Rigorous model based model predictive control of a glass melter and feeder

Rigorous Model-based Predictive Control (RMPC) is getting widely accepted in the glass industries as a promising new technology. First applications have been completed and these have demonstrated the benefits of the use of rigorous process models for process control. RMPC is applied to increase furnace and feeder stability and to improve transition behaviour (e.g. load changes, glass composition/colour changes, recovery from major process upsets etc.). Stability of temperature distribution and flow pattern is essential for production at the highest possible yield. It is commonly known that upsets (even small ones) in the furnace should be prevented as each upset leads to defects. RMPC has demonstrated that reproducible and easy controlled production is possible. As RMPC is able to anticipate the effects of measured disturbances through feed-forward corrections, it prevents large drifts from nominal operating conditions and it is much faster in returning to nominal conditions than conventional PID controllers that only use feedback of measured deviations from set points. The resulting temperature variations using RMPC are therefore much smaller than in a furnace controlled by operators and PID controllers. This creates additional freedom for choosing process settings (controller set points) to meet objectives such as reduction of energy consumption and emissions. Conventional MPC controllers apply models that are determined by means of step tests on feeder or furnace in which the inputs (i.e. fuel distribution, boosting, and load) are changed in a stepwise fashion. The output response of the feeder or furnace is measured (e.g. variations in feeder temperatures in the zones and at the exit grid thermocouples or furnace crown profile and bottom temperature variations). To minimize risk of production losses during testing, many small steps are taken and measurements continue for several days or even up to weeks to enable modelling of the melting tank or feeder dynamics. From the measurement data a model is derived that relates the observed variations of the measured outputs to the manipulations applied to the input signals at the working point (set of process settings: pull, total fuel consumption, required electrical boosting power) for which the tests were performed. TNO Glass Group and IPCOS have developed a totally new approach for the determination of control models. In a first step, a rigorous, fast 3D detailed model (TNO's Glass Process Simulator or GPS) ofthe feeder or furnace is set up and validated. When GPS describes the dynamic behaviour of the process (feeder or furnace) sufficiently accurately, the rigorous 3D CFD (Computational Fluid Dynamics) model is applied for control system design. A newly developed model reduction technique based upon POD (Proper Orthogonal Decomposition) techniques is applied to derive the required, fast, dynamic simulation model for this RMPC design. The resulting control model that is derived from dedicated model reduction tests on the GPS simulator can be used for a large set of working points instead of for one single working point, as the response of the feeder/furnace to large variations in disturbances and process settings is determined through simulation. Consequently, the control model does not have to be rebuilt when a different working point for the feeder/furnace is selected due to e.g. the production of a different product. This new technique has been applied to various melters and feeders used in container, float and technical glass manufacturing.