Premium
Design of advanced process controllers
Author(s) -
Palmor Z. J.,
Shinnar R.
Publication year - 1981
Publication title -
aiche journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.958
H-Index - 167
eISSN - 1547-5905
pISSN - 0001-1541
DOI - 10.1002/aic.690270513
Subject(s) - control theory (sociology) , pid controller , dead time , relation (database) , process (computing) , controller (irrigation) , control engineering , stability (learning theory) , nonlinear system , function (biology) , phase margin , process control , identification (biology) , engineering , computer science , control (management) , mathematics , temperature control , cmos , database , artificial intelligence , electronic engineering , amplifier , operational amplifier , biology , operating system , quantum mechanics , machine learning , evolutionary biology , agronomy , statistics , physics , botany
This paper deals with defining the relation between modeling of process units and controller design. In the process industries, it is often difficult to obtain accurate process models. It is shown that advanced algorithms such as dead‐time compensators require more detailed process information for design than do with PI and PID controllers. A method is described to systematically evaluate the relation between process identification and controller design, stability, and performance. It is shown that conventional gain and phase margins do not provide proper safety margins for dead‐time compensators and optimal controllers. Methods for safe design of dead‐time compensators are derived, and the approach could be useful in a wide class of problems. To illustrate the approach presented, the design of PI controllers by conventional methods is analyzed. The conditions, as well as the exceptions, are specified in which methods such as described by Ziegler and Nichols or Cohen and Coon will give good results. It is shown that for any stable nonlinear system with an input/output function y = G* p ( u ), a linearized design function G ′ pd can be constructed and identified that guarantees stability and reasonable performance of the controller (Eqs. 21 and 22 and Figures 4–7) over a wide range of operating conditions. A rigorous framework for identifying the exceptions and understanding the reason why such methods work is presented.