Stan: Many processes would be operating far below their capability if it was not for consultants with extensive practical expertise guided by a sound fundamental basis in the math and first principles of process control to see through the complexity often introduced by local excursions into patchwork solutions that are at best a distraction to the simple control solution. It is like these consultants have an overhead view of the maze, seeing the clear path to the end avoiding the dead ends. These people do not get the recognition deserved and their lessons are often lost over time. In Control Talk we are doing our best to showcase the achievements of individuals and capture the understanding gained the hard way by working in the plant and making sure the solution is sustainable and maintainable. It is our way of saying “Thank You”.
Greg: An exceptional example of this type of consultant is Sigifredo Nino, the founder and owner of Summa Control Solutions. Sigifredo worked five years at a coal-fired power station (EEB), 10 years with a pulp and paper mill (Smurfit), 13 year with Foxboro (pulp and paper, power generation, mining, industrial boiler, oil refining, petrochemicals), and for the past two-and-a-half years as an independent process control consultant with Summa Control Solutions, Inc. (oil refining, mineral processing, power generation, petrochemicals).
Stan: What is your philosophy and general approach?
Sigifredo: Process control is not just about the technology, but also the ability to question its soundness and the exertion of good criteria to either use it or reject it.
Process control is a very mature discipline, however, there are fewer and fewer people who really know what it entails to design a sound control strategy or even how to properly tune a PID controller. Superfast computers and fancy mathematical constructs will never replace the thorough understanding of engineering fundamentals and applied control theory.
While many authors have been trying to exclude mathematics and equations to make the field more “accessible,” I do actually tend to go in the opposite direction along with Roger Penrose: “The understanding that we have of the principles that actually underlie the behavior of our physical world indeed depends upon some appreciation of its mathematics.” (The Road to Reality – A Complete Guide to the Laws of the Universe, Jonathan Cape, London, 2004).
My approach to process control problems, paraphrasing Einstein: I always look for the simplest possible solution, but not simpler. Simple solutions work better and stay forever in service … except during the times when a field device fails, or there has been a change in the process, e.g., a bigger or smaller impeller in a pump.
Innovation is not only about new inventions, but also about using a technology existing in a field and then to apply it to another one. Innovation doesn’t mean to forget the fundamentals, and this is something I believe has happened to process control.
Finally and siding along with Greg Shinskey, who said: I have never found a problem I have not been able to resolve by using advanced regulatory control and PID algorithms; and control strategies that are not understood by the operators will be put out of service following the first malfunction.
Greg: My video recorded interview with Greg Shinskey at his home in New Hampshire featured in a special 2011 ISA Automation Evening session included a brief synopsis by Nick Sands, Terry Tolliver and me about what Shinskey meant to us in our careers. In the interview, Shinskey mentioned he was extremely proud of a protégé who was applying what he had learned from his books. That protégé turned out to be you. How did your relationship with Shinskey get started and progress?
Sigifredo: Since meeting Greg Shinskey in November of 1993 in Atlanta, when I attended to his “Process Control Systems” seminar, I have successfully implemented several of the control strategies presented in his books. And over the past seven years I have been privileged to work very close to him in several projects involving combustion controls, oil refining and petrochemical processes.
What has impressed me the most is seeing how his profound understanding of the chemical engineering fundamentals and control theory allow him to come up with brilliant and yet simple and functional practical designs.
Probably the first thing I learned from working with Shinskey is that the most important part in process control is the process. You can’t design or tune up a control strategy by treating the process as a black box.
Also from Shinskey I have learned to weigh the “unintended consequences of the added complexity” in a control strategy; to make sure that the benefits counterweigh the additional intricacy.
I think that Greg Shinskey’s work has not had the acknowledgment and dissemination it deserves, and I may venture to say that one of the reasons is that if any of his books was going to be used in a course, it would require from the instructor to have a very broad hands-on experience in process control in order to be able to successfully teach the concepts and theories presented in his writings, and that kind of knowledge does not normally exist in the academic environment, and it’s getting scarcer in the industry as well.
Greg: In our phone conversation, you mentioned one of your many successful solutions that standout was a power boiler. What was the situation you faced?
Sigifredo: The power boilers had been converted from fuel oil to natural gas, and this changed the combustion dynamics. The steam pressure and temperature were operating far below the desired setpoints, and the steam generating capability could not be ramped as fast as needed. The result was more steam flow was used than needed and the boiler could not meet shifting demands, reducing the ability to sell electricity. The inability to keep the steam temperature within acceptable limits and to raise the setpoint was the culmination of the problem. Not only was there a loss in boiler efficiency and sale of electricity, but there was also premature aging of the steam turbine blades.
Stan: What had been done previously in an attempt to solve the problem?
Sigifredo: Various specialists had tried a patchwork of feedforward strategies that resulted in more problems from positive feedback and counterproductive complexity, undermining the operators’ confidence in the control system.
Stan: What were some details of the solution and importance?
Sigifredo: The natural gas control valve with a slow actuator and excessive backlash was replaced with a v-ball valve with a digital positioner. A dead time-compensated, interacting PID controller was used for the steam temperature loop with all the modes and the dead time adapted as function of the steam generation rate, plus an air flow feedforward. This power station is what is called spinning reserve, so readiness to change the megawatts (MW) is key in the ability of the power station to sell its production.
Greg: The ability to put a dead time block in the external reset feedback path of the PID is one of the many benefits of the positive feedback implementation of the integral mode. Only the dead time needs to be set in the block to achieve dead time compensation. In comparison, a Smith predictor requires the process gain and time constant to be accurately set as well. There are many other benefits from this implementation of the integral mode, such as preventing a burst of oscillations from a slow secondary loop in cascade control as discussed in the May 2006 Control article “The Power of External-Reset Feedback.”
Sigifredo: The customer was really impressed by the accomplished results, although somewhat concerned about the permanence of the solution. My advice was, “We have been running this strategy for over two months now without a single instance of instability or lack of performance. If it ever stops working fine…don’t blame the control strategy or the tuning. Verify that the field devices and equipment are working fine; e.g., RTD and attemperator.”
Greg: What general guidance can you offer for steam temperature control?
Sigifredo: An advanced regulatory control strategy like that outlined above is the best you can have. Model-based control strategies are not required and will not outperform the former.
Stan: What other boiler applications have you worked on?
Sigifredo: Three of the most important strategies I have used are:
i. Combustion controls: Dual fuel firing cross limiting including fuel ratio and limit on either fuel; and heat release combustion control strategy for CO boilers
ii. Three elements drum level using gain-only controller with the steam feedforward connected to the external feedback connection in a positional PID controller, so the integral time is used as a dynamic compensator.
iii. Four elements level control using drum pressure to provide shrink and swell compensation, as proposed by Shinskey in his process control systems book.
Greg: What is different about boiler control problems?
Sigifredo: Boilers are based on a very mature technology that is highly integrated with lots of interaction among the different subsystems, including distributed lag processes. An entire course in process control can be given around boilers starting with single loops and progressing to advanced regulatory strategies.
Stan: What is your general approach to solving boiler control problems?
Sigifredo: Boiler control strategies tend to be built with a lot of unnecessary complexity. I have more often than not succeeded in improving the overall performance by decommissioning the troublesome add-ons and eliminating positive feedback loops created by some of those strategies.
Greg: To provide a humorous twist to this column of thankfulness we have the “Top 10 Thank You Notes”.
“Top 10 Thank You Notes”
10. Thank you for batch processes and highly exothermic polymerization reactors so I had to learn to minimize overshoot and peak error for integrating and runaway processes and address a window of allowable PID gains where a PID gain that is too low or too high can cause severe oscillations.
9. Thank you for the logarithmic relationship between pH and hydrogen ion activity and the process temperature effect on dissociation constants, so I had to learn how to deal with the ultimate nonlinearity and discover hidden truths on the effect of ionic strength, water concentration and process temperature on the actual solution pH.
8. Thank you for on-off piping valves posing as throttling control valves so I have a mission in life to expose a 50+-year mistake of extraordinary proportions.
7. Thank you for compressor surge and RCRA pH limits so I had the motivation to develop an open-loop backup triggered by an extreme measurement rate of change.
6. Thank you for signal filter and sensor time constants larger than the process time constant so that I have great stories about attenuation and deception where a slower measurement makes the trend chart look smoother while the process gets worse.
5. Thank you for suppliers of pressure transmitters who use a 1-second default damping setting so that the compressor shuts down when a transmitter is replaced in the middle of night, causing users to buy transmitters from my favorite supplier with 0.2 second default.
4. Thank you for suppliers who proudly promoted a pH electrode with a 2-minute, 86% response time so that effluent diversions to avoid pH RCRA violations cause users to buy electrodes from my favorite supplier with a 0.1 minute 86% response time.
3. Thank you for unmeasured disturbances on the input to a process with an integrating process gain or large process time constant in nearly all of the most important loops in my career, such as pH and liquid concentration and temperature control and gas pressure, so that the performance of a PID is nearly optimal, and I do not have to pay attention to the 1001 controller algorithms developed by graduates students over the past 50 years.
2. Thank you for the near integrator approximation for self-regulating processes with a large primary process time constant so that equations and associated egos behind the 101 tuning rules so passionately advocated take a back seat to improving the loop.
1. Thank you most of all for dead time and unmeasured disturbances. Otherwise I would be out of a job.