The following tip is from the ISA book by Greg McMillan and Hunter Vegas titled 101 Tips for a Successful Automation Career, inspired by the ISA Mentor Program. This is Tip #66.

101 Tips for a Successful Automation CareerThe automation profession requires knowledge of thousands of details on hardware, software, interfaces, and process applications. The diversity of the industries served and the natural inclination of people to think their case is special often lead to the conclusion that each plant and application is unique. Automation engineers are particularly vulnerable because of overwhelming details to “not being able to see the forest because of the trees.” Maybe because of my orientation as a physicist, I am always trying to find some commonality via a concept based on a first principle. I use applications of the principle and user feedback to either confirm or update the concept, an essential part of the scientific method and the reason that humans have advanced compared to animals (except for the special case of fraternity house beer blasts).

A good example of the power of concepts involves deadtime (Tip #70). With the guidance of Greg Shinskey, I found that PID tuning and errors for load disturbances could be simplified to simple functions of deadtime (Tip #71). Hundreds of equations developed over the last 60 years reduce to these relationships when the objective is load disturbance rejection. Then I realized that a near-integrating process gain could be universally applied if a deadtime function block was used to identify the process variable (PV) ramp rate. The initial PV responses of self-regulating, integrating, and runaway processes all look like a ramp to the PID. The deadtime block even enabled the unexpected extension of the near-integrator ramp concept to loops where the loop deadtime is much larger than the time constant (as noted in the July 19, 2012 Control Talk blog “Deadtime Dominance Does Not Need to Be Deadly.”)

Subsequently, I found that the same deadtime block could be used to compute the slope of a batch profile for the optimization of feed rate and cycle time. I also discovered that the block could compute a future PV value for a full throttle batch and start-up response. Greg Shinskey clued me in to the concept of simply using a setpoint filter with the time constant equal to the reset time to eliminate overshoot on setpoint changes, enabling the use of load disturbance tuning. My realization (that of external-reset feedback) enabled the use of direction setpoint rate limits to achieve a multitude of other process objectives, including abnormal situation prevention, coordination, and optimization with retuning (Tip #72). Furthermore, this external-reset feedback as part of an enhanced PID could, without retuning, handle large wireless and analyzer delays, stop limit cycles, and eliminate interactions. Control engineers can move on from tuning to taking advantage of a huge step increase in PID capability (Tips #89-97).

I did a batch project just to see how different the batch world is from the continuous world, which is the subject of university classes and most control theory textbooks. The essential aspects are an integrating process response from a closed liquid discharge valve during the batch and the use of sequences. I then realized that a continuous process can be treated as a near-integrator when running and treated as a batch during start-up. Sequences used to automate batches could be used to automate start-ups of continuous processes.

Concept: Conceptual knowledge should be sought to see the commonality in seemingly different application requirements. Personal pride in the uniqueness of solutions should take a back seat to the elegance of simplification and unification, which will extend the capabilities of the automation profession. Conceptual knowledge should be supplemented with procedural knowledge as a guide for effective implementation of solutions based on the knowledge.

Details: Conceptual knowledge is best developed in the time domain from trend charts, thought experiments, and from the principles of material, energy, and charge balances in Appendix F. The Control Talk blog “Where do Process Dynamics come from” discusses how these balances can lead to a better understanding of process dynamics. For example, all of the individual ionic equilibrium calculations for pH in a half dozen textbooks can be simplified to a single charge balance equation that is solved by the simplest of all search techniques: “interval halving.” Use the checklists in Appendix C to help provide procedural knowledge for effective implementation.

Watch-outs: You have to keep an open mind and avoid preconceptions. You should look for cases that provide exceptions to the concept along with cases that provide evidence that support the concept. The developers of new tools are (unintentionally or intentionally) often looking for proof that their creation is valuable. The use of tests to prove the value of an innovation is commonly seen in papers on control algorithms. Better tuning, tests on unmeasured disturbances as process inputs, and the use of key PID features may have considerably reduced the perceived advantage of an algorithm. Because loop performance depends on tuning (Tip #71), you can tune a controller to prove any point.

Exceptions: The first requirement is to get the job done right and on time. Flashes of insight that reveal concepts for unification of diverse and seemingly contradictory relationships cannot be predicted or scheduled.

Insight: Principle-based conceptual thinking frees you to move to a higher level of accomplishment.

Rule of Thumb: Seek to find the commonality rather than the uniqueness of solutions.

Pin It on Pinterest