Index of /skoge/publications/2023/skogestad-advanced-regulatory-control_arc
Abstract.
The paper explores the standard advanced control elements commonly used in industry for designing advanced
control systems. These elements include cascade, ratio, feedforward, decoupling, selectors, split range, and
more, collectively referred to as "advanced regulatory control" (ARC). Numerous examples are provided, with
a particular focus on process control. The paper emphasizes the shortcomings of model-based optimization
methods, such as model predictive control (MPC), and challenges the view that MPC can solve all control
problems, while ARC solutions are outdated, ad-hoc and difficult to understand. On the contrary, decomposing
the control systems into simple ARC elements is very powerful and allows for designing control systems for
complex processes with only limited information. With the knowledge of the control elements presented in
the paper, readers should be able to understand most industrial ARC solutions and propose alternatives and
improvements. Furthermore, the paper calls for the academic community to enhance the teaching of ARC
methods and prioritize research efforts in developing theory and improving design method.
UPDATES to the paper - Comments from the author
Associated one-day NPC workshop from August 2023:
Slides from Advanced process control course at NTNU
MISPRINTS
Figure 34 caption. The correct is: "Inventory control of single unit for case with desired feed flow Fs (can be given up)"
Comments on the paper (mostly from Linkedin)
Pawel Domanski, Warsaw University of Technology
I find your paper very interesting and to speak frankly it is a pity that such an article appeared only after so many years,
because the discussion and reflection on the state of process control engineering is completely missing.
[more - full email]
Pouria Sarhadi, Univ. of Hertfordshire:
Similar to your other work, but in a more informal and practical tone, this paper explains the essence of control –
what truly works in real-world applications.
It's a highly interesting and striking paper that I recommend to all control engineers. The more experience (of course not mine), the smoother grasp of its material!
Miguel Mauricio Iglesias:
Amazing work, kind of 40 years of research in process control summarised in one single paper.
Tao Chen, Univ. of Surrey:
Great read Sigurd, many thanks for this work!
Atique Malik, AIControl LLC:
Wonderful. This is great. A good overview and very practical. Nice to have access to useful material of a high standard.
Timothy Salsbury:
Nice work that highlights the value in solving complex problems with multiple simple and replicable “blocks” rather than making the solution as inherently complex as the problem.
David De Sousa. OMV, Austria:
Excellent work Sigurd. Thanks for making it Open Access. Timothy, indeed simple and replicable "blocks" that industry practitioners are very familiar with and that are applied on a regular basics in the process industry, which adds a lot of practical value to this work.
Olav Slupphaug, ABB:
I strongly encourage everyone in my network dealing with education, research, innovation, and practise related to control and optimization - in particular process control - to integrate the insights by Prof Sigurd Skogestad moving forward.
Kent Rasmussen, Process control engineer at Tengizchevroil:
Thank you for making this paper open access and sharing the slides from the course. It is always inspiring to read your work.
The ARC and MPC can often hapilly coexist to provide economic optimization.
Atique Malik, AIControl LLC:
(comment to Kent Rasmussen). Quite so. The two technologies of MPC and ARC have been a useful and necessary complement to each other for at least the four decades that I have worked at oil refineries and petrochemical plants.
So much has been written about MPC but very little cogent and insightful work on ARC. This paper is the exception. Some previous work by Luyben and Morari comes to mind, but it’s not as practical.
Damien Hocking, Digital Oilfield:
This is a great read.
David de Sousa:
This very insightful paper explores the standard advanced control elements commonly used in our industries for designing advanced control systems. These elements include cascade, ratio, feedforward, decoupling, selectors, split range, and more. These are collectively referred to as ‘‘Advanced Regulatory Control’’ (ARC).
Numerous examples are provided by the author, with a particular focus on process control. The paper emphasizes the shortcomings of model-based optimization methods, such as model predictive control (MPC), and challenges the view that MPC can solve all control problems, while ARC solutions are outdated, ad-hoc and difficult to understand.
On the contrary, decomposing the control systems into simple ARC elements is very powerful and allows for designing control systems for complex processes with only limited information.
Michael Caracotsios:
A truly outstanding review article (as expected by Professor Skogestad!)
David de Sousa:
Outstanding not only from an academic standpoint but also from an industry practitioner perspective. Great that He and NTNU have made it Open Access. Michael.
Victor M. Zavala, Professor at University of Wisconsin, Madison:
I love the paper for many reasons, Sigurd. I do think that we should challenge MPC as a control paradigm from an interpretability/explainability stand-point.
Like a neural net, MPC is a powerful and flexible tool, but it is difficult to make sense of it. I like the principle of decomposition and modularity to help overcome some of these issues. Great work!
------------
MISPRINTS
-------------
Page 26: It says
"Note that Shinskey (1978) has proposed a separate scheme for complex MV-CV switching, see Figure 9 in Reyes-Lúa et al. (2019). "
1. However, no paper Reyes-Lúa et al. (2019) is found in the reference list. It should be Reyes-Lúa and Skogestad, S. (2020b):
"Systematic Design of Active Constraint Switching Using Classical Advanced Control Structures". Industrial and Engineering Chemistry Research, 59(6), 2229–2241.
The reason for the confusion is that the journal claims that the paper is both from 2019 and 2020 (depending on where you look).
2. Also, I could not find this scheme in Shinskey (1978) (maybe I did not look good enough), but I do find it in Fig. 6.15 in Shinskey (1979):
Process Control Systems, 2nd edition, McGraw-Hill.
------------
COMMENTS
------------
see separate document: Comments.....doc