Repository hosted by TU Delft Library

Home · Contact · About · Disclaimer ·

Advanced feeder control using fast simulation models

Publication files not online:

Author: Verheijen, O.S. · Camp, O.M.G.C. op den · Beerkens, R.G.C. · Backx, T. · Huisman, L.
Institution: TNO Industrie en Techniek
Source:Drummond III C.H., 65th Conference on Glass Problems, 19 October 2004 through 20 October 2004, Columbus, OH, Conference code: 66450, 1, 26, 1-10
Ceramic Engineering and Science Proceedings
Identifier: 238984
Keywords: Automation · Computer simulation · Glass · Glass furnaces · Process control · Redox reactions · Fast simulation models · Feeder control · Glass quality · Control systems


For the automatic control of glass quality in glass production, the relation between process variable and product or glass quality and process conditions/process input parameters must be known in detail. So far, detailed 3-D glass melting simulation models were used to predict the effect of process input variables, such as fuel consumption and fuel distribution, load and load distribution, and electrical boosting, on the flow pattern (residence times, short cut flows), temperature distribution and redox state of the glass in the furnace. For feeder control, the main objectives are stable temperature and temperature uniformity in the spout section of the feeder just before the glass melt is delivered to the forming process. However, computations of detailed 3-D simulation models are very time-consuming: one steady state simulation of a complete furnace including refiner(s) typically takes about a day. This time demand indicates that the currently used detailed 3-D simulation models are not suitable for rigorous (CFD) model based Model Predictive Control (MPC). To make the 3-D simulation models suitable for control purposes, simulation tools that are much faster than real time with still a high level of reliability are now developed. These fast simulation tools (GPS: Glass Process Simulators) open up a wide variety of applications of glass furnace models: process monitoring, what-if scenarios and model based predictive control (MPC) of temperatures or glass melt quality or redox/color. Now we realized time-transient simulations with GPS for feeders, which are about 10000 times as fast as real time. This paper will show the capability of an MPC that is based on fast GPS to control temperature and temperature uniformity of one of the feeders connected to an emerald-green container glass furnace.