Identification and control of fractional and integer order systems

The main focus of this thesis is on developing parsimonious models using measured process data and subsequent use of these models in the design of controllers for chemical engineering processes, in particular, for processes with large dead times. The application case studies presented and discussed in this thesis include diverse examples such as a classical heat transfer wall problem, a continuous stirred tank heater with transportation delays, an industrial scaled primary separation cell, and froth heaters.;Two different types of processes are discussed in this thesis: 1) processes that can be described by fractional order transfer function models and 2) industrial processes that are modeled using conventional rational (integer) order models, as is the normal practice in industry. For fractional order systems, this thesis proposes a nested loop optimization method where model parameters including time delay are estimated iteratively in the inner loop and the fractional order model is estimated in the non-linear outer loop. The proposed method is applied in simulation on distributed parameter systems such as a classical heat transfer wall problem and on identification data obtained from laboratory experiments of a continuous stirred tank heater (CSTH) with transportation delays and industrial froth heater process. A fractional order PI controller tuning method using Bode’s ideal transfer function as the reference system is also developed for fractional and integer order systems. The proposed tuning method is evaluated by simulation on fractional and integer order systems and experimental application on a computer-interfaced pilot scale CSTH process.;Application examples, related to conventional (integer order) models, discussed in the thesis involve two industrial case studies in the oil sands industry. The first of these is the regulation of the froth bitumen and middlings Interface level in a separation cell process which is part of the oil-sands extraction unit. Internal model control (IMC) and model predictive control (MPC) using linear models are designed, implemented and tested in real time on the industrial separation cell. These controllers yielded better performance over the existing control strategy which uses PID control. The second application is concerned with temperature control of the bitumen froth which is part of the froth treatment unit. Using the linear models obtained from the industrial data, a gain scheduling multivariable MPC is designed, and tested in simulations and compared with the current operation which uses a number of local PID controllers. Results presented in the thesis illustrate the first successful industrial implementation of an MPC controller on a separation cell in the oil sands extractions unit at Suncor Energy Inc. in Fort McMurray, Alberta.;Overall, this thesis presents results on identification and model based control design case studies on fractional order systems, distributed parameter systems and two industrial oil sands processes.