The 80/20 rule, also known as the Pareto Principle, states that approximately 20% of factors account for 80% of outcomes, highlighting the importance of a few critical elements. A comment from an enthusiastic forum user reads: “Why is PI control so prevalent in industrial applications? The 80/20 rule applies—PI accounts for 80% of usage. But don’t forget the performance rule: most outcomes are acceptable but not exceptional, with ideal or poor performance being the minority. The basic fact is that with PI, regardless of tuning or optimization, the result is typically just satisfactory.” While I don’t advocate for new algorithms, I support the 80/20 rule. It explains why we remain so committed to PID. My doctoral research focused on advanced control, and my work primarily involves advanced control. For control theorists, researching challenging control problems and proposing solutions is essential. In equipment that doesn’t require on-site engineer tuning, complex control algorithms may be a manufacturer’s closely guarded secret. In large-scale systems, these advanced algorithms are carefully protected to prevent competitors from accessing them. Self-tuning technologies are widely applied in many fields, but the inherent uncertainty in process control gives it a unique character. Advanced control algorithms and self-tuning techniques are less common in process control.
Engineers working on industrial systems must prioritize cost consciousness, achieving appropriate goals with simple, affordable, and reliable methods. Attempting to “boil the ocean” is unrealistic. In the process industry, the uncertainty of controlled processes and disturbances necessitates feedback control to cost-effectively reduce uncertainty. It is precisely this uncertainty that allows PID, which does not rely on precise process models, to maintain over 95% of the market share in process control, with most applications being PI. This uncertainty also requires engineers to perform PID tuning themselves, as well-designed control schemes often fail when feedback does not adequately address uncertainty. When uncertainty is not sufficiently reduced, causal relationships become unclear, and higher-level applications often underperform. The key to improving process control lies in effectively utilizing PID. However, practical applications face various challenges. While control schemes and tuning are critical, years of implementing higher-level applications have not prioritized these foundational elements. As a result, despite significant investments, the adoption of advanced applications has been slow. Industry has already invested heavily in new algorithms and advanced applications. Even model predictive control, with numerous successful industrial case studies, encounters various issues in practice. If foundational control problems remain unresolved, what can a new algorithm—or even AI—achieve? I believe caution is warranted, especially in feedback control.
Focusing on PID—studying it, applying it effectively, and tuning it properly—is the key to improving process control. Within the feedback control framework of error-based correction, dividing errors into present, past, and future represents a fundamental philosophical approach to solving all problems, with PID being its mathematical expression. Control schemes and tuning are the critical drivers of improved process control. Developing and promoting simple, universal, consistent, robust, and flexible tuning methods is the ultimate strategy. This is why the Lambda tuning method is so important. As Goethe said, “People always seek the extraordinary, unaware that the extraordinary lies in the profound understanding of the ordinary.”