Batch Process Fisher

(1)

Batch Control

Systems

Design, Application, and Implementation,

2nd Edition

(2)

Because neither the author nor the publisher have any control over the use of the information by the reader, both the author and the publisher disclaim any and all liability of any kind arising out of such use. The reader is expected to exercise sound professional judgment in using any of the information presented in a particular application.

Additionally, neither the author nor the publisher have investigated or considered the affect of any patents on the ability of the reader to use any of the information in a particular application. The reader is responsible for reviewing any possible patents that may affect any particular use of the information presented.

Any references to commercial products in the work are cited as examples only. Neither the author nor the publisher endorses any referenced commercial product. Any trademarks or trade names referenced belong to the respective owner of the mark or name. Neither the author nor the publisher makes any representation regarding the availability of any referenced commercial product at any time. The manufacturer’s instructions on use of any commercial product must be followed at all times, even if in conflict with the information in this publication.

Printed in the United States of America. 10 9 8 7 6 5 4 3 2

ISBN-10: 1-55617-967-7 ISBN-13: 978-1-55617-967-9

No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the publisher.

ISA 67 Alexander Drive

P.O. Box 12277 Research Triangle Park, NC 27709 Library of Congress Cataloging-in-Publication Data Hawkins, William M.,

1938-Batch control systems : design, application, and implementation / William M. Hawkins and Thomas Fisher.-- 2nd ed.

p. cm.

Rev. ed. of: Batch control systems / Thomas G. Fisher ISBN 1-55617-967-7

1. Process control--Data processing. 2. Process control--Automation. I. Fisher, Thomas. II. Fisher, Thomas. Batch control systems. III. Title.

TS156.8.F573 2005 660'.2815--dc22 2005029956

(3)

Thomas G. Fisher

The following is extracted from Bill Wray’s obituary for Tom, written for the World Batch Forum:

It is with profound sadness that we announce the loss of our friend and colleague, Tom Fisher. Tom passed away peacefully on December 6, 2001 after a long battle with cancer. He faced this struggle as he faced other challenges in his life, with dignity and courage.

Tom had a long career with The Lubrizol Corporation, joining as a process engineer in 1967 and rising to become Operations Technology Manager. He had previously worked for DuPont and NASA. His vision led to the formation of the ISA SP88 committee in 1988 and to the subsequent development of the batch automation standards we benefit from today. He served as Chairman and Editor of this committee as well as a leader of the IEC SC65A Working Group 11 for batch control. Further, Tom served as ISA Publications Vice President, was active in the ISA SP84 committee, and a member of the Process Control Safety subcommittee of the Center for Chemical Process Safety. His many contributions were recognized by ISA when he was named a Fellow of the organization. Tom was also a Registered Professional Engineer and member of the AIChE.

Without Tom’s leadership and vision, there would be no ANSI/ISA-S88.01-1995, Batch Control Part 1: Models and Terminology, he is therefore rightfully known as the Father of Batch Automation.

Tom was a teacher and taught courses through the ISA, WBF, and other organizations on topics such as safety interlock systems, programmable controllers, and batch automation. Above all else, teaching was his passion. Further, he was an accomplished author, having written several books and articles on batch automation and safety interlock systems.

If you ever met Tom, you know he was one of the truly nice guys in this world. Tom is survived by his wife Shirley, his children, Gary, Jeff, and Rebecca and several grandchildren. We will miss him greatly, not only as a mentor, but more importantly as a friend.

(4)

As a general principle, manufacturing is nothing more than changing raw materials into something else that is, hopefully, of greater value than the raw materials them-selves. In many types of manufacturing, machinery attacks raw materials directly. For example, machines are frequently used to shape chunks of metal to make parts that can then be assembled to create something like an automobile. The raw material is changed into another form because we do something physical and easy to understand directly to the raw materials. Process manufacturing is a little special and often dif-ferent in approach. While some processing activities do make overt changes in the material with activities like mixing or heating, many other processing activities merely set the conditions and the environment to which the raw materials are exposed and then let nature take its course. Chemicals react because the conditions are right, not because we have a machine that does the job. Bacteria or cells or corn mash grow and ferment because the environment is conducive for such goings on. In those cases, all we can do directly to the raw materials is to get them to the conditions that will let them do what we would like for them to do.

Process manufacturing has been around for a long time. Most societies since prehis-toric times have found a way to make beer and/or wine. For both products, the process consisted of putting the right stuff in the pot and then keeping the pot just warm enough without letting it get too warm. What was going on there might have been ancient and it may have been manual rather than automatic, but it was process control. Without control, beer can happen, but don’t count on it.

All modern process manufacturing requires control if it is to function with any relia-bility at all. It is an essential element of process manufacturing even though it is often looked on as “instrumentation” that is added on to a process to somehow make it work better. The reason for this slightly derogatory viewpoint is that manual control doesn’t show and doesn’t seem to count as control in the minds of many. The only type that gets credit as “real” control is the automatic version. Part of this view comes from the traditional approach to automatic control in continuous processes where “loops” and “devices” are applied to bits and pieces of a process so that a human oper-ator can manipulate targets for controllers. The controllers (the perceived “real” control) then set small groups of equipment or process variables to a desired condi-tion and twiddle valves and things to maintain that condicondi-tion.

A simple controller-based view of control may well be justified, or at least tolerated, in continuous processes where the whole intent of the technology is to maintain the same operating conditions indefinitely. The only times targets for control have to be adjusted are at startup, shutdown, and when things are not behaving well. That is hopefully much less than 5% of the time the process is running so the control exerted

(5)

That distorted view of control was one of the factors that hampered the development of more effective control of batch processes. The fool things would simply not stay in one state. They had to keep changing from one operating condition to another. Of course, the operator was making that happen so it was traditional to consider most batch processes as fundamentally manual in nature. However, it was obvious that the operator was doing control and was necessarily doing it almost 100% of the time the process was running. In that setting, the operator’s role in the control of the process couldn’t be ignored. In spite of that, it was difficult to figure out how to go about doing manual control tasks with more automatic control even after computer based tech-nologies gave us the tools to effectively confront the problem. Traditional state oriented and regulatory control could solve part of the problem and sequences could be imposed but problems remained.

Because people are very flexible, it was not immediately obvious that batch control had to also be flexible. Fixed sequences would work for a while, but often had to be redone as the process changed or other process requirements were added — changes and minor improvements that had seldom bothered the operator. In addition to that, the operator usually had to still do a lot of the control. Many approaches to the problem were proposed and tried, some quite inventive, but it took a while for a broadly accepted approach to batch control to emerge; a simple little standard pro-duced by ISA called ANSI/ISA-88.01-1995 (aka ISA 88).

To discuss control of batch process manufacturing without mentioning ISA 88 is a little like discussing a sandwich while avoiding the mention of bread. The standard defines an internally consistent approach for control of batch processes and the standard terminology that allows it to be described and understood. Because it addresses the entire scope of manufacturing activities, including procedure and coordination as well as more traditional single point control, it deals with essentially all of the functionality required in a batch processing plant. It also provides tools to help define the way the processing equipment should work with manual control, automatic control or some mixture of both. Whether the challenge is improvement to an existing process or a major new installation, the standard is a pretty reliable guide for automation.

What, precisely, is ISA 88? Physically, it is a combination of five documents; some still in preparation as this is written. The first part entitled Models and Terminology is a document that defines most of the important concepts the standard contributes to manufacturing automation. The committee that wrote the standard realized that a common set of terms would be needed to even start work on a standard, so termi-nology was addressed first. However, names can only be applied to things or concepts that are understood. At the time that work on the standard began, much if not most accepted terminology was related in some way to commercial products, most of which differed significantly from supplier to supplier. In addition, functions common to multiple suppliers often were assigned unique proprietary names. Also, no name

(6)

The models are general so that they can represent almost any batch manufacturing process and are valid no matter how control is actually implemented. Terms were then matched up with elements of the models. This may sound unacceptably

abstract, but it actually works. Once the basis for the models is clear they are intuitive and the terminology can be applied to any physical process.

The purpose of standard terminology was to provide a common language free of techno babble. Where words in common use could be applied, they were used. Where more than one word in common use meant the same thing in the context of the standard, either one of the conflicting words was chosen or the committee came up with a more appropriate term. Naming concepts that had never had a name was one of the more difficult tasks. The goal was to arrive at a common language that would allow people to communicate clearly and unambiguously with each other — engineers with other engineers and specialists, engineers with suppliers, managers with both, etc. The committee came pretty close. The terminology provides an ade-quate way to communicate concepts as well as references to tangible components of control, automation and process elements.

The models provide a generalized view of the manufacturing process. To be useful, they have to be abstract enough and flexible enough to fit essentially any batch process and still be detailed enough to represent reality. The models that resulted serve their purpose well. They are able to establish a basis for modularization and allow unsynchronized activities in a manufacturing facility to be visualized and under-stood. They represent required functions in each part of the process and identify the relationship between the various parts.

Defining models and terminology seemed fairly simple to begin with. Just draw up some models, plug in some terms, and go on to the important stuff. However, some-thing funny happened on the way to the models. In order to create them, it was

necessary to really understand and properly represent almost everything that goes on to make a product in a batch process. That ended up requiring five years in which up to 50 batch manufacturing experts met for several days every month or two to define a much more perceptive view of the organization of manufacturing equipment and the hier-archy of control functions needed to enliven that equipment. The result was a

structured view of manufacturing function that is very complete. Concepts like modular grouping of equipment and control, the meaning of recipe, the importance of the schedule, the inherent layering of control functionality, etc., were fitted together to create an internally consistent and structured view of essentially any batch manufac-turing process. It became the accepted road map for batch automation. It is important.

ISA 88 is important for several reasons but is best known because it has been proven to deliver measurable benefits when properly applied to define and implement batch process automation. Those and many other benefits are the subject of this book. Naturally all benefits come with some cost and the use of ISA 88 is no exception.

(7)

manager, the time required to start participating with a design team may be no more than few hours. Engineers, who must learn the technology as well as the principles, will spend more time. The time spent is definitely worth the effort, but it is absolutely necessary to invest that time in order to derive the benefits that are there for the taking. This book is a good start.

The fundamental difference between traditional control and control as outlined in the standard is that ISA 88 adds control of procedure and a level of coordination control necessary to keep multiple procedures sorted out. The concepts that are spelled out in terms of a manufacturing environment are consistent regardless of whether the control is provided manually or automatically. It treats control as a function that causes equipment to do the things necessary to make a product - no matter how that control is achieved. There is nothing wrong with manual control. It has worked well for many years and is not to be taken lightly. However, there are usually benefits to be gained by automating at least some of the procedural control in a manufacturing facility. ISA 88 works with either case, or both. It is based on the premise that it is the function that is important. This is particularly important in modern manufacturing approaches where smooth integration of manual and automatic control activities is needed. Few processes actually run all night with the lights out and people are not yet obsolete so we will be mixing automatic control and people for some time to come.

There is another signal principle that provides profound benefit. Products change; processes change and products are added. All of these changes require that the con-trol of the process be modified in some way. Product related changes are much more frequent than changes to the way that process tasks and actions are implemented. ISA 88 leverages the infrequent changes to process-oriented tasks and separates them from product processing information. Process oriented tasks are defined, modular-ized and separated from product specific information. A recipe contains the

information specific to a product. That recipe is used to orchestrate the sequence in which process-oriented tasks are executed, and to provide the values they will use for such things as amount, rate, pressure, etc. Most process changes can be made in recipes written by product experts who understand the process but are not necessarily expert in control. This allows control experts to design and execute the process-ori-ented tasks and for the equipment control to be effectively locked away in a controller while the robust and much less complex recipe can be executed in another part of the system. It is a powerful concept that definitely helps keep an automated plant run-ning in automatic.

The journey through the concepts and practicalities of batch control is a challenge, but a rewarding one. The following chapters should make that journey toward expertise more interesting and even more fun. Enjoy yourself along the way. Lynn Craig

Medford, NJ July 2005

(8)

xxi

The books that you read depend on many people to bring an idea to chapters to pub-lication. This one is no exception. It is not possible to list the names of everyone who contributed, if only because of memory fade. If I tried, I'd leave out someone that would be less than happy or even offended by the omission. The following is a short list of people who made the book possible.

Three individuals were key to the project, because the project would have failed without them. Lois Ferson of ISA acted as agent to get a book contract, motivator when the project looked impossible, and editor when there was something to edit. Lynn Craig read everything, and discussed it with me at great length, especially mod-ules and control activities. Joellen Burns, my wife, made it possible by not leaving me when I got totally wrapped up in writing. She supported me through it all and helped me to understand databases.

This book would be incomplete and not very interesting without the education and experience that I received from working with people in the batch control and fieldbus communications fields. Many thanks to Bill Wray and the people at the Channelview polyols plant for the most challenging startup. Not all of my education took place at startups, though.

Most relevant is the time spent in SP88 meetings and discussions, as well as time with the people in IEC SC65A WG11 in Europe. Some of the people who made me think about batch control in different ways were Tom Fisher, Lynn Craig, Dennis Brandl, Bruce Braunstein, Edgar Bristol, Harry Burns, Guido Carlo-Stella, Dave Chappell, Leo Charpentier, Tom Crowl, Dave Emerson, Larry Falkenau, Ashish Ghosh, Niels Hax-thausen, Tom Hollowell, Thomson McFarlane, Thomas Muller-Heinzerling, Albert Pampel, Lou Pillai, Michael Saucier, Keith Unger and Allen Weidenbach. WG11 had an unforgettable meeting with Reiner Uhlig. Time was also spent in SP95, but that's another long list with considerable overlap. John Vieille and Charlotta Johnsson are exceptional European members who have also worked in batch control.

The World Batch Forum has been a source of new ideas and current practices since Michael Saucier started it in 1994. There are many people that I'd like to recognize, but the list got too long. See the wbf.org web site for the list of officers and board members. WBF also sells a CD of all of the papers presented from 1994 to the latest meeting.

SP50 and later the Interoperable Systems Project and the Fieldbus Foundation pro-vided an education in both control and communications. Richard Lasher of Exxon led the User Layer, teaching all of us about function blocks, cascade control, and block

(9)

xxii

block work in ISP and FF, with Bill Hodson and Marcos Peluso as major contributors. Dave Glanzer managed the High Speed Ethernet project, and Lee Neitzel led the tech-nical work. Richard Corles provided support and valuable insights. Marten Hinrichs provided a European viewpoint, and many more interesting people were involved with control communication.

There is a really long list of people who provided me with process control experience and guided me through the tough parts, who worked for Hercules, Rosemount and Rosemount's customers around the world. Thank you all.

(10)

William M. Hawkins has more than forty years in the field of process control as an engineer, twenty with Hercules as a chemical industry user and twenty with Rosemount as a control system vendor. He has been an engineering consultant for the last five years. He showed engineering talent early, and received a BSME from MIT in 1960.

Bill started at his career at the Hercules Powder Company blasting cap plant, where he worked on assembly machines and invented special measurement equipment. He transferred to the Covington Fiber and Film plant, where he worked on electronic and instrumentation problems in fiber and blown film. He transferred to central engineering and established an instrument laboratory, where he designed and built Hercules’ first computer control system. He designed a batch control system for the TPA facility using a Foxboro 2/30 minicomputer and wired interlock logic mounted in the control panel. He also did a batch control project for polypropylene catalyst with a PLC.

Things changed, and Bill took a job with Rosemount in 1980 as Systems Engineering Manager, but soon became the control architect for Rosemount’s next control system, RS3. He started up the first RS3 at a corn milling plant, with a transfer network of 165 valves and 22 pumps. No PLC was required because RS3 could do that logic. He devel-oped ABC Batch with the help of the NAMUR recommendations in 1985. He started up the first ABC installation at a urethane polyols plant in Texas.

Bill joined the SP50 fieldbus communication standard and its User Layer in 1988, where he contributed to the design of control function blocks and learned a lot about fieldbus communication. He joined SP88 in 1992, where he helped to develop the concept of separation of recipe and equipment. He served as secretary for several years with Lynn Craig as chairman and Tom Fisher as editor, through the final ballot on ISA-88.01. He also served on the matching IEC standards committees, where he learned about the European viewpoint. He left SP50 in 1992 when a commercial consortium split away to get the job done. He wrote half of the function block specifications.

The World Batch Forum was started in 1994, and Bill was elected Treasurer, a post he held for eight years. He is still active in the Forum, currently involved with the confer-ence technical program.

Bill formed HLQ, Ltd. when Emerson closed the Minnesota RS3 plant in July,1999. He began by working for the Fieldbus Foundation on development and testing of the High Speed Ethernet fieldbus. Most of his time is now spent writing technical material.

(11)

Scope

This book is about process control as it is practiced in batch process industries. That means mostly recipe-driven procedural control and the equipment entities that make recipes relatively simple. The book does not include discussions of implementing most basic control, like temperature and pH control. There are many books on the subject of process control. The ISA 88 series calls it basic control relative to procedural control, which does not have a wide selection of books. It does not cover specific con-trol systems or computer operating systems or database applications for batch control, but does talk about those things in general.

Purpose

This book is intended to update ISA’s “Batch Process Control” book with the findings of SP88, both as discussion of 88.01 and as experience in applying 88.01, and to update the technology in the first edition.

Organization

There are sixteen chapters, each building on details discussed in previous chapters (except Chapter 1, of course). The first three chapters introduce processes, projects, and basic control. Experienced engineers could skip this material, but it might be interesting to see how someone else views those topics.

Chapters 4 and 5 introduce topics that are specific (but not unique) to batch control. Chapter 4 is about controlled equipment. Most people think of equipment and con-trol as separate things. 88 combines them to build an entity that can perform a process function. Process functions are specified by product recipes, not separate equipment and control functions. Chapter 5 introduces recipes as a way to specify products that require assembly procedures that vary from product to product.

Chapters 6 through 13 discuss the 88.01 standard from beginning to end. The stan-dard is mostly described in different words in this book, rather than as quotes from the standard. The different words are intended to give you another viewpoint, not as criticism of what the standard should have said. This should help you to understand what 88.01 says by looking at it in another way. It is too late to criticize the standard.

Chapter 6 begins the discussion of 88 with the physical model. At first, the model is intuitively obvious, as the professor used to say. All kinds of misunderstandings begin at the Unit and get worse as you go lower in the model. Alternative physical models are presented for perspective. A marketing person would say that this chapter clears everything up, but an engineer would be more restrained.

(12)

xxvi

related to the process functions that are performed by equipment entities. Chapter 8 explains the recipe models for the four types of recipes and the five categories of recipe information. These two chapters contain much of what is new in 88. Try to understand these chapters before proceeding. If the material won’t come together, try reading Chapter 12, which discusses how these concepts are used. Chapter 10 might help because it reviews chapters 7 and 8, among other things.

Chapter 9 discusses schedules and production information as a foundation for mate-rial in chapters 11 and 12. As a catchall chapter, it also discusses the 88 subjects of allocation and arbitration, modes and states, and exception handling. This finishes section 5 of 88.01.

Chapter 10 summarizes Chapters 6 through 9 and discusses the changes that sections 4 and 5 of 88.01 have made to previous ways of doing batch control. It also presents work done at Purdue that preceded and influenced 88, and reviews the NAMUR work presented in Chapter 5.

Chapters 11 and 12 cover section 6 of 88.01 on the subject of control activities. Sec-tions 4 and 5 presented new concepts. Section 6 describes how those new concepts are used to make products.

Chapter 11 discusses the three activities that require interfaces to the activities described in ANSI/ISA-95.00.03-2005. Recipe Management creates and maintains the recipes that are used to make products. Production Planning and Scheduling determines when what products will be made, and may specify the equipment, materials and personnel to be used. Production Information Management processes information generated by the batch control activities for use by plant operators and business functions.

Chapter 12 describes the three activities that are essential to make batch products, as well as the activity that maintains personnel and environmental protection. Process Management coordinates the distribution of control recipes to process cell units in a timely manner. Unit Supervision executes that part of a recipe than can be done in that unit, and sends commands to the Process Control activity, which applies the commands to equipment procedural and basic control in the modules belonging to the unit.

Chapter 13 presents the definitions in section 3 of 88.01. Some comments will help the reader and some are intended to improve the definitions in the next revision of 88.01.

Chapter 14 attempts some further clarifications of 88.01 from the author’s viewpoint. They would probably not gain consensus in an SP88 meeting. You may find them helpful in understanding what 88.01 did not say.

(13)

xxvii

the basis for the example. Hope you find it useful.

Chapter 16 discusses the 88-95 interface problem, which is still being discussed by a joint working group that doesn’t have enough control engineers to balance the data-base fashion designers. A solution is proposed, but probably too late to be considered. 88 seems like it might be useful in areas other than batch control, so possibilities and pitfalls are discussed. Finally, it seems like the largest future impact on batch control could come from the FOUNDATION™Fieldbus and their ability to do control in field

devices. This could lead to standard module interfaces for standard process functions, which could lead to the holy grail of useful standard phase libraries. These projects lack leaders, resources, and qualified people at the present time.

There is a lot of ground to cover in these 16 chapters. This book is not intended to be read in one evening. The plot is too tenuous and new characters keep appearing. There are passive voice statements that are intended to keep your excitement from becoming uncontrollable. There are some attempts at humor. Your mileage may vary.

All learning is based on making people ask the questions so that they will retain the answers. It will be helpful to be familiar with the contents of 88.01. Read far enough to form some questions and then spend some time thinking about possible answers to those questions. Then plow into the book again to find out what a group of industry experts came up with when they asked those questions.

Jumping into the middle of the book with a specific question is not recommended, unless you have gone through the book and understand all that is implicit in the answer to your question.

(14)

Chapter 1

Figure 1-1 Cross-Section of a Blasting Cap. . . 4

Figure 1-2 Approximate Layout of Explosives Plant. . . 6

Figure 1-3 Sample Specialty Baking Plant. . . 17

Table 1-1 Process Properties. . . 9

Table 1-2 Three Unambiguous Properties. . . 11

Table 1-3 Batch vs. Continuous Processes. . . 12

Chapter 2 Figure 2-1 Example of a PFD (TPA Process). . . 23

Figure 2-2 Rough Example of a P&ID. . . 25

Figure 2-3 Example of a Loop Sheet. . . 27

Figure 2-4 Example of a Block Diagram. . . 35

Table 2-1 Material Balance for One Reactor. . . 29

Table 2-2 Gantt Chart for Four Reactors/2-Hour Cycle. . . 29

Table 2-3 Overall Material Balance. . . 29

Chapter 3 Figure 3-1 Pressure vs. Flow Curves. . . 41

Figure 3-2 The Elements of a Regulatory Control Loop. . . 41

Figure 3-3 Discrete Control Loop. . . 47

Figure 3-4 Simple SFC. . . 50

Figure 3-5 Override Control by Value Position. . . 55

Chapter 4 Figure 4-1 A Combination of Process Equipment and Control. . . 64

Figure 4-2a Instrumented Reactor. . . 65

Figure 4-2b Bioreactor . . . 67

Figure 4-2c Mobile Reactor. . . 68

Figure 4-3 Shell and Tube Heat Exchanger. . . 71

Figure 4-4 Heat Exchanger for Heating and Cooling. . . 72

Figure 4-5 Heat Exchangers for Jacketed Batch Reactor. . . 72

Chapter 5 Table 5-1 Condensation of NE 33-Figure 7.1. . . 82

(15)

Figure 6-2 Physical Model. . . 94

Figure 6-3 ISO CIM Model Compared to 88.01 Model. . . 97

Figure 6-4 Comparison of Physical Structures. . . 99

Figure 6-5 Uppermost Levels of Communication. . . 106

Figure 6-6 Process Control Levels of Communication. . . 106

Chapter 7 Figure 7-1 Three-Level Hierarchy of Control. . . 109

Figure 7-2 Procedural Control Model. . . 111

Figure 7-3 Procedural Control/Equipment Mapping to Achieve Functionality. . . 116

Chapter 8 Figure 8-1 Recipe Types. . . 126

Figure 8-2 Relationship of General or Site Recipe and Master Recipe. . . 133

Figure 8-3 Linking of Control Recipe and Equipment Entity. . . 137

Figure 8-4 Unit Procedure Built into Equipment Entity. . . 138

Chapter 9 Figure 9-1a Simple Basic Control. . . 151

Figure 9-1b Transfer Header. . . 151

Figure 9-2 Sequence for Two Directions. . . 152

Figure 9-3 State Transition Diagram. . . 154

Chapter 10 Table 10-1 Hierarchy for Recipes. . . 160

Table 10-2 Generic Functions of a CIM System. . . 162

Chapter 11 Figure 11-1 Control Activity Model. . . 171

Figure 11-2 Concurrent Definition/Selection. . . 175

Figure 11-3 Recipe Management. . . 177

Chapter 12 Figure 12-1 Process Management. . . 188

Figure 12-2 Unit Supervision. . . 194

Figure 12-3 Process Control. . . 199

(16)

Chapter 14

Figure 14-1 Complete Physical Model. . . 219

Figure 14-2 The Physical Model and the Purdue Reference Model. . . 225

Figure 14-3 The Common Valve Problem–Two Units. . . 240

Figure 14-4 The Common Valve Problem–More Than Two Units. . . 241

Chapter 15 Figure 15-1 Procedure for Designing a Batch Process Control System. . . 256

Figure 15-2 Procedure in Three Stages and Three Operations. . . 258

Figure 15-3 Process Stage in Three Operations and Three Actions. . . 258

Figure 15-4 Control Module Pattern Drawing. . . 262

Figure 15-5 Example of Module Grouping. . . 265

Figure 15-6 Reactor Temperature Control. . . 275

Figure 15-7 Control Modules Required for Interfacing with Phase Logic. . . 277

Figure 15-8 Control Block Configuration for Reactor Temperature Control. . . 279

Table 15-1 Table for CM PID Pattern Drawing. . . 263

Table 15-2 Description of PID Pattern. . . 263

Table 15-3 Table for Specific Module . . . 264

Table 15-4 Table of Common Parameter Names. . . 264

Table 15-5 Generic 88 Implementation for Heat Transfer. . . 281

Table 15-6 Generic 88 Implementation for Heat Transfer Shut Down. . . 282

Table 15-7 Generic 88 Implementation for Heat Transfer-Modified. . . 286

Table 15-8 Generic 88 Implementation for Heat Transfer Shut Down-Modified. . . 288

Chapter 16 Figure 16-1 Activity Model of Production Operations Management. . . 295

Figure 16-2 Control Activity Model. . . 295

Figure 16-3 Comparison of Ideas from 88 and 95. . . 296

Figure 16-4 Possible Interface of 88 and 95. . . 300

(17)

ANSI/ISA-88.01-1995 - Batch Control Part 1: Models and Terminology (Formerly ANSI/ISA-S88.01-1995).

ANSI/ISA-88.00.02-2001 - Batch Control Part 2: Data Structures and Guidelines for Languages.

ANSI/ISA-88.00.03-2003 - Batch Control Part 3: General and Site Recipe Models and Representation.

ANSI/ISA-88-00-04-2006 - Batch Control Part 4: Batch Production Records. Automation, Systems, and Instrumentation Dictionary. Fourth edition. ISA, 2002. Bernard, John W. CIM in the Process Industries. Independent Learning Module Series. ISA, 1989.

Fisher, Thomas G. Batch Control Systems. Resources for Measurement and Control Series. ISA, 1990.

Lipták, Béla G., editor-in-chief. Instrument Engineers’ Handbook.

Volume 1 – Process Measurement and Analysis. Fourth edition. CRC Press/ISA, 2003. Volume 2 – Process Control and Optimization. Fourth edition. CRC Press/ISA, 2006. Volume 3 – Process Software and Digital Networks. Third edition. CRC Press/ISA, 2002.

McMillan, Gregory K. Tuning and Control Loop Performance. Third edition. ISA, 1994. Merriam-Webster’s Collegiate Dictionary. 10th edition. Merriam-Webster Inc., 1993. Nisenfeld, A. E., editor. Batch Control. Practical Guides for Measurement and Control Series. ISA, 1996.

Parshall, Jim and Larry Lamb. Applying S88: Batch Control from a User’s Perspective. ISA, 1999.

Shlaer, Sally and Stephen J. Mellor. Object-Oriented Systems Analysis: Modeling the World in Data. Yourdon Press Computing Series. Yourdon Press, 1988.

Webster’s New Collegiate Dictionary. G. & C. Merriam Co., 1958.

(18)

This book is the second edition of ISA’s Batch Control Systems: Design, Application and Implementation written by Thomas G. Fisher and published in 1990.

Twenty years later, process engineers still select batch processing for products that can’t be made continuously or discretely. Biotechnology has been added to the list in a big way.

Hardware prices have come down, but vendors no longer own the computer-related technology. System prices will not go much lower. Field devices have enough metal work in them that they will not go much lower. Raising the price is the cost of software development, even though tools for doing software are improving.

Twenty years ago there were few standards applicable to batch control. RS-232 was an electrical standard, not a communication standard. The first edition was the precursor to ISA’s 88 Batch Control - Models and Terminology standard. That standard improved the communication of batch control needs and solutions, and has affected all modern systems. An even bigger change in communication standards, such as TCP/IP and various kinds of fieldbus, has also impacted the way we do batch control today.

This book brings the subject up-to-date with a discussion of ANSI/ISA-88.01-1995, Batch Control Part 1: Models and Terminology, as the central part of batch process control. This is basically a new book, because many of the original subjects were either made obsolete by advancing technology or redefined by SP88. The focus is batch con-trol, but some of the techniques will improve any kind of process.

Tom Fisher died in 2001. ISA asked me to update the book, now that 88 is widely rec-ognized (but not widely understood). Tom wrote the first edition with what was available in the eighties and two years of SP88 meetings, led by his Lubrizol colleague, Rick Mergen. The book influenced SP88 to the extent that Tom’s work was a starting point for many of the discussions that eventually shaped the standard. I hope that this book is the second edition that he would have written, but perhaps he considered ANSI/ISA-88.01-1995 to be his second edition.

Tom was responsible for founding the committee and was the recognized leader, serving as chairman for several years and as editor for many more. Although Rick and Lynn Craig served terms as chairman and left their own marks on the standard, Tom was the chairman and leader during the most critical period of SP88’s growth. That period included debates that defined most of the fundamental principles of 88. Tom served as the editor from the committee’s inception until his untimely death. He saw ISA-88.01-1995 published as a national and as an international standard. Tom also

(19)

Most of us know that standards do not create themselves. Strong leadership is required, with extra skill in herding cats that all want to go their own way. Creative leadership is required when no standard exists. This is not to say that 88 was created from thin air. Human history is a story of incremental progress, with the occasional major setback. Isaac Newton said, “If I have seen further it is by standing on the shoul-ders of Giants,” and many have found it fitting to apply that saying to their own subject. 88 was based on the work done by giants like Reiner Uhlig of NAMUR and Ted Williams of Purdue. 88 was also based on the work of many who have labored by the vats of a wide variety of batch processes.

The analogy of standing on tall shoulders implies a certain instability. Perhaps it is more accurate to think of laying bricks in a wall. Many people have worked on batch process control over the years, each laying one or more bricks to contribute to the height of the wall. The result would have been a mess without leaders to keep the wall straight and level. SP88 has laid a lot of brick, more or less straight and level, but the wall isn’t done yet.

Bill Hawkins August 2005

(20)

Forward . . . xv

Preface . . . xix

Acknowledgments . . . xxi

About the Author . . . xxiii

About the Book . . . xxv

Chapter 1 Introduction to Manufacturing Processes. . . 1

Process Classification. . . 3

Process Properties. . . 9

Batch Process Definition . . . 12

Auxiliary Processes . . . 16

Process Boundaries . . . 17

Summary. . . 18

Chapter 2 Introduction to Process Design and Construction. . . 19

Process Flow Diagram . . . 22

Analysis . . . 24

Piping & Instrument Diagram. . . 24

Loop Sheets . . . 26

Example of Process Design and Construction . . . 28

Process Flow Diagram. . . 29

Batch Design . . . 30

Construction . . . 32

Modular Design. . . 33

Summary. . . 36

Chapter 3 Introduction to Process Control . . . 39

Types of Control . . . 40 Regulatory . . . 40 Discrete Control . . . 46 Sequential Control . . . 49 Constraint Control . . . 51 Alarms . . . 52 Overrides. . . 54 Interlocks . . . 55 Interlock Variations . . . 56 Summary. . . 58

Chapter 4 Introduction to Controlled Equipment . . . 59

The Role of Humans in Process Control . . . 59

Process Equipment . . . 62

Controlled Process Equipment. . . 63

Examples of Controlled Equipment . . . 64

(21)

Chapter 5 Recipes. . . 77

Definitions . . . 77

NAMUR . . . 79

Recipes. . . 83

Evolution of Recipes . . . 83

Changing from Drums to Digital Computers. . . 84

Modular Programming . . . 86

Separating Recipe and Equipment Programming . . . 86

Summary. . . 87

Chapter 6 88 Physical Models . . . 89

Modeling . . . 89

Models in 88.01 . . . 90

Batch Processes and Equipment in 88.01 . . . 91

Batch Processes. . . 91

Physical Model . . . 93

Process Cell Classification . . . 97

Other Physical Models . . . 101

Control in the Physical Model. . . 104

Summary. . . 107

Chapter 7 88 Batch Control Concepts, Part 1 . . . 109

Structure for Batch Control . . . 109

Basic Control . . . 110

Procedural Control . . . 110

Coordination Control . . . 114

Equipment Entities . . . 115

Relationship Model . . . 116

Control in Equipment Entities . . . 117

Structuring Equipment Entities . . . 121

Summary. . . 123

Recipes. . . 125 Recipe Types . . . 125 Recipe Contents . . . 129 Recipe–Equipment Relationship . . . 134 Recipe Transportability. . . 139 Summary. . . 140

Production Plans and Schedules . . . 141

Production Information . . . 142

Allocation and Arbitration . . . 145

(22)

Summary. . . 156

Chapter 10 88 Perspective and Review . . . 157

Before SP-88 . . . 157 Purdue Workshop WG4. . . 159 NAMUR. . . 160 Batch Control Systems . . . 161 Computer Integrated Manufacturing. . . 161 Purdue Reference Model . . . 161 During SP-88 . . . 165 Summary. . . 167

Chapter 11 88 Control Activities and Functions, Part 1 . . . 169

Control Activities . . . 169 Control Activity Model . . . 170 Information Handling. . . 172 Process and Control Engineering . . . 174 Recipe Management . . . 176 Production Planning and Scheduling. . . 180 Production Information Management. . . 181 Summary. . . 186

Chapter 12 88 Control Activities and Functions, Part 2 . . . 187

Process Management . . . 187 Unit Supervision . . . 194 Process Control . . . 198 Personnel and Environmental Protection . . . 201 Summary. . . 201

Chapter 13 88 Definitions . . . 203

Other Definitions . . . 215 Summary. . . 215

Chapter 14 Further 88 Clarifications. . . 217

Definitions . . . 217 Physical Model . . . 218 Batch Control Concepts . . . 224 Modules and Phases . . . 224 Equipment Entities and the Purdue Reference Model . . . 225 Exception Handling. . . 226 Modes and States . . . 232 Recipe Operations . . . 237 Alternative to Unit Procedures . . . 238 The Agitator Problem . . . 239 The Common Valve Problem . . . 239

(23)

Executing Procedural Elements . . . 242 Recipe Computations . . . 243 Operator Interaction . . . 245 Procedure Function Charts . . . 246 A Word about Control Systems . . . 247 Summary. . . 250

Chapter 15 Generic 88 Implementation. . . 253

Designing Batch Process Control. . . 254 Know Process Design . . . 255 Know How to Operate. . . 260 Know Modules, Phases . . . 269 Know Normal Operations . . . 274 Know Exceptions . . . 285 Know All There Is to Know. . . 289 Summary. . . 291

Chapter 16 Role of 95 and Other Things. . . 293

Joining 95 and 88 . . . 294 Interfaces . . . 296 Missing Interfaces . . . 300 The Role of STEP. . . 301 Expanding the Scope of 88. . . 302 Why Expand 88 . . . 302 Features of 88 . . . 303 Back to 95 . . . 309 Fieldbus. . . 309 History. . . 310 Future for Batch Control . . . 311 Control in the Field . . . 311 User-Defined Function Blocks . . . 312 Summary. . . 313

Bibliography . . . 315 Index . . . 317

(24)

Introduction to

Manufacturing Processes

The goal of this book is to explain batch process control as specified in ISA’s 88 Batch Control standards. It begins with that goal in the distance, so it looks attractive and interesting instead of the slithering mass of slippery details that it actually is. By the end of the book, most of the details will have stopped slithering and revealed their true nature. It is not possible to pin them all down, of course, because people will do unique things.

Batch process control refers to one of three classes of manufacturing processes: dis-crete, batch, and continuous. This chapter will introduce some concepts to help you to identify processes, with examples to illustrate them. It will clarify the 88 definition of a batch process and briefly cover other kinds of processes. Finally, this chapter will discuss how to use this process knowledge to define boundaries in a set of processes that will simplify process design.

The result of reading this chapter should be that you can at least tell a batch process from a buffalo, much like the Gilbert and Sullivan’s Modern Major General who could “tell at sight a Mauser rifle from a javelin.”

Manufacturing

Manufacturing transforms raw materials into different forms that have more value than the materials that went into them. Whether the result is useful depends on your point of view. Lumber is manufactured from trees. Titanium ingots are manufactured from Australian beach sand. Aircraft are manufactured from parts manufactured from titanium ingots, among other things. Fuel is manufactured from crude oil that was manufactured by the Earth at a rate no human manager would tolerate.

Manufacturing is performed at facilities whose locations are determined mostly by customer demand, shipping distance, and the complexity of the manufacturing process. Nobody manufactures aircraft from beach sand or houses from trees at a single location. These facilities are called “plants,” “sites,” or “plant sites.” Site refers to an area of ground and plant refers to a container for processes, so it is possible to have several plants on a site.

(25)

Process

Process has different meanings in law, medicine, business, and computer science. In this book it is used as a contraction for “manufacturing process,” meaning the set of things that are done to incoming materials to produce at least one product that is more valuable than the inputs. In the words of the International Technology Educa-tion AssociaEduca-tion (www.iteawww.org), process is “a systematic sequence of acEduca-tions that combines resources to produce an output.” As a verb, process means doing the things to materials that will produce products.

Plastic toys are manufactured from crude oil using a series of processes. A process performs a part of the total transformation. To put it another way, manufacturing can be broken down into processes. Raw materials are processed into intermediate prod-ucts that are used in other processes to manufacture an end product. The end product is sold to customers who have it shipped to their manufacturing or distribu-tion locadistribu-tions. For example, the manufacture of lumber requires one process to strip the bark from the logs, another process to saw the logs into rough lumber, and a fur-ther process to produce finished lumber. Manufacturing plywood or particle board requires entirely different processes.

Processes are divided into three classes. This is so because one of the laws of human nature says that most of the audience will get lost if a book introduces more than three things at one time. The classes are discrete, batch, and continuous:

Discrete

The product of a discrete process is an object that maintains its shape after processing. Each processed object could be labeled to distinguish it from other products of the same process. Liquids and gases can be packaged so they become part of the output of a discrete process, such as a beverage bot-tling line. Discrete processes usually resemble assembly lines, in which each part is carried past a machine that does something to it.

Batch

A batch process is similar to a discrete process in that a sequence of manipu-lations is performed in order to make one product. However, in a batch process the product is not a discrete part. Batch processes do not involve a constant flow of materials in and out of the process, and the output is nor-mally a homogeneous mass, not discrete objects. After a useful chemical transformation is developed in a laboratory, skilled people reduce the condi-tions for the transformation to the minimum number required to make the product. These conditions include the amounts of each raw material and the procedures for transforming them. The result is a narrative recipe for making a product. The products of a batch process may be dough for bagels, tereph-thalic acid (TPA) for polyester yarn, the finish solution for permanent press goods, or the base for various shades of paint. The laboratory process may be scaled up to a commercial batch process in which the beakers and Bunsen

(26)

burners become highly agitated tantalum tanks and multimedia heat exchangers, or something less extreme.

Continuous

Continuous processes involve a relatively steady flow of materials in and out of the process, and are most efficient when they have reached a steady state. Any sequence of manipulations in a continuous process is confined to startup and shutdown. Continuous processes are developed when customer demand exceeds the ability of batch processes to make the lowest-cost product. Petro-leum refining is an example of a continuous process, as is the conversion of fuel into electricity. Many specialty chemicals and some pharmaceuticals cannot be made using a continuous process.

Process Classification

The purpose of this section is to introduce the classification of processes to those who have not yet seen them or to provide a review for those who are certain that they know what will be said. Knowing the classification of a process helps engineers determine what the terms used in a process mean and the scope of the control that a process requires.

This section will apply the three classes of manufacturing processes to a real manufac-turing plant, which made blasting caps and other electro-explosive devices in the 1960s. This historical example allows us to discuss the plant’s processes in some detail, because the plant has been sold and the processes are obsolete. While the example is historical, the present tense will be used to describe it.

Blasting caps are used to initiate explosions, most often to fracture rock so that it can be removed from a construction or mining site. They may seem an odd choice for an example, but the blasting-cap plant is rich with processes, unlike a boring refinery. Though the plant will be novel to almost everyone, the relevant process classifications should not be difficult. If you can classify the processes in a blasting cap plant, you can classify processes anywhere.

(27)

Blasting Cap Plant

A blasting cap consists of a drawn copper shell about 0.25” (6 mm) in diameter and long enough to contain the detonating ingredients, time delay fuse, and plastic end plug. A pair of wires is attached to the two pins in the end plug, to be connected to a source of electricity. Different lengths of wire are used to accommodate the various depths of the holes that are drilled in the rock to be blasted. The major raw materials for the blasting cap plant’s processes are copper sheet for the shells, bare copper wire for the leads, explosive and time delay powders, plastics for the wire insulation and shell plugs, and fine wire for the bridge wire that heats and sets off the explosives. Figure 1-1 shows a cross-section of a blasting cap.

Figure 1-1 Cross-Section of a Blasting Cap

Blasting caps are sold to mining and construction companies with different lengths of lead wires and various time delays between the electrical impulse and the detonation of the cap, including zero delay. When a tunnel bore is being blasted from rock, it is impossible to explode the entire tunnel face at once and still maintain the tunnel diameter. Instantaneous caps are used for a small area in the center of the bore. After that rock is loosened, millisecond-delay blasting caps are used to reduce a ring of rock around the center to rubble, and so on out to the bore diameter. Different kinds of rock require different delays.

The plant has the following processes:

1. Shell Plant—Stamping presses draw tubular shells from copper sheet. The length

of the shell varies depending on the time delay. The shells are collected in boxes sorted by length. The diameter does not vary.

2. Delay Fuse Plant—A powder mixing process produces a batch of fuse powder with

a known rate of combustion. The powder is poured into lengths of lead tubing on a vibrating fixture. The lead tubes are drawn through a die that reduces the diameter to the ID of the shell and compacts the powder. The tubes are X-rayed to detect voids that would fire early or not at all. Then they are cut to length for a specific delay, and samples are tested.

(28)

3. Loading and Pressing Bays—Shells are manually placed into 500-shell press blocks.

Each shell is loaded with a measured volume of explosive powder by a machine. The loaded shells are manually placed in a 500-pin press and pressed to compact the powder. The loading process may be repeated with a time delay fuse and is repeated with a second powder. For safety reasons, the operator steps behind a thick wall while operating the press. The press bays have one weak wall to relieve blast pressure, away from the operator, of course.

4. Wire Plant—Spools of thick copper wire are drawn down to #19 gauge wire through

a series of dies. When a spool of wire ends, it is butt-welded to a wire from a new spool. The thick wire moves slowly relative to the drawn wire. Two wire drawing stands provide a pair of wires to the insulation extruder. Plastic insulation that has been treated to reduce static electricity is applied to the pair of bare wires in a spe-cial die. The result is a pair of insulated wires that are joined by a thin plastic bridge, which can be broken by pulling the ends of the wire apart. The paired wires run through a water trough that cools the plastic and detects pinhole faults. A capacitance gauge that contacts the insulation senses the location of the wire within the insulation, which tells a dead-time compensating controller how to adjust the position of the wire guides. Finally, the paired wires are wound into 5000 foot (1500 meter) spools in about ten minutes. One spool is enough wire to make an average of three hundred caps.

5. Assembly Plant—Assembly is done in three processes. Each process uses several

assembly machines that are designed for one of the three processes. A basic assembly machine is a large round steel table that is rotated by a Geneva wheel positioning mechanism, which makes a precision partial rotation and stops. The table carries as many fixtures for holding the part to be assembled as there are stops on the Geneva wheel, typically six to twelve. Assembly operation stations are precisely located around the table in fixed positions. A part is placed in a fixture on the table at the feed station. The three assembly processes are as follows:

a. Wire Folding—The table fixture is designed to hold a coiled length of wire on two pins. The feed station winds the desired length of wire around the fixture pins in a figure-eight pattern. The next station strips the insulation from one end of the pair. Next, a high-potential insulation test reveals any holes or shorts and flips a toggle on the fixture if the test fails. Then a color-coded strip of adhesive paper is wrapped around the center of the figure-eight coil. Finally, test rejects fall into a recycle container, and good product is stored in boxes. Each wire bundle will become the lead wires for one blasting cap.

b. Plug Molding—5.5 mm plastic plugs are cast for the shells around two tinned copper wires. A fine wire is then welded to bridge the two wires in the plug. The bridge wire will contact the sensitive explosive. Each plug is pressed into a loaded shell. The fixture is designed to contain an explosion.

(29)

Some empty assembly stations provide time for a delay cap to detonate. Basic cap assemblies that pass a bridge-wire resistance check are boxed for transport to the final assembly machine.

c. Cap Assembly—The folded lead wires are welded to a basic cap assembly. Plastic insulation is extruded over the welds, the bridge wire continuity is tested, and the assembled cap is either accepted or rejected.

The plant also makes time-delay caps. This process requires that special powders be mixed to achieve the right delay time. The powder mix is pressed into a lead sleeve, which is pressed into the shell between the bridge wire and the explosive charge. Statistical quality control methods are used because most tests are destructive. That is, one cannot sell a blasting cap that has been given a functional test.

Figure 1-2 shows the approximate layout of buildings on the plant site. Explosives plants tend to have small buildings spread out according to the amount of explosive contained in the building. The distances to prevent explosive propagation have been learned from hard experience.

(30)

Characterizing the Processes in a Blasting Cap Plant

Now that we have described the plant processes, we can characterize them as discrete, batch, or continuous.

1. Shell Plant—Copper sheets are transformed into copper shells. No chemical

change or mixing is involved, although the metallurgy of the copper has changed. The product, a drawn shell, is distinct from other shells, and each shell could have been identified with a number, although this was not common practice forty years ago. The process is therefore discrete because the output is discrete. The press does go through a sequence of operations to make products, but this alone is not enough to make it a batch process.

Of course, the operator says that “a batch” of shells has been made when the lid is closed on a box. This is the problem with using common terms. The box is actually a “lot” that can be statistically sampled, so that a bad lot of copper sheet or a bad machine can be identified.

2. Delay Fuse Plant—Powders made elsewhere are selected according to a recipe and

mixed to produce a batch of fuse powder that will burn at so many inches per second. This qualifies as a batch process. The powder is poured into lengths of lead tubing, which are drawn through a die. The tubes are 100% X-rayed. They are cut to length for a specific delay, and samples are tested. These are discrete processes. The result is one lot of fuses that will almost certainly all provide the same time delay.

3. Loading and Pressing Bays—Empty shells are filled with measured amounts of

powder according to a recipe, and they are pressed to compact the powder to a cer-tain density. The powder is made at another plant, stored in bunkers, transferred to the loading room, and weighed into each shell. There is the following sequence: insert shells, load, press, load, press, and deliver to storage. The sequence must be repeated to make another rack of five hundred shells (a finite quantity). Neverthe-less, the rack of shells is not a batch; it is a lot for statistical purposes. Chemistry is not involved, nor is mixing. Discrete parts come in and go out. Loading requires weighing charges of powder, but the powder hopper is periodically refilled as required. The charging process takes a continuous supply of powder and continu-ously supplies discrete charges to discrete shells. The loading and pressing processes are discrete.

4. Wire Plant—Wire is drawn, insulated, and spooled. These are three separate

processes. Wire drawing does not change the chemistry, but it runs continuously for all practical purposes. No sequence of operations is required to make the product, once the startup procedure of threading the dies has been performed. Threading the dies is not a trivial procedure, which is why a butt-welder is included to attach the end of one spool to the beginning of another. Wire drawing is a con-tinuous process.

(31)

Extrusion heats plastic pellets and mixes them with dye. But, more pellets are fed to the process as plastic is extruded, and no sequence of operations is required to produce product. So, extrusion is continuous.

Spooling wire is a process that takes a continuous input and produces a discrete output. There is no mixing or chemical change, but the output can be distin-guished from the stream of wire and labeled as a distinct item or part. A sequence of operations is required, but this is typical of discrete processing. The sequence is required to make a physical change in a discrete part. In this example, a wire spool is changed from empty to full.

5. Assembly Plant—No question here: assembly machines that move a discrete part

to different tool stations (or vice versa) so as to make physical changes are the essence of a discrete process.

The only batch process in the plant is the manufacture of the time delay powder. The rest of the book will provide many examples of batch processes.

Other Process Examples

Consider a plant that converts scrap baling wire from hay and cotton bales into high quality iron oxide for magnetic tape. This process is referred to as a “bucket and paddle” operation because it is simple, but it is also highly profitable. There are only a few buckets, so each must be charged with bits of baling wire and secret ingredients, processed according to a secret procedure, and emptied so the next batch can begin. This is a classic example of batch processing: a vessel is filled, sequentially processed to chemically transform the added materials, and emptied before another cycle can begin.

A continuous plant may also have sequenced equipment, such as the switch condensers that extract waxy hydrocarbons from the process gas stream. A switch condenser runs for a while with process gas flowing over tubes of chilled water that congeal the waxy product. Then the process gas is diverted to another condenser, and the chilled water is replaced with steam to melt the wax, which drains into a product pipe. The unit is switched back to cold water, and process gas once again flows through it. There is a sequence of operations, but it is designed to allow continuous flow through a set of process equipment. Other examples are gas dryers and centrifugal separators that must be stopped in order to remove product. The result is a continuous flow of product and the sequence never varies, so the processes are continuous.

The procedure for starting up or shutting down continuous processes is definitely sequential, probably has chemical changes, and does not have a continuous output. But no oil refinery engineer would call startup a batch process. Startup is viewed as a temporary aberration in the life of a continuous process.

The point is that the three categories of manufacturing processes are convenient divi-sions that make it possible to break a large problem into smaller parts. That is the

(32)

engineering use of categories. Marketing uses them to be able to say, “Mr. Customer, what you have there is a batch process. What we have is far and away the best batch control system available, and you need it.” This bold assertion can be used to bypass all sorts of pesky little details.

The following sections will explain how to distinguish a batch process from other processes with a high probability of being correct.

Process Properties

The following table presents some properties of processes, in an attempt to find one or two properties that always characterize the correct process. There are other processes in a plant, such as transportation and storage, but they will not be described here.

Table 1-1 Process Properties

Input materials are mixed or chemically changed

Chemical change transforms the ingredients into materials that have a dif-ferent chemistry. Mixing or blending transforms the ingredients into a product that is homogeneous, not discrete. A discrete process may cause a chemical change such as oxidation or flame hardening, but the change occurs in a discrete part. A continuous process usually changes chemistry, but it may only cause a physical separation.

Input materials are only changed physically

Physical change transforms one or more physical properties of the material, such as shape or hardness, or it adds other physical parts to make a different part. Some chemical change may occur, but that is not the principle reason for the process. Separation is a physical change, such as liquid from solid or indi-vidual hydrocarbon gases from natural gas. Discrete separation is called “sorting.”

Property Discrete Batch Continuous

Input materials are mixed or chemically changed Maybe Yes Maybe Input materials are only changed physically Yes No Maybe Input materials must be fed continuously Maybe No Yes Product materials leave while materials are coming in Yes No Yes A sequence of operations is required for normal production Yes Yes Maybe

A recipe is required Yes Yes Yes

One or more production vessels require a mixing agitator No Yes Maybe Output is multiple objects that can be individually labeled Yes No No Output is an unbroken stream of product No No Yes Output requires that the production vessel(s) be emptied Yes Yes No

(33)

Input materials must be fed continuously

The law of conservation of mass requires that a process with a continuous output have a continuous feed. The transition from discrete barges or trucks or rail cars to continuous feed is done with a feed storage tank. Filling bottles is a discrete process that may be able to take a continuous feed. Such a process requires multiple filling nozzles, arranged so that a new bottle starts filling as a full bottle stops.

Product materials leave while materials are coming in

Continuous and discrete processes may be thought of as assembly lines. Input materials go through a series of operations and become exiting product while more material is coming in. Only a batch process contains all of the material that is being added and stops adding material before product is withdrawn.

A sequence of operations is required for normal production

A set of procedures performed in a prescribed order that may depend on current conditions is a sequence of operations. A discrete process example is the following:

If part is in place, lower drill to surface of part. Feed drill to specified depth.

Raise drill and signal completion.

A batch process example:

If reactor is ready, add a measured amount of ingredient A. Start agitator and heat to reaction temperature.

Add ingredient B until reaction is complete. Cool to a stable temperature.

Stop agitator and wait for permission to dump.

A continuous process example:

If time is up, divert process stream to dryer A for specified time. Heat and vent dryer B.

Cool dryer B.

Divert process stream to dryer B and repeat cycle with dryers reversed.

A recipe is required

All processes require a specification that defines how the product is to be made, including the quantities of raw materials and the specific processing sequence. The difference between the recipes of the three process types lies in the quantity and complexity of the details. A continuous process may have only a few specified products and does not call the specifications “recipes.” A discrete recipe may be called a “bill of materials.” The number of products that a discrete process can make is limited by the degree of specialization of the machinery. A wire folder cannot make washers. Almost anything that can

(34)

be homogenized can be made with a bucket and paddle, so the number of products possible from a batch process may be much larger than for either a continuous or discrete process. True, there are batch processes that are spe-cialized by their physical limits or auxiliary equipment, but a reactor that has a heat exchanger and agitator is able to make more diverse products than the other processes.

One or more production vessels require a mixing agitator

It is difficult to induce chemical change without some form of mixer. Few dis-crete processes want to mix parts. A continuous process usually relies on flow to do the mixing but may require the help of an agitator.

Output is multiple objects that can be individually labeled

The objects may look identical, but each could have a unique label. The size of the label is not a consideration.

Output is an unbroken stream of product

The output of a bottling line or a pill press may seem continuous because of its speed, but the output is composed of discrete objects. A stream can be gas or liquid or an extrusion of solid material.

Output requires that the production vessel(s) be emptied

The “vessel” in a discrete process is the fixture that holds the part being processed. It has to be emptied so a new part can be inserted and processed. Continuous distillation must not empty the tower during processing. About the only thing that gets emptied in an operating continuous process is one of a set of devices that sequentially receive the process stream, perhaps to dry it or to separate a crystal slurry from solvent in several centrifuges. A batch reactor is filled with reactants until the required processing is complete; then the contents are transferred for further processing or packaging.

Properties for Process Classification

The table presented in the preceding section shows just three properties that are not ambiguous:

Table 1-2 Three Unambiguous Properties

Property Discrete Batch Continuous

Output is multiple objects that can be individually labeled Yes No No Output is an unbroken stream of product No No Yes Product materials leave while materials are coming in Yes No Yes