A new, fast, and efficient automatic tuner for the ABB AC 800M family of controllers

Abstract: The most common controller in the process industry today is the PID controller due to its simple structure. A regular factory may have thousands of different PID controllers in use where each of them has to be tuned to work satisfactory. Doing this manually can become both a difficult and time consuming task where some knowledge for each process is required. Therefore, it is highly beneficial to have a way to automatically tune the controllers, which was the motivation behind the relay autotuner first introduced in the 1980’s. Since then, many alterations of the original relay autotuner have been proposed as technology development have progressed, concerning both available computing power as well as PID controllers in general. This master thesis is proposing a new autotuner, based on a short asymmetrical relay experiment, that is running at sufficiently low cost to be executed directly in an industrial controller unit. This is highly beneficial for end customers who are to control a process with unknown dynamics. An extensive simulation study is conducted using a test batch of a wide range of processes representative for the process industry to ensure that the approach is adequate. A suitable first order model with time delay (FOTD) is found from which a controller structure is based, following a set of well-known tuning rules. The method is then implemented in the ABB Ability™ System 800xA where the proposed method was shown to successfully estimate a model and control a process with unknown dynamics in a robust way.