12 Pump priming, 84
264 Index
Q Surge margin, 199

Quality control, 93 Surge point, 61, 227
Surge reference line, 64, 133, 135, 140, 145,
147, 227
R Surge trip line, 147, 227
Ratio of specific heat, 30 Surge volume, 86, 98, 107, 108
Real gas, 28 System resistance, 9
Recycle flow, 76, 102
Reduced flow, 139
Reduced flow squared, 130 T
Reduced head, 128 Throttling valve, 15
Reference condition, 123 Throughput, 9, 69
Regulatory control, 92, 93, 107, 151 Total-recycle mode, 83, 84
Relay point, 95 Transient behavior, 15
Resistance curve, 37, 60 Transient response, 49, 94
actual, 61 Troubleshooting, 236
constant, 61 Turndown mode, 13, 112
Reverse acting, 157, 164 Turndown ratio, 164
S V
Safeguarding, 208 Valve
Safety envelope, 91 characteristics, 15
Safety instrument system, 13, 181 check valve, 9
Separator, 8 control valve, 15
Simple PID controller, 14 deadband, 226
Site-acceptance test, 223 fast-open, 222
Soft sensor, 106 hysteresis, 226
Specific heat, 30 non-return valve, 9
Speed of response, 107, 186 on/off valve, 15
Stall, 63 overshoot, 227
Standard operating procedure, 69, 85 positioning, 15
Startup, 220 speed of response, 15
plant, 227 sticky, 226
sequence, 229, 230 stroke test, 225
Startup sequence, 229 throttling, 15
Step change, 70, 227 Valve stiction, 226
Stonewall, 61 Variable
Stonewall line, 64 controlled, 18
Stonewall point, 62 manipulated, 18
Supply and demand, 94 Variable-frequency drive, 9, 166
Supply driven, 94 Variable-speed gearbox, 116, 166
Supply-and-demand model, 168, 208 Vena contracta, 177
Supply-driven control, 155, 167 Venturi tube, 129
Surge, 9, 61
Surge control line, 147, 196, 227
Surge indicator, 136 W
Surge line, 64, 133 Work process, 242

Processs Control in Compresoor

Uploaded by

Copyright:

Available Formats

Processs Control in Compresoor

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Processs Control in Compresoor

Uploaded by

Copyright:

Available Formats

Advances in Industrial Control

Process Control for Pumps

ISSN 1430-9491 ISSN 2193-1577 (electronic)

Paper in this product is recyclable.

Control engineering is viewed rather differently by researchers who produce general

Glasgow, UK Michael J. Grimble

Melbourne, Australia George Prattos

Industrial operations involve the continuous processing and movement of mate-

equipment control. The recent acquisition of Compressor Control Company (CCC)

1 Excel® is a registered trademark of Microsoft Corporation.

Houston, USA Steve S. Niu

1 Introduction to Pumps and Compressors . . . . . . . . . . . . . . . . . . . . . . . . . 1

2.3 Behaviors of Dynamic Machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

4.3.4 Recycle Valve and Manipulated Variables . . . . . . . . . . . . . . . 107

6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

A Performance Analysis and Control Design with Software Tool . . . . . . . 241

β Impeller backlean angle

g Gravitational Acceleration, g = 9.80665 m/s2

s Suction side, e.g., suction pressure Ps

Italics Used for emphasis or for special terminologies

MGI Miscible Gas Injection

Fig. 1.1 Type of rotating equipment by principle of work . . . . . . . . . . . . . 2

Fig. 2.6 Three basic variables for compressor performance

Fig. 6.31 Orifice-based flowmeter datasheet . . . . . . . . . . . . . . . . . . . . . . . . 179

Fig. A.4 Manual digitization of compressor performance curve . . . . . . . . 248

Table 1.1 Comparison of centrifugal and reciprocating machines . . . . . . . 5

Table 6.2 Minimum flow calculation for pump, with anti-surge

1.1 Type of Pumps and Compressors

1.1.1 Rotating Equipment

Fig. 1.1 Type of rotating equipment by principle of work

1.1.2 Pumps and Compressors

Fig. 1.2 Type of compressors by capacity

of the consumption. In addition, a 2001 report by the Department of Energy of the

1.1.3 Centrifugal and Reciprocating Machines

Fig. 1.3 Reciprocating compressor and bicycle pump

It increases the fluid pressure by operating on a fixed volume in a confined space.

"Non-Return Valves" "Non-Return Valves"

Fig. 1.4 Example: human heart as a reciprocating machine

Table 1.1 Comparison of centrifugal and reciprocating machines

1.1.4 Fans and Blowers

1.2 Applications of Pumps and Compressors

1.2.1 Liquid Pumping

Fig. 1.5 A simple pumping process

Fig. 1.6 Large pumps in pipeline operation

1.2.2 Gas Compression

Complex compression processes may involve multiple compressors working

Fig. 1.9 A compressor system with two compressors in parallel

Fig. 1.10 Multi-phase depletion compressors

Fig. 1.11 A typical ethylene plant with compressors

1.3 Operation and Control of Pumps and Compressors

1.3.1 Operating Requirements and Control Objectives

The primary objective of pumps and compressors in an industrial setting is to safely

The rise in global energy consumption and environmental consciousness, coupled

1.3.2 Overall Control Strategy

1.3.3 Process Control Technologies

Measure → Compare → Correct . (1.1)

Fig. 1.13 Closed-loop control (feedback control)

Fig. 1.14 A process with noise and disturbance

Fig. 1.15 Process measurement components

Fig. 1.16 Valve characteristics