Vassilios Tzouanas

Expert multivariable control

Abstract Multivariable control techniques have not been as widely adopted in process plants as their developers had hoped and expected. This can be attributed to their dependency on an accurate process model and their inability to effectively handle abnormal conditions. To safeguard against such shortcomings, an expert multivariable controller (EMC) is designed to make the control of multivariable processes more effective and fail-safe. It diagnoses process malfunctions and makes appropriate changes to the controller structure, control algorithms and control parameters. The controller uses plant-independent control systems knowledge as well as process specific knowledge. Its design is facilitated by using artificial intelligence (AI) techniques. The application of this controller to a distillation process is demonstrated.

Adaptive pole-assignment control of a batch polymerization reactor

Abstract This paper is concerned with the control of important polymerization properties in a batch solution polymerization reactor of methylmethacrylate. Properties such as weight avarage molecular weight or monomer conversion are controlled using an advanced adaptive pole-assignment controller whose performance is compared to that of a generalized minimum variance controller and a fixed gain PID controller. Simulation and experimental results on the batch polymer reactor demonstrate the need for fast on-line measurements in the polymerization industry as well as the usefulness of this adaptive pole-assignment control algorithm.

Adaptive Pole-Placement Control of a Batch Polymer Reactor

Abstract This paper is concerned with the theoretical and practical aspects of adaptive pole-placement controllers. Following a brief review of the theory, the paper outlines an evaluation of an adaptive pole-placement controller. Simulation and experimental results on a batch polymer reactor for the solution polymerization of methylmethacrylate demonstrate the usefulness of this adaptive pole-placement control algorithm.

Towards an Expert Multivariable Controller

Abstract Multivariable control techniques have not been as widely adopted in process plants as their developers had hoped and expected. This can be attributed to the dependency on an accurate process model and their inability to handle effectively abnormal conditions. To safeguard against such shortcomings an Expert Multivariable Controller (EMC) is proposed to make the control of multivariable processes as effective and fail-safe as possible. It diagnoses process malfunctions and makes appropriate structural changes to the controller structure, control algorithms and control parameters in order to achieve as tight control of the process as possible. The expert controller uses plant-independent control systems knowledge and plan dependent knowledge. A preliminary version of this controller is demonstrated in the control of a batch PVC polymerization reactor.

Constrained PID design & implementation: A co-simulation framework

A simulation environment is presented for tuning proportional-integral-derivative (PID) controllers that optimizes a performance criterion subject to various constraints. The tuning environment is cast under a simulation-based control design or co-simulation framework not requiring elaborate analytical optimization techniques or computing platforms; it is applicable to open loop stable or unstable processes; it is independent of the PID controller type; it accommodates controllers with proportional and/or derivative action on the error or process variable; and can be applied to the simultaneous tuning of PID controllers in a series or parallel cascade structure. The method can also account for robustness in response to modeling errors by including magnitude constraints on the nominal sensitivity function. Performance is demonstrated in simulation using Microsoft Excel and its Solver function in a Plug-and-Play approach. A PID design is applied to a two-tank level laboratory system to compare experimental results against the performance obtained from classical tuning methods.

Adaptive pole-assignment control of processes with large time delays

By the use of a simple predictor for dead time compensation, we show that an explicit pole-assignment algorithm can become or practical importance for processes with relatively large time delays. With the omission of the delay term in the closed-loop characteristic polynomial, solution of the pole-assignment control algorithm becomes computationally more efficient. The closed-loop system with an adaptive pole-assignment algorithm and a dead time compensation scheme is shown to be globally stable. The stability analysis is applicable to both minimum-phase and non-minimum-phase systems. For a particular class of processes, it is shown that the pole-assignment algorithm can take the form of a velocity-type, three-term, adaptive PID controller.