The various control acronyms that take us through the evolution of process automation include: direct digital control (DDC), supervisory computer control (SCC) and advanced process control (APC). DDC marked the beginning of the age of the digital computer in the process control
world. SCC evolved out of the early DDC projects in response to the lack of tangible benefits realized from the early DDC projects. APC came of age with the introduction of commercial multivariable predictive control (MPC) products. Those of us who are veteran practitioners appreciate that APC has a history and would agree that it is a history worth recounting.
APC, like most (if not all) engineering exercises, has been, and will always be, a blend of art and science. As a specific engineering discipline, APC can be considered a subset of the general topic of process automation. As illustrated in Fig. 1, the art and science of process automation can be defined in terms of four paired concepts: function vs. form, technique vs. technology, solutions vs. tools, and people vs. machines. The logical union of the art and science shown in Fig. 1 represents the active universe of the practice of process automation.
Fig. 1. The art and science of process automation.
Art and science.
The art and science of process automation are about function in contrast to form, with function being what is to be done vs. the form of where it will be done. For process automation, the form is the computing device. Technique, as opposed to technology, draws the distinction between methods used vs. how those methods are implemented. In this case, technology is the basic software within the computing device. Solutions imply problems solved using the tools on hand. Consider tools as a collection of pre-programmed software functions. And, finally, it is about people as opposed to machines. Machines, in this context, are illustrative of the processes within the scope of the automation task vs. those individuals, the people, who are the users.
The general concept of process automation can be extended to the specific concept of APC with the use of the traditional hierarchy of the control pyramid shown in Fig. 2. A few words of clarification for this version of the pyramid: It shows a clear distinction between Level 1 and all those above it. Level 1 regulatory control is that portion of the control system required for the safe and stable operation of the process under normal operating conditions. In practice, this includes all of the single-loop controllers, as well as the safety-instrumented systems essential for safe and stable process operation.
Level 2 differs from Level 3 only in the number of variables included within the scope of the higher-level regulatory controllers. Level 4 differs from all those below it in terms of optimization being done in steady-state vs. the dynamics of Levels 1, 2 and 3 control. While the distinction may be somewhat a matter of semantics, it remains a useful one for purposes of discussion.
Finally, the arrows in Fig. 2 are meant to reflect the flow of information between the process and the APC system while Levels 2, 3 and 4 together comprise the concept of APC as presented in this discussion.
Fig. 2. Process control hierarchy.
For the Silver Jubilee issue (January 1970) of Instrumentation Technology, the readers overwhelmingly selected the digital computer as the most significant development during the previous quarter-century. Furthermore, virtually all the experts saw the role of the digital computer expanding in all aspects of process control.1
In the preface to his book, Automation: Its Purpose and Future, published in 1955, Dr. Magnus Pyke observed scientists do not, as a rule, distinguish themselves as commentators on general affairs. Indeed, the more distinguished they become as scientists the more they tend to restrict their thinking to their own concentrated field. This modern attitude of scientists arises in part from the fact that science has become so technical that its practitioners spend all their intellectual efforts in mastering the facts and techniques of their particular branch.2
In describing the new style of industrial work, Dr. Pyke goes on to point out that, early on, machines were developed to do away with the necessity of mechanical work. Electronic computers, however, do away with the necessity for mental effort. In recognizing some potential limitations on that score, he points out that computers are only machine tools and their usefulness depends on the skill and accuracy of the people using them.
With the arrival of the digital computer, the new style of process control became known as direct digital control, or DDC. In DDC, the computer calculates the values of the manipulated variables (like valve positions) directly from the values of the set points, controlled variables and other measurements on the process. The decisions of the computer are applied directly to the process. The early efforts to implement DDC reflected a time when the basic control system was analog, be it electronic or pneumatic. Thus, there was a clear distinction between the analog and digital parts of the system.
The first DDC systems were installed on a commercial basis in the mid-1960s. One of the earliest systems to be publicized was installed and used in the startup of a 500 million pound per year ethylene plant at the Phillips Petroleum refinery in Sweeny, Texas.3 Since this was a first-time installation of a DDC system, Phillips decided to retain conventional electronic analog control on critical loops and use DDC on all others. Critical loops were defined as those loops that would be difficult to control manually if the DDC system should fail. Of the 180 total control loops, two-thirds (120) were put on DDC.
The initial investment in DDC systems was justified on the basis of providing a lower cost alternative to conventional analog controllers. As the Phillips experience showed, this justification proved to be a myth. The lower cost of the analog controllers was offset by the higher cost for backup capability in case of DDC system failure. While intangible benefits were recognized, other more tangible sources of benefits were needed to justify the investment. Phillips found that the most promising area for additional benefits lies in the implementation of supervisory computer control, or SCC. While not included in the original project scope, SCC was implemented after the unit startup was completed.
The DDC system at the Phillips refinery was judged to have been a technical success from the beginning. It was available for plant start-up and was, in fact, essential to plant startup since initially the system was provided without analog backup. It failed, however, on the economics used for the initial justification of the system. The DDC system was made an economic success after the fact with the addition of SCC functions using the DDC system as a base.
SCC was born out of the need to provide tangible benefits for the use of the digital computer within the process control hierarchy. As demonstrated by Philips success, the efficacy of SCC was established early on for not only new plants but also as an add-in technology for existing plants. The success of SCC through the 1970s and into the early 1980s is well documented. Both the art and science developed rapidly as computers became more powerful and interfaces to a variety of instrument systems were developed. The introduction of real-time database software, along with support of higher-level programming languages, became powerful tools in the hands of the practitioner. The creativity of the engineer was put to good use implementing solutions for a variety of control problems that could not be addressed in the analog control system.
While SCC progressed rapidly, the Phillips project team gave a word of caution. This is a direct quote from the project team: Implementation of supervisory and optimizing control has not progressed as rapidly as technology would permit; and, as a result benefits have been delayed. This is due, in part, to the time required to build up operations confidence in the system. Is this perhaps a lesson we are still learning today?
The next big thing in SCC came along with the commercial introduction of the multivariable predictive controller (MPC). IDCOM and DMC are two of the more notable that achieved early commercial and technical success. It was during this time, through the marketing campaigns of the MPC suppliers, that the moniker of APC came into general use. And again, many successful applications of MPC have been noted in the literature.
While MPC achieved demonstrated success as an integral part of the APC system, the experience with optimization was less than stellar. Mr. Latour gave an early optimistic view of online optimization in this magazine in 1979, along with a number of applications that showed promise.4,5 With a few exceptions, the concept has never quite lived up to its potential.
Things have come a long way in terms of the new style of process control. The overall efficacy of APC is well established and has been thoroughly documented. At the same time, cases where expectations were not met and systems failed to deliver a sustained rate of return are also well known. An assessment of the state of APC was given by Mr. Wang in the July 2011 issue of this magazine.6 It is disheartening to note that he estimated that more than 50% of APC applications are in off mode or do not work at all. He goes on to say that only about 10% are fully working. There is a plethora of opinions as to the reasons for this; two of the more impassioned are cited here.7,8 What I find particularly interesting about these gentlemens views is that they are as much about the business of APC as about the technology of APC.
I think everyone would agree that much has changed in the evolution from DDC to APC. We now work within a totally digital world with ever more powerful machines with limitless capability to, as Mr. Pyke pointed out, do away with the necessity for mental effort.
Further overlap and blending.
Keen observers of the stock market say the trend is your friend. To me that suggests that it is important to not only understand where you are at but to also appreciate how you got there. I submit that a better appreciation of the blend of art and science, and more importantly how it has changed over time, will give us insight into where we might be going. To that end, an informal and unscientific survey was conducted asking a number of veteran APC practitioners to complete an exercise comprised of a series of questions. The questions were:
1. What is the relative contribution of the art vs. the science of APC?
2. At what level in the control hierarchy has the efficacy of APC been demonstrated?
3. How has this blend changed over time?
In terms of Fig. 1, consider the answer to the first question being described by the relative sizes of the two circles. The second question would define the degree of overlap of the two circles, with the higher up on the control hierarchy as defined in Fig. 2, the greater would be the extent of the overlap.
As might be expected, the opinions expressed in the survey were as varied as the number of people who participated. However, there was some degree of consensus concerning a trend and the reasons behind that trend.
The results of the survey suggest that, in the beginning, it was all about the science (maybe a 10/90 split between art and science). Since DDC had never been done before, the engineers were making it up as they went along. As the Phillips project showed, they did a pretty good job. By 1985, the split had shifted to about 50/50 and the overlap was growing as MPC began achieving its early success. Today, the split is estimated at 90/10, a complete reversal from the beginning, with the overlap about the same as it was in 1985.
For me, the contemporary view gleaned from the survey seemed, at first, counterintuitive. The science has certainly improved with the evolution of more powerful commercial toolkits for implementing the MPC and optimization components of the APC system. But, paradoxically, as more of what used to be the art is now captured in the science, the problems that can be solved and the techniques required to define the solution have become ever more complicated, requiring ever more creativity to be successful. The majority of the reasons for this current view are related to the people (in terms of the skills of the APC practitioner), the motivations of operations personnel and the support of management. These are all issues that have been in play since the beginning but have now become the determinate factor in the success or failure of APC.
As a group, the survey participants remain optimistic about the future. While much has been accomplished and much remains to be done, the journey continues to be an interesting one. In contemplating the future, a quote by Charles Lyell from 1863 seems to be apropos. He wrote: The rate of progress in the arts and science proceeds in a geometrical ratio as knowledge increases, and so, when we carry our retrospect into the past, we must be prepared to find the signs of retardation augmenting in a like geometrical ratio.
The optimist believes that the overlap of the circles will continue to grow. The realist may view the future as Mr. Lyell sees it, with the overlap staying about the same. There will always be the pessimists among us who view the whole thing falling apart. Personally, I choose to remain an optimist. HP
1 Smith, C.L., Digital Computer Process Control, Intext Publishers, New York, 1972.
2 Pyke, M., Automation: Its Purpose & Future, The Scientific Book, London, 1955.
3 Parsons, J.R., M.W. Oglesby, and D.L. Smith, Performance of a direct digital control system, ISA Conference, Philadelphia, Pennsylvania, October 1970.
4 Latour, P.R., Online computer optimization 1: What it is and where to do it, Hydrocarbon Processing, June 1979.
5 Latour, P.R., Online computer optimization 2: Benefits and implementation, Hydrocarbon Processing, July 1979.
6 Wang, J., What is the outlook for advanced control engineering? Hydrocarbon Processing, July 2011.
7 Latour, P.R., Demise and keys to the rise of process control, Hydrocarbon Processing, March 2006.
8 Friedman, Y.Z., Has the advanced process control industry completely collapsed? Hydrocarbon Processing, January 2005.
|The author |
Michael Delaney is a chemical engineering professional with over 30 years of experience in the field of process automation and control. He has practiced the art and science of APC with Standard Oil of Ohio, Setpoint, Honeywell, Pavilion Technologies, Mustang Engineering and BPEC Consulting. He is currently a senior consulting engineer with ProSys, Inc.