Tools for Autonomous Process Control

There is an ongoing trend towards higher automation level in process control systems. The purpose of a plant is to manufacture some product at a high production rate with a consistent quality, while minimizing the use of resources in terms of energy consumption, raw material and labor. With a higher degree of autonomy in the control system, it is possible to improve all these aspects, thus increasing the profit of the plant. When the control system is not able to solve a task by itself, it should support the human interaction with good software tools. This thesis treats different aspects of autonomous process control. A functional architecture of the control system is presented. The focus is on autonomy on the local control loop level. A list of desired functionality for an autonomous single loop controller is presented. This list consists of methods for initialization, assessment of basic process features, selection and tuning of on-line controller, monitoring of the on-line control performance, and fault diagnosis. Implementation aspects and software architectures of an autonomous single loop controller are also discussed. In particular, sequential control using the graphical language Grafchart is studied. The thesis also gives detailed presentations of three different tools and methods that can be used in an autonomous control system. An interactive tool for preliminary assessment of the process dynamics is presented. It is based on analysis of data from one or more step response experiments. A simple dynamical model, possibly combined with a static non-linear function, can easily be obtained. The model can be used for calculating parameters for a PI or PID controller. A new method for automatic tuning based on relay feedback is developed. A relay with time-varying hysteresis is used in order to achieve excitation over a large frequency interval. An estimation of the frequency response of the process is obtained by frequency domain identification. This estimation is used together with optimization methods for robust PI and PID controller design. A simple strategy for fast set point response is presented. It mimics what experienced process operators often do manually to obtain fast set point step responses with no overshoot. The strategy consists of a short sequence of steps in the control signal. Conditions for good switching times are given. These conditions can be applied with varying degrees of process knowledge