The goal of the innovation is to develop world leading technology for model predictive control and online optimization which is applicable to critical and demanding industrial processes. This technology is based on the use of nonlinear, "first principles" models, that is mathematical models which are developed from detailed knowledge of the industrial processes to be controlled. This technology is suitable for control of chemical reactors and metallurgical process units, which are typically the most critical units in an industrial processing plant. The project has focused on developing model predictive control technology which is tolerant to faults in instrumentation, process equipment and process operation.
The innovation is expected to increase the marked potential for advanced process control technology in industry sectors where safety and product quality are essential. The technology has been developed as extensions to Cybernetica's industrially proven system for nonlinear model predictive control - Cybernetica CENIT.
The project was executed in close collaboration between Cybernetica, NTNU, and the process industry partners Inovyn Norge, Eramet Norway, Arclin (North-America), and Aica (Asia-Pacific). Pilot installations of the new technology were accomplished at the industrial partner's processing facilities in Norway, Canada, and New Zealand.
This control technology offers superior performance in terms of control and operation of critical and exothermic processes, which are characterized by nonlinear dynamics, frequent grade changes, batch or semi-batch operation. Benefits are related to stable operation of processes which are difficult to control, increased production volumes, more consistent product quality, and increased safety compared to alternative control methods.
The objective of the project is to advance beyond current state-of-the-art and develop technology for nonlinear model predictive control (NMPC) and online optimization with unprecedented properties; The system will be based on 'first principles models' an d made tolerant to faults in instrumentation, process equipment and process operation. Among the planned, new features are techniques for model based fault detection and diagnosis; optimization of constraints and reference trajectories for the controller, so as to achieve acceptable performance in spite of faults; Continuous re-optimization of reference trajectories and constraints so as to meet terminal quality specifications.
The innovation is expected to increase the marked potential for advanced proc ess control technology in industry sectors where safety and product quality are essential. The technology will be developed as extensions to Cybernetica's industrially proven NMPC products. Implementations are planned for the control of phenolic resin po lymerization, amino resin polymerization, suspension PVC polymerization and ferromanganese refining processes.
The R&D challenges are related to the application of advanced methods for model based fault detection and diagnosis, and to the application of optimization based methods to ensure that the NMPC is robust with respect to faults and process disturbances.
This control technology is expected to offer superior performance in terms of control and operation of critical and exothermic processes, which are characterized by nonlinear dynamics, frequent grade changes, batch or semi-batch operation. Anticipated benefits are related to stable operation of processes which are difficult to control, increased production volumes, more consistent product quality , and increased safety compared to alternative control methods.