Distributed Control

Why all the hatin' on advanced process control?

Is it too hard, too complicated, too expensive, or does it just have a bad rap? Maybe all it needs is better marketing.

By Greg McMillan and Stan Weiner

Stan: I have gotten the distinct impression from a number of users that the term "Advanced Control" is viewed negatively. I was wondering in general who feels this way (e.g., plant management, process design, operators) and why (e.g., support, understanding, training, overload).

Greg: I wonder if the term "advanced" is a detractor. ISA inserted the term "advanced" in the title of the next editions of my pH book (Advanced pH Measurement and Control) and Astrom's PID book (Advanced PID Control) back in the 1990s and sales fell off. To be consistent with title of my pH book, I included the advanced in the title of my Temperature book (Advanced Temperature Measurement and Control). I regret the use of the word "advanced."  In my new book on reactor control, I avoided the use of the term "advanced," but maybe the term "advances" is too close as well (Advances in Reactor Measurement and Control). The good thing is that the titles appear early in the alphabetical order on the ISA "Find Books" webpage.

I think in the 1980s, "advanced control" was looked on favorably, and maybe it still is in some industries. I think it will be useful and interesting to figure out what happened.  Toward this end I have asked the ISA Mentor Program participants Brian Hrankowsky, Danaca Jordan, Leah Ruder, and Hector Torres and the co-leader Hunter Vegas for their perception.

Brian: For someone like me, "advanced" means more value, cutting edge, juicy technology and the opportunity for learning.

For many of my peers and certainly for the process engineers, industrial engineers, electrical engineers, operators, it means "too complicated," "no one will understand," "no one else will be able to support," "will cost a lot and be hard to justify" and "We've gone this long without needing advanced controls so we don't need it."

For example, in buildings where override, cascade, ratio and feedforward control are just part of process control, they are viewed as simple. In buildings where they are lumped in the term "advanced controls," they are exotic and avoided. I can usually get past the cascade issue when I point out the valve positioner in the loop is already a cascade and so they are found everywhere.

This may also stem from the college training that people are getting. I have had to explain that proportional only (P) control is not less stable than proportional-integral (PI) control (that offset is not instability) and many other basic items along those lines. So for people who think PID is magic already, advanced control sounds like sorcery. I think it would better if when they learned unit ops, they also learned the various control strategies and optimizations described in Liptak's Instrument Engineers Handbook.

Similarly, "fuzzy logic control" (FLC) did a poor job of advertising itself. People (even smart people in controls) think it is not able to be validated because it is not deterministic – which is total false.

Greg: "Fuzzy logic" is an example of another term with negative connotations. There are times when FLC is the right solution as discussed in the Nov 2012 Control Talk Column "Ruel Rules for Use of MPC, PID, and FLC.

When I made a presentation on how the addition of a PID controller to an ethanol plant as a front-end production rate controller (i.e., flow controller) could significantly reduce corn raw material cost (largest cost factor), I was dismayed that no one in the audience of about 10 people from the ethanol manufacturer viewed a PID as simple. It turns out the audience consisted of marketing and research people, who did not understand PID control. These people were making the decisions. There was not a single process control person in the audience. I have to admit the strategy takes PID control to more innovative heights for batch processes, but I don't think the audience ever got beyond the overwhelming concept of a PID controller. The heart of presentation can be seen in my May 2015 ISA interchange post "Simple Strategies for Optimizing Ethanol Plants"

Danaca: I often assume when people use the term "advanced control" that they mean advanced process control (APC). Control Global had a great article "Advanced Process Control Ain't Easy" a few years back that may explain some of the negativity.

Basically, the thesis is that APC is difficult to implement well, requires trained resources to maintain and has had an unfortunate failure rate due to vendors over promising how easy it will be to drop in their premade software and routines. We have a few wonderful, sustained uses of model predictive control (MPC) and other APC at our company, but these are the exceptions, and they usually require algorithms configured in-house by process experts.

When I read Greg's book Advanced Temperature Measurement and Control, I do not immediately get the same connotation, but I also don't know if I will have the opportunity or background needed to implement these control methods. This has more to do with my role as a generalist plant engineer instead of a specialized controls expert, rather than a negative connotation with the term "advanced."

Leah: I think Danica and Brian nailed it. I would add that the decline in plant resources doesn't help. It takes patience and dedication to model and optimize some of these control strategies, and current staffing doesn't allow that kind of time and focus. When a plant has one automation engineer that is running around putting out fires, that individual probably doesn't have bandwidth for optimizations. Generally, I would say a lot of plants don't have their existing "standard" controls healthy enough to support the addition of an advanced control layer. Adding more strategy doesn't correct fundamental issues.

Hector: The word "advanced" makes me think of sophisticated techniques in controls; something maybe "above my head." The word would not necessarily have a negative connotation itself, but rather it might make people hesitate, especially if they feel their knowledge is still basic and might have difficulties understanding what is presented.

I like the word "advances." It seems friendlier. Gives the idea "I will let you know what is out there, what the trends in this topic are; not necessarily complex, but rather recent."

For those non-practitioners, those in other areas or management, the sole word "control" might tell them nothing; do not even think about saying "advanced control." That might have to do with our own capability to sell our job and ourselves. In my early years, I limited myself to developing DCS code; nothing more. I had a vague idea of what was behind PID and how to tune it; never worried about learning the "advances" in my area. Now I have to broaden my skills and approach, and I still find it difficult to remove the "DCS Guy" label form my forehead when the people in the plant look at me.

Hunter: All of these comments are good. I would like to elaborate on Leah's comment as it is probably the biggest barrier to successfully implementing any advanced control. If the instrumentation is not reliable and accurate, or the simple PIDs are not working well and consistently, then it is an utter waste of time to even consider advanced control. Ironically when a little time and effort is expended to get the right instrumentation installed and the underlying PID controllers tuned well, many plants find that they immediately start reaping a large part of the benefits promised by a much more complicated (and expensive) advanced control solution.

Stan: The most successful APC projects had a person knowledgeable in basic control go into the system and improve the instrumentation and tuning. Also, the valves were already precise and responsive in that they were designed for throttling service and had good positioners. I suspect a major portion of the final benefits of the APC project were the result of these basic improvements, but all of the credit went to APC. In the end, maybe this doesn't really matter, in that the APC project achieves benefits from basic and advanced control and advances the recognition of the value of our profession.

Greg: I propose instead of "advanced process control" we use the term "smart process control." Smart instrumentation has an excellent record and reputation of making a step improvement in accuracy and reduction in maintenance. Operating and ambient conditions of "dumb instrumentation" had an installed error that was an order of magnitude greater than the catalog accuracy. Drift was large enough to require yearly recalibration. "Smart instrumentation" largely achieves catalog accuracy and may not need a calibration check for five or more years. Could we do the same with "smart process control?" Could the proper implementation require less maintenance than "dumb control?" There are some examples. The use of external reset feedback, setpoint rate limits, cascade control and ratio control (e.g., flow feedforward) can make life easier for the operator and the automation engineer if properly implemented. Signal characterization, auto tuners and adaptive control can enable the controllers to deal with the inevitable nonlinearities even if not perfect. The enhanced PID developed for wireless transmitters can deal with discontinuous updates and the associated delays of analyzers without the need for retuning. If MPC had features akin to external reset feedback and the enhanced PID and could provide diagnostics, had 24/7 online support and recognition of what it was doing, including online key performance indicators (KPI) and be self-healing, MPC could be included in "smart process control." If we can achieve this goal of better performance and less maintenance, we can ask operators, process engineers and plant management, "Do you want smart or dumb process control?"

Top ten things you don't want to hear from operators

10. What the heck is the APC doing now?
9. Those newfangled controls are causing me to miss lunch.
8. Turn off whatever is new.
7. APC guy? What APC guy? I never met him.
6. I just finished retuning the PID controllers again.
5. I just can't get enough reset action.
4. I am going to move up my retirement date.
3. I like manual so I get to fix what is going on.
2. Did that air system failure cause the loss of measurements?
1. These control systems are dumb and dumber.