For a regenerative TV-panel furnace as an example, the effect of changes in combustion firing profile on glass quality has been investigated for two different pull rates and two different cullet fractions. This has been done by simulation of the temperatures, heat transfer, flow patterns and elementary processes in the complete furnace, which includes melting tank, melter combustion space with port necks and burners, doghouse and throat. As the reliability of the predicted glass melt quality indicator values very much depends on the accuracy and the details of the predicted glass melt flows and temperature levels, the glass depth temperature profiles on various locations are validated against actual measured temperature profiles in the existing furnace, at two process settings. Glass quality parameters have been determined by statistical analysis of simulated trajectories of 50,000 particles that are released in the batch chargers, as well as for bubbles starting form different sources. The rate of growth and shrinkage of bubbles has been determined from relations describing the physical fining process. These derived glass quality parameters from modeling are correlated to actual glass defect production data - in terms of number of bubbles per ton of glass - found at two process settings. The simulation study is defined as Round Robin Test 5 (RRT5) by TC21 of the ICG. The case study and parameter variations study has been carried out by use of GTM-X, a computer program for Computational Fluid Dynamics (CFD) owned, developed and licensed by TNO Glass Group.
