Getting innovation back into process control

Virtual plants and online metrics can capitalize on the synergy between modeling and control to unleash engineers' imaginations to push the process control envelope

By Greg McMillan and Stan Weiner

Greg: I don't have the statistics, but I have the general impression that process control innovation is on the decline. We are into a copy approach to automation projects with an objective of just keeping the plant running. This step backward can be attributed to the retirement of specialists, downsizing of corporate and plant staffs, overwhelming emphasis on project schedule and cost, lack of technical resources that provide practical guidance on the implementation and value of process control improvements (PCI), and an attitude that the profession is mature, and it has all been done before. In reality, new advanced control tools and an increase in PID and tuning software functionality can empower the individual to the point where imagination is the limit. We conferred with Todd Jaco, a recent graduate of the Missouri University of Science and Technology (Missouri S&T) and new business manager for Mynah Technologies to see what we can do to turn this around and open up minds. Todd has extensive practical experience as a simulation consultant and is a great communicator with exceptional interpersonal skills.

Stan: When did you become interested in process control?

Todd: In the Chemical and Biochemical Engineering Department of Missouri S&T, there were very practical courses on the application and importance of process control. The focus at Missouri S&T is preparing engineers to be immediately productive in the process industry. The partnering with companies such as Emerson, Monsanto and Mynah has strengthened this approach, as seen in the new Bertelsmeyer Hall with a state-of-the-art virtual plant and unit ops lab with an industrial distributed control system (DCS). Most  Mynah employees and interns are from Missouri S&T. My interest in process control started at Missouri S&T and was fostered by Greg's Series on Dynamics with demos by Pierce Wu using Mimic simulations to make control technology come to life. This lead to numerous questions and a motivation to see what more can be done. The series was supported by ISA book 101 Steps for a Successful Automation Career, which helped detail more of the practical aspects of the job, including how the knowledge translated to better automation project execution.

See Also: Prototyping for small and large APC applications

Greg: What do you see as the next step?

Todd: We plan to develop the ability to provide the customer with online process metrics. These metrics can show what the process and automation system is currently doing and how well the operator is managing and improving the process operating conditions. We can put more detail into the automation system model as part of a virtual plant to show how knowledge of automation system design and process objectives can affect the metrics, using the new Momentum Press book, Tuning and Control Loop Performance – 4th Edition, as a guide. This will open the door to seeing how automation details such as analyzer cycle time, measurement span and update time, sensor lag and transportation delay, valve size, flow characteristic, stroking time, resolution and dead band affect tuning and performance. The chapters on process control improvement and batch optimization can be employed to see how control strategy innovations can improve plant performance.

Stan: What do you envision as process metrics?

Todd: I would start out with metrics of process capacity and efficiency on an individual unit operation basis. The equipment component outlet, raw material and utility flows would be totalized over various time intervals. Shorter time intervals would be used to provide a more current view (e.g., last hour or batch phase) and longer time intervals would be used to provide a longer-term view (e.g., last shift or last batch). The time interval would start at the beginning of raw material and utility flows (e.g., start-up of continuous operations and the beginning of a new phase for batch operations). The component outlet total divided by each individual raw material total provides a metric of raw material efficiency (e.g., yield), and the component outlet total divided by a utility flow total provides a metric of energy efficiency. The utility flow can be converted to an energy usage rate to give metrics in units of kilograms per kilojoule. By using totals over sufficiently large intervals (e.g., time intervals larger than equipment residence time), the synchronization issue is largely avoided until the metric is extended to process areas.

Greg: I expect each flow would be averaged over a time increment (e.g., one minute) to smooth out short-term changes and errors in synchronization, and to reduce the array size for totalization. The oldest flow in the array is subtracted from the total as the newest flow is added to the total after each is multiplied by the time increment and the phase speed-up factor. Synchronization of input flows with the product output flow could be achieved by passing input flows through dead time and filter blocks corresponding to equipment delays, lags and residence times. How do you introduce the user to these capabilities?

Todd: We plan to offer a half-day extension to our short course to cover metrics and a hint as to how PCI can play an important role to get the user interested. As interest grows a one-day basic and two-day advanced course on PCI would be offered. A key part of these courses would be the use of advanced modeling objects to bring home the possibilities.

Stan: What do you expect are some of the PCI opportunities?

Todd: The most straightforward opportunity is the simple change of a loop setpoint to gain a more productive or efficient operating point. Good process simulations can reveal this opportunity. The setpoint improvement may be done up-front based on better process knowledge by exploring the operating region. If reduction in variability is needed to get closer to the optimum, greater attention in the model to automation system design and dynamics can demonstrate these possibilities of better control strategy and tuning.

Greg: The simple addition of a valve position controller (VPC), which involves a simple configuration change to include a PID acting as a VPC, can be used to optimize feed and utility setpoints automatically to increase efficiency and capacity, as described in the Control November 2011 article, "Don't Over Look PID in APC." The optimization can be easily and bumplessly turned on and off by switching the VPC between the manual and auto modes. The tuning is simplified by the use of external reset feedback and directional velocity limits on the optimized setpoint. The book, Advances in Reactor Measurement and Control, details the use of VPC for various types of reactors. Opportunities to optimize batch operations are more extensive and often neglected because the focus of advanced PID control and model predictive control has been primarily on continuous operations. The Control July 2008 article, "Unlocking the Secret Profiles of Batch Reactors;" the Control September 2012 article, "Get the Most Out of Your Batch;" and the 8/28/2014 blog, "Batch Control Optimization Recommendations," provide a spectrum of batch opportunities and show how a virtual plant is the key to finding and testing them.

Stan: How do you expect this initiative to progress?

Todd: We will encourage and help users to present and publish their applications so that we can all learn from their successes and difficulties. Application examples and demos will be given to Missouri S&T and the Rose-Hulman Institute of Technology for use in their well-established, extensive process control labs. These virtual plants can increase the communication between industry and academia, and motivate new engineers to look for process control opportunities, as described in the October 2013 Control Talk column, "The Education of Future Automation Engineers." These applications can provide the momentum to eventually have user group sessions on the use of virtual plants for PCI.

Greg: I am excited about the potential for not only reversing the trend, but also putting more process knowledge into control knowledge, and vice versa. I come from a culture and heritage in Monsanto's Engineering Technology Department where modeling and control were inherently linked. A good portion of my deeper understanding of process dynamics (summarized in the Control December 2014 white paper, "First Principle Process Relationships" that led to process control improvements was gained by doing simulations. You can quickly explore possibilities in a matter of hours that would take months in a real plant. In many cases, operating plants no longer allow such exploration. A simulation is needed to build the case and work out the details before improvements are implemented. More recently, I have developed a Bioreactor Model running 1,000 times real time in a virtual plant for knowledge discovery, process control development and exploration that can lead to more repeatable batches. Todd has opened his mind to the incredible possibilities offered by the increased power and flexibility of simulation software and the DCS. Have you?

Greg: We conclude with the Top 10 Uses of a Virtual Plant in a University

Top 10 Uses of a Virtual Plant in a University

(10) Predict the questions and answers on the next exam
(9) Develop "fantasy control," kind of like "fantasy football"
(8) Develop metrics like "time to complete lab assignment"
(7) Optimize capacity: "credit hours per semester"
(6) Optimize efficiency: "credit hours per hour of time invested"
(5) Optimize quality: "grades"
(4) Explore and prototype new opportunities;: "internships"
(3) Have more fun than a Sunday afternoon at Cracker Barrel
(2) See if you can stump the professor
(1) Demonstrate production of crafty beers