Advances Beyond Imagination
Sporting more than 50 years in the automation and control industry, Dr. Irving Lefkowitz has seen it all when it comes to the modern advancement of industrial technologies. His perspective offers valuable insight for those pondering the future of the engineering profession.
Dr. Irving Lefkowitz is professor emeritus of systems engineering at Case Western Reserve University and vice president and cofounder of ControlSoft Inc.— company that develops and markets software solutions for advanced industrial control applications. Over the past 50 years, he has contributed to advances in process dynamics, advanced control, and computer-based integrated and hierarchical systems control, specifically focused on their applications in industry. In recognition of his lifelong commitment to the field of automatic control, he was awarded the Control Heritage Award of the American Automatic Control Council.
At Case Western Reserve University, he served as chairman of the Department of Systems Engineering and also director of the Control of Industrial Systems Research Program, which was supported by a consortium of industrial concerns. He also served as: co-director of the Integrated Industrial Systems Project at the International Institute for Applied Systems Analysis in Vienna; United Nations Industrial Development Organization consultant on several advanced industrial control projects in India; and U.S. Agency for International Development consultant on computer control applications projects for Egyptian industry. He has authored numerous papers and lectured and consulted extensively on computer process control in the U.S. and other countries.
Dr. Lefkowitz holds an M.S. and Ph.D. in systems and control engineering from Case Western Reserve University, and a B.S. in chemical engineering from the Cooper Union School of Engineering.
Furthermore, Dr. Lefkowitz has a history with Control Engineering magazine. Along with Don Eckman—founder of the CaseSystems Research Center and initiator of the University’s Control of Industrial Systems Research Program—Lefkowitz co-authored “A Report on Optimizing Control of a Chemical Process” in the September 1957 issue of Control Engineering. The research conducted by Eckman and Lefkowitz demonstrated the feasibility of using a computer, in real time, for optimizing control of a batch chemical re-actor. Specifically, the computer solved a set of differential equations, which were derived from a mathematical model of the process, so as to force the reactor composition to follow a minimum processing time trajectory going from a measured initial composition to a specified end composition. Also demonstrated was the effectiveness of the approach when the model only approximated the reaction kinetics. This was achieved by a scheme of repetitive control where the optimal trajectory was recomputed at every sample time based on the current measured composition value. Test results carried out on a laboratory-scale chemical reactor showed reductions in processing time of the order of 23% compared with a conventional control strategy.
What do you consider to be the top three advances in control and automation over the past 50 years? Which do you consider the most important?
My choices are:
The phenomenal advances made in computer technology resulting in exponential increases in speed and memory capacity, and corresponding reductions in hardware costs. This, coupled with corresponding advances in software capabilities, made economically feasible the widespread applications of the computer to industrial process control and automation.
The development of a systems approach to the control of complex industrial systems, incorporating concepts of distributed control and systems integration.
Blurring of the distinctions between continuous process control and discrete manufacturing—the coming together of the PID world and the PLC world.
I consider advances in computer technology to be the most important of the three because it provided the means for the commercial realization of my other two choices. In addition, it provided the engine that drove the commercial feasibility of automation and inte-grated industrial systems control. In fact, without the developments in computer hardware and software, I doubt whether there could have been the range and extent of commercial applications of large-scale integrated systems control and automation that we’ve seen in both the manufacturing and processing industries.
What do you think will be the next significant advances in control and automation in the next 5 to 10 years? In the next 50 years?
The following areas seem promising for promoting significant advances in control and automation in the near term:
Automated systems diagnostics for processing and manufacturing plants. This is important for management’s ability to monitor and evaluate plant performance in real time, for effective preventive maintenance scheduling, and for rapid fault detection and remedial actions. As plants become more highly integrated and computer control systems more complex, overall plant performance becomes increasingly vulnerable to malfunctions of sensors and actuators, breakdowns in processing components, and major changes in the interactions with external systems.
Further developments in the area of smart, solid-state-based sensors. These developments have the promise of broadening the scope and improving the quality of the infor-mation provided to the computer control system, thereby improving system reliability and overall economic performance.
More extensive applications of advanced control concepts (for example, model-based control and predictive control) to improve control of processes with poor response characteristics.
More extensive and effective applications of computer control and systems integration so as to further improve economic performance, quality control, system adaptability to changing conditions, and system reliability and fault tolerance.
When asked to “crystal-ball gaze,” say, 50 years into the future, I am reminded of an experience I had some 50 years ago. I was invited to participate on a national panel using the Delphi method, which was then considered a hot new tool for prediction of future events and trends. This was in the early days of commercialization of digital computers. The objective of the panel was to forecast trends in computer technology and potential applications of computers in industry, commercial services, the home, et cetera. What we all considered wild-eyed predictions fell so far short of the developments that actually came to pass, that I vowed never again to prognosticate on long-term predictions of technological advances.
Looking back over your career, have there been any surprises with respect to advances in the field of automation and control?
I could cite a number of “surprises,” but the one that sticks out in my mind goes back to the early 1950s. The new commercial availability of digital computers for scientific (as opposed to just data processing) applications gave rise to much armchair philosophizing about the advent of complete plant automation and manless plants. In general, the idea was that a central computer would automate and control all aspects of the operation of an industrial plant so as to maximize such economically based objectives as productivity, operating efficiency, and product quality. All the relevant information about the plant would be transmitted to the computer, which would determine, by appropriate algorithms and logic-based rules, the control signals needed to be transmitted back to the plant to achieve the desired objectives.
Early digital computers were not designed for real-time applications; however, a number of small machines were being developed specifically for control of industrial processes. The RW 300, in particular, was actually applied to supervisory control of chemical reactors with some success. However, this machine was an 8,000-word drum machine—a bear to program and very limited in the size and complexity of problems it could handle. New machines were developed, each bigger and faster than the previous generation, culminating in huge mainframes, exemplified by the GE 4060, with increasingly larger and more complex plant applications.
What surprised many of us in the field was that this evolution stopped so suddenly. The overall goal was still there, but the market decided that bigger main frames were not the answer. The hardware was becoming prohibitively expensive but, more significantly, the costs associated with programming, debugging, and updating control algorithms often exceeded even the high cost of the hardware.
All this changed with the development of the microprocessor and subsequent commercialization of the minicomputer. The new hardware was much faster, more compact, and much less expensive. An accompanying evolution in software made programming, debugging, and updating easier and faster. This significantly cut the time and manpower costs of implementing a computer control system. Finally, there was the development of new ap-proaches to the control of complex systems, based on decomposing the overall complex plant system into more easily handled subsystems controlled and integrated by a hierarchy of computers, each tailored to the specific control and decision-making tasks required.
One of the early and very successful implementations of the new technologies and systems concepts was in the steel industry. An entire steel mill was controlled by a system of computers distributed among the various processing units, organized hierarchically, and integrating decision and control functions that ranged from direct operational control to scheduling and planning functions—thus approaching the goals of the early visionaries.
What should engineers in process control and automation focus on to advance their careers?
I believe a major challenge facing manufacturing engineers is the ability to design and implement industrial production processes and manufacturing systems that are not only more completely integrated and automated, but also more adaptable to changes in demand, economic factors, processing technologies, et cetera. It is only by this means that industrial enterprises can maintain a competitive edge, particularly with respect to foreign competition that benefits from lower wages and operating costs and fewer constraints.
To meet this challenge, an engineering team responsible for a particular project must, collectively, be reasonably conversant with basic systems engineering tools including modeling from first principles, systems dynamics and automatic control, optimization and decision-making techniques, computer simulation of continuous and discrete systems, methodologies associated with large-scale systems integration and control, and computer hardware and software relevant to the application of these tools.
Please provide a favorite example of how you have been an advocate for change in the field of control and automation.
I directed the Control of Industrial Systems Research Program at Case Western Reserve University for almost 30 years, starting in 1954. We played a pioneering role in advancing the concepts and technologies that contributed to the present advanced state of computer control of industrial systems. The research embraced studies in process dynamics, advanced control techniques, computer-based process control, distributed and hierarchical systems control—all specifically focused on industry applications. A consortium of diverse industrial concerns representing computer and control manufacturers, chemical and oil companies, steel plants, paper, rubber, and others funded the program and helped motivate its directions and priorities. It was a very exciting experience to be at the forefront of a newly evolving field and to see it develop from concept to wide-ranging implementation.
For more information, visit the following sites: