ANALYSIS OF ELEMENTARY PROCESS STEPS IN …

1 INTRODUCTIONR esidence time distributions, determined for batchand the glass, after batch fusion, in glass melting fur-naces show that the flow of material in the tank partlyfollows plug flow regime, but that also sections withstrong mixing and re-circulation flows exist. In the glassmelting tank, different zones should preferably corre-spond to different ELEMENTARY PROCESS STEPS . But, inmost industrial glass melting tank furnaces, these zonesare not physically separated from each other, and thedifferent processes in the tank may interfere. Thesebasic processes are: a). heating of batch up to "meltingor fusion" temperatures, b). fusion reactions to form asilicate melt and c). dissolution of sand grains in thisprimary silicate melt and final molten glass composi-tion, d).

2 Removal of gas bubbles and dissolved gases ande). chemical homogenization plus thermal homogeniza-tion of the viscous of the melt in these tanks is relative-ly intense: in most cases the glass melt passes a zone5 up to 8 times before the melt leaves the tank. Most ofthe well-molten en completely degassed melt from thehot spot area, returns to the batch area and is mixed withfreshly molten non-homogeneous glass, still containingvery large numbers of bubbles and seeds. The high qual-ity hot glass melt from the hottest spot of the tank justbrings part of the required melting energy to the batchtip. The mixing with the freshly molten glass, that con-tains many un-molten grains and seeds, spoils this highquality with a melting history representing 4 hours ofmelting and with 48 hours of melting can be in the sameglass container or glass sheet product.

3 This means thatif the glass quality is still acceptable, the parts witha residence time of 48 hours has been in the tank up to12 times longer than essentially necessary. N mec andJebav [1], showed that dead water zones and the socalled space utilization factor (depends on the ratiosbetween minimum and average residence times) have alarge influence on the energy consumption of a glassfurnace for given pull or the maximum pull obtainablefrom a melting tank. They also showed, that very hightemperatures may not be the optimum solution for highpull and low energy consumption and that certain opti-mum conditions in terms of optimum temperature, stir-ring, enhanced bubble growth rate can be derived fora certain glass melting tank to get high pull and rela-tively low specific energy consumption and high flexi-bility.

4 They propose developments in the direction ofOriginal papers206 Ceramics Silik ty 52(4) 206-217 (2008) ANALYSIS OF ELEMENTARY PROCESS STEPS IN INDUSTRIALGLASS melting TANKS - SOME IDEAS ON INNOVATIONSIN INDUSTRIAL GLASS MELTINGRUUD BEERKENSTNO Science & Industry, Glass Group, De Rondom 1, Box 6235, 5600 HE Eindhoven, The NetherlandsE-mail: May 6, accepted October 9, 2008 Keywords:Fast fining, Glass furnace design, Compact melter, Controlled meltingConventional industrial glass furnaces show broad glass melt residence time distributions in the melting tanks and averageresidence times may be up to more than two days for high quality glass products, such as float glass or TV glass, despite theminimum residence times of 8-10 hours (or even less than 4 hours for container glass furnaces).

5 Long residence times areassociated with large melt tank volume/pull rate ratios and high structural heat losses. The recirculation flows, necessary tosupply the batch blanket with sufficient energy for the fusion processes of the batch materials, are often poorly controlled andcause these shown residence time differences. For each ELEMENTARY PROCESS step the most relevant PROCESS parameters in termsof desired flow patterns, mixing behavior, temperature, chemistry and required time for completion of the PROCESS step areanalyzed. From this ANALYSIS , it is concluded that the different STEPS need very different conditions comparing the processeswith each other. Therefore glass furnace design modifications should be focused on the development of furnaces with seg-ments, each of them dedicated and optimized for the a certain stage of the melting PROCESS : melting -in of batch, sand graindissolution, removal of gases (bubbles and dissolved gases), re-absorption of residual bubbles and glass melt (chemical andthermal) homogenization.

6 Some developments in fast fining and rapid melting -in of batch will be shown and shortly presented at the seminar "Advanced Glass Materials and Innovative Glass MeltingTechnology in the Year 2020", Brig, Switzerland, March 26-29, space utilization (avoid to many different trajec-tories in the melting tank and dead water zones) andtechniques to speed-up sand grain dissolution and finingwithout increasing temperatures. Conclusion: most industrial glass melting tanksoperate far from optimal and furnace designs todayallow very strong recirculation flows, keeping the glassmelt longer in the tank than would be necessary. Lessrecirculation and improved heat transfer to the batch inorder to minimize the need for hot spot glass re-circula-tion offers a potential for considerable energy savingsand space reduction of the glass tank paper first addresses the features of industrialmelting tank furnaces operated today by more than90 % of worlds' glass production and identifies the weakpoint of this concept existing since about the essential PROCESS STEPS will be describedand the optimum conditions for each of the sections inthe furnaces accommodating the PROCESS STEPS will bepresented.

7 Finally some suggestions for improving glass fur-nace tank designs will be discussed and some dilemmasand problems still to be solved will be elements seem to be:zImproved heat transfer to and into batch blanket: mostof the batch material melts (fuses) in the direct vici-nity of the batch tip;zControl of flow patterns and optimization of the hotspot plus spring zone position in the tank (for glassquality it is important that spring zone and hot spotlocations in the melt are close together and that flowpatterns are stable);zFaster fining (gas removal from melt) processes;zThe availability of refractory materials allowinghigher temperatures that enables faster fining (seeFigures 3).RESIDENCE TIME DISTRIBUTIONIN CONVENTIONAL GLASS melting TANKSIn general, the molten glass obtained from the batchafter fusion of these raw materials, does not follow adistinct trajectory in the melting tank [2].

8 Many differ-ent paths will be travelled by glass melt volumes in thesame tank. In the worst case, the freshly molten glassmoves from the batch blanket into the direction of thetank bottom and flows directly (short cut flow) alongthis relatively cold bottom of fossil fuel fired meltingtanks, to the neck or throat to leave the melting -end. Thetemperature history of this melt shows generally that theglass has hardly been exposed to temperatures thatallow all seeds or bubbles to be removed from this high-ly viscous melt. On the other hand, in case of a well-developed spring zone in the tank, located between thebatch tip and the exit of the tank, the molten glass flowalong the bottom is interrupted by an upward flow at thespring location.

9 The glass melt may reach the surface ofthe melt. Ideally, the spring zone position covers the hotspot area of the tank or is very close to the location ofthe highest temperatures in the melting tank. Theupward flowing glass enters the highest temperaturesection of the melt and viscosity decreases, fining agentbecomes active, fining gases are formed and growingbubbles can reach the glass melt modeling shows that the largest partof the melt reaching the surface of the spring zone area,flows back to the tip of the batch 1 shows residence time distributions of threedifferent industrial glass melting tanks. One can observethat the minimum residence time is typically 15 to 20 %of the average residence time. Figure 2 shows as an example the projection of theflow pattern of a piece of glass melt in a typical floatglass furnace with a residence time close to the averageresidence time.

10 The figure shows that this 'particle' ANALYSIS of ELEMENTARY PROCESS STEPS in industrial glass melting tanks - some ideas on innovations in industrial glass meltingCeramics Silik ty 52(4) 206-217 (2008)207 Figure 1. Residence time distribution determined by mathe-matical modeling [2] for 3 different melting tanks, producing:TV panel glass, clear float glass and container time (h)% per - Container2 - TV-panel3 - FloatFigure 2. Flow pattern of glass melt volume with an averageresidence time. Note the strong re-circulation patterns in thefloat glass melt tank with a bubbler endbubbler areaworking endcanal,outletrecirculationinto melting endrecirculationin melting endentrancedoghousesideneck(in modeling we call a small piece of melt or volume ofmelt a 'particle') re-circulates a few times between thebatch zone and the spring zone of the from the residence time, the melting behav-ior depends strongly on the temperature level at whichthe glass has been exposed: fining is characterized by afining-onset temperature [3].