Electromagnetic effects on glass melt flow in crucibles

Transcription

1 Proc. Eighth Advances in Fusion and Processing of Glass Glass Technol.: Eur. J. Glass Sci. Technol. A, February 2008, 49 (1), Electromagnetic effects on glass melt flow in crucibles U. Krieger, 1 B. Halbedel, D. Hülsenberg & A. Thess* DFG-research group Magnetofluiddynamik, TU Ilmenau, Germany Department of Inorganic Nonmetallic Materials, P.O. box , D Ilmenau, Germany *Department of Thermo- and Magnetofluiddynamics Manuscript received 17 July 2006 Revision received 23 July 2007 Manuscript accepted 30 July 2007 Knowledge and control of the vortex flow in melting systems play an essential role in improving the homogenisation of glass melts. Although from a historical perspective the influence and effect of electromagnetic forces on the melt flow is not a new technique it still has no industrial application. This paper addresses this alternative method resulting from the application of Lorentz forces. So called external Lorentz forces are generated by the interaction of an electric current density and a magnetic flux density realised by direct electric heating via electrodes and an external magnet system. Experimental results on the electromagnetic modification of the flow in stacked melts in a crucible, using coloured and colourless glass are presented. In addition the temperature fluctuations enabled the calculation of the velocity and the direction of the flow in the melts by the application of cross-correlation. The results show an enhanced thermal homogenisation of the glass melts by the external Lorentz forces and provide possibilities for the optimisation of glass production using magneto-hydrodynamic effects. 1. Introduction The progress and development of glass production are driven by new production ideas and by applications where glass acts as a key material. The need of higher quality levels results in new requirements for the production processes. Innovations in the past were sometimes simple such as mechanical stirring of glass melts. However, high quality levels were a precondition for the production of glass components for the light-optical microscopy and other optical products (1) and they are still an issue of concern. (2) Some significant difficulties exist in controlling the melt flow during glass processing. Unavoidable changes in melt properties are caused by their temperature dependence, fluctuations in chemical composition, or variations in the outlet flow which lead to irregularities and variations in the residence time and the chemical and thermal homogeneity of the glass melt. (2) Therefore the techniques of glass production have to be optimised in relation to the chemical composition of the glass and the required material properties of the product. This paper is organised as follows: Section 2 gives a short overview of existing possibilities for the optimisation of glass melting processes by various methods 1 Corresponding author. Uwe.Krieger@tu-ilmenau.de Now at: JSJ Jodeit GmbH, Am Nasstal 10, D Jena-Maua, Germany. service@jsj.de Proceedings of the Eighth International Conference on Advances in Fusion and Processing of Glass, June 2006, Dresden, Germany and their influence on the melt flow. Although from a historical perspective the influence and effect of electromagnetic forces on the melt flow is not a new technique, (3,4) it still has no industrial application and thus in Section 3 we evaluate the potential of the Lorentz forces followed by the description of the experimental set-up and parameters used here (Section 4). Then we report experimental results (Section 5) obtained from in situ measurements of temperature distribution during the electromagnetic stirring of glass melts; these include consideration of striae formation in the glass and the calculation of the velocities and the flow direction using temperature fluctuations. The summary of our investigations shows that external Lorentz forces provide an additional method for the optimisation of glass production. 2. Convection in glass melts In practice the manipulation of glass melt flow is mainly based on free convection, which is caused by density differences resulting from the inhomogeneous temperature distribution in the melt. Therefore, the effects of free convection are always interconnected with the direction of gravity and are optimised by the construction of the melt system. Techniques of forced convection like bubbling, which acts as a local flow barrier in tank furnaces, and direct electric heating via electrodes are mainly suitable for vertical melt flows. (5,6) In these cases the vortices that are produced in the melting system Glass Technology: European Journal of Glass Science and Technology Part A Volume 49 Number 1 February

2 Table 1. Traditional methods for the optimisation of melting processes and the improvement of quality level Method Characteristic Improvement construction of the melt system modification of convection, outlet flow formation of vortices bubbling flow manipulation formation of vortices, residence time electric boosting direct electric heating via electrodes, Joule heat effect improvement of melting, formation of vortices mechanical stirring local flow manipulation chemical homogenisation drainage elimination of polluted glass melt prevention of cat scratches and striae are additionally superposed on the outlet flow. The most common methods for the optimisation of melting processes that are related to the manipulation of the melt flow are summarised in Table 1. Chemical homogenisation in the melt is mainly affected by the velocity gradients of the flow. Local homogenisation is often realised by mechanical stirring which affects the velocity gradients in the zone near the stirrer due to the high viscosity of the glass melt. Mixing efficiency is increased with the number of revolutions but is very dependent on the geometric size and position of the stirrer. (7) This leads to the conclusion that in terms of chemical homogenisation the time dependence of the processes must be considered. Further problems arise from the abrasion products of ceramic stirrers and the cooling effects. Nevertheless, this method is profitable in the forehearth and the feeder, for example when glass melts should be coloured with a dopant, as they permit easy change of the colouring agent. (11) Due to the limited impact mechanical stirring cannot be used for the elimination of heavy cords originating from the corrosion of the refractory material. If the quality level requires the elimination of these cords, it is profitable to apply a drainage system located at the bottom zone of the feeder and to separate this glass from the main flow. (9) The generation of Lorentz forces is an additional technique that can influence the glass melt flow and thus affect both thermal and chemical homogenisation. The method has already been applied to special devices, (3) experiments on laboratory scale, (4,14) and has been investigated by numerical calculations. (11 13) However, until now it has not been applied industrially. To overcome this situation we have investigated the electromagnetic modification of glass melt flow in a crucible using a special laboratory equipment. It has already been shown that an enhancement of thermal homogenisation can be achieved by using external Lorentz forces. (14,15) In this paper we examines the influence of the external Lorentz forces on the glass melt flow in crucibles in relation to the number and size of vortices and the dependence of velocity on the external Lorentz force direction. 3. Generation of Lorentz forces in glass melts The Lorentz force density is given by (16) f L =σ(e+e i +(v B)) B (1) This equation involves the electric current density j=σ(e+(v B)) (Ohms law) where σ is the electric conductivity, E is the electric field strength, v is the velocity of the liquid and B is the magnetic flux density. The component σe describes the electric current due to the electric potential drop (E= gradφ) and σe i the electric current due to the induced electric field strength E i. An induced electric field strength E i is generated in the fluid because of the alternating magnetic field (B=B(t)). The term σ(v B) arises from the convection in the liquid. The electric current density j in the glass melt causes simultaneous heating (Joule heating), which is applied in practice for electric boosting and melting. In fact one has to consider a natural generation of Lorentz forces when the melt is directly heated via electrodes. (11,12,17) We refer to the force arising in this fashion as the internal Lorentz force, which is created by the interaction of the electric current density in the melt j and the flux density B=B int of the magnetic field of the electrode current. This magnetic eigenfield of the electrode current is governed by Ampere s law µ 0 j E = B int where µ 0 is the permeability of free space. As a result the effects of the internal Lorentz force exist in the vicinity of the electrodes and always act towards the tip of the electrodes. Thus the internal Lorentz force influences the stability of the batch layer in all electric melt sytems with bottom electrodes. (11,12) Significant values arise with rod shaped electrodes and with high currents in the electrodes, I E 800 A. (12) Figure 1 shows a schematic drawing of the internal Lorentz force in the vicinity of a top-down electrode. Furthermore, Lorentz forces are created as a result of (i) the interaction of the convective flow (moving ions) and the magnetic eigenfield of the electrode current (v B int ); (ii) the interaction of the electric current in the melt (σe) and the magnetic field which is produced by that current ( eigenfield ); (iii) the induction of eddy currents in the melt (σe i ) which interact with the magnetic fields. On the basis of the conditions in typical glass melts (v 1 mm/s, j 1 A/cm², f 10 khz, σ 100 S/m) the lastmentioned forces are several orders of magnitude lower than the internal Lorentz force described above and thus they can be neglected. (17) Induced Lorentz forces created as the result of eddy currents are negligible in our experiments due to the low frequencies (50 Hz) and the low electrical conductivity σ 5 S/m 34 Glass Technology: European Journal of Glass Science and Technology Part A Volume 49 Number 1 February 2008

3 Figure 1. Generation of internal Lorentz force F L in the vicinity of a top-down electrode, I E electrode current, B int flux density of the internal magnetic field (eigenfield of the electrode current), j current density in the glass melt (at 1300 C) of the glass melts. For induction heating of glasses (Skull Melting) frequencies of 200 khz are applied in practice. (18) We call the Lorentz force produced in the glass melt by an electric current density and an externally generated magnetic field the external Lorentz force; (14) this force can be produced in the melt via electrodes and an external magnet system. The direction of the external Lorentz force density f L is affected by the direction of the vectors of the electric current density j and the externally generated magnetic flux density B ext f L =j B ext (2) Test 3a Test 3b Figure 2. Schematic arrangement of two top-down electrodes for the generation of external Lorentz forces F L in a crucible; phase difference ϕ between the electric current density j and the externally generated magnetic flux density B ext : ϕ=0 test 3a; ϕ=180 test 3b; position (0,0,0) centre of crucible base We assume that both the electric current and the magnetic field are harmonically oscillating. Therefore, the direction of the external Lorentz force can be easily switched to the opposite direction with a shift of the phase angle, ϕ, from ϕ=0 to ϕ=180. (1) Figure 2 shows a schematic drawing of top-down electrodes in a crucible along with the possible directions of the external Lorentz forces. A comparison of the calculated values of varying force densities acting in different glass melts enables a better characterisation of the electromagetic stirring effects (see Table 2). For the calculation of the buoyancy force density we use f g =gβρ T (3) where g is the acceleration due to gravity, β the coefficient of volume thermal expansion, ρ the density of the glass melt and T the temperature difference in the melt. (17) The internal Lorentz force densities can be estimated using f L I E ² µ/v (4) where I E is the electrode current, V the active volume of the melt and µ is the magnetic permeability of the melt. (11) The internal Lorentz force densities were calculated for electric melting in glass tanks and are presented in Table 2 together with the calculated external Lorentz force density in the laboratory experiments. The calculations were carried out using an in-house program developed by the Department of Electroheat (PROMETHEUS) in combination with commercial software (FLUENT). The data in Table 2 show that the internal Lorentz forces in glass melts mainly depend on the electric current in the electrodes, and that the internal Lorentz Table 2. Comparison of different force densities in glass melt furnaces and in the laboratory experiments Force Melt type Value Known data type (N/m 3 ) buoyancy force - f (17) g ash melt 406 ΔT=300 K; ρ=2 3 kg/dm³; β= K 1 (17) internal Lorentz force - f L ash melt 51 I E =1000 A; B int = T; σ=5 S/m; v max =5 mm/s (11) internal Lorentz force - f L Na 2 O CaO SiO 2 72 I E =3000 A; B int =0 012 T; v max =4 4 mm/s (11) internal Lorentz force - f L Pyrex 3 4 I E =500 A; B int = T external Lorentz force - f L BaO B 2 O 3 SiO I E =25 A; B ext =0 055 T; σ=5 S/m; P dir =575 W Glass Technology: European Journal of Glass Science and Technology Part A Volume 49 Number 1 February

4 forces are much lower than the external Lorentz force density (laboratory experiments). Moreover, in the case of our experiments the electrode current is about two magnitudes lower than in industrial praxis. The comparison of the buoyancy force and the external Lorentz force indicates that the external Lorentz forces could be the only significant source of melt flow if the temperature difference T decreased. For practical application one has to consider that here only Lorentz force densities are calculated and that the flow pattern in glass melts results from the interaction of a sum of forces including buoyancy, Lorentz and friction forces. 4. Experimental equipment and preparation of specimens A special facility was developed and built for the systematic experimental investigation of the influence of Lorentz forces on glass melt flow in crucibles. The equipment has been described in detail elsewhere. (14) It consisted of a furnace heated with bifilar SiC-rods which were symmetrically arranged around a mullite tube. The furnace was positioned in the centre of the air gap of an external magnet system in such an arrangement that the generated magnetic field penetrated the ensemble. In Table 3 we present the chemical compositions of the BaO B 2 O 3 SiO 2 glasses studied. The glasses were initially melted in a muffle furnace using a Pt-crucible. For the experiments with the stacked melts the colourless glass TUI-1 was transfered (in the molten state) into an alumina crucible (height:100 mm). After cooling down, the alumina crucible was placed in the special furnace. In the other experiments we utilise a platinum crucible (height 100 mm, diameter 80 mm). The glass melt level reached a height of approximately 80 mm. In order to ensure a homogeneous temperature distribution as starting point for all tests at first indirect electric heating was applied (test 1). After remelting at 1300 C two electrodes were immersed into the glass melt from above to produce electric current (test 2). The arrangement was similar to Figure 2, but in addition also plate electrodes were applied in order to investigate the influence of the electrode design. (14) The position (0, 0, 0) was defined to be the centre of the crucible base. The simultaneous measurement of the temperatures in the melt (0, 0, z) was realised by a special PtRh-thermocouple (Electrotherm GmbH) consisting of three single thermocouples arranged in different z positions (0 mm, 30 mm and 60 mm). With the magnetic flux generated in test 3a the external Lorentz force acted upwards supporting free convection (+z direction). In test 3b the external Lorentz force was orientated in the opposite direction (see Figure 2). The test parameters are given in Table 4. The experimental results presented in section 5 were obtained with two stacked glass melts (colourless and coloured) using 0 02 wt% CoO as dopant and the same basic composition. The melts were layered in the ratio 1:8 (TUI-1-Co1 : TUI-1). At the beginning of the experiments the colourless glass in the alumina crucible was heated up again to approximately 1300 C. After a residence time of 60 min the coloured glass (grain size <1 mm) was filled-on on top of the melt. Then a 10 min heating period followed. After that the electrodes and the thermocouples were immersed into the melt and the power of the indirect electric heating of the furnace was reduced from 2070 to 1075 W in order to prevent overheating. After several minutes direct electric heating was started by application of direct electric power (575 W). In addition the magnetic flux density B ext was generated before the electric currents were turned on in the case of tests 3a and 3b. At the end of the tests the electric current and the magnetic field was switched off and the electrodes and the thermocouples were removed from the glass melt. Then the cooling of the furnace down to 650 C was started ( 20 K/min) in order to diminish the stress by a residence time at the transformation temperature of 12 h. A low cooling rate of 2 K/min was used down to room temperature. Of course the removal of the electrodes and the thermocouples influences the flow profile, but the use of expensive platinum prevented cooling with dipped in electrodes and thermocouples. Additionally some diffusion may occur during cooling and influences the situation of the solidified material. Then the crucibles were cut with a diamond saw blade in the x z plane (cf. Figure 2). In relation to the visualisation of the striae formation disks with a thickness of about 5 mm were cut parallel to the x z plane. After grinding and polishing they were scanned with a flat bed scanner using transmitted light for the visualisation of the CoO distribution. The local chemical composition (CoO-content) was investigated with atomic absorption spectrometry (AAS) for some top and bottom positions in the glass. Table 3. Chemical compositions of the glasses studied (wt%) Glass SiO 2 B 2 O 3 BaO Fe 2 O 3 CoO TUI TUI-F TUI-1-Co Table 4. Experimental parameters Test Indirect Direct External electric electric Lorentz heating heating force 1 X 2 X X 3a; 3b X X X 36 Glass Technology: European Journal of Glass Science and Technology Part A Volume 49 Number 1 February 2008

5 Due to low concentration of CoO the required weight of the samples of about 5 g results in a relatively global determination of the chemical distribution. Thus the differences in the chemical composition in the dimensions of the striae thickness could not be characterised by AAS. Using in situ measurement for the temperature distribution T(x, y, z) in the melt the first systematic results of the thermal homogenisation were published in Hülsenberg et al. (14) To calculate the flow velocities in the centre of the crucible the glass composition TUI-1 was modified with 3 mol% Fe 2 O 3 (TUI-F3, Table 3) in order to reduce the heat transport by radiation. The existing temperature fluctuations were used to determine the running time of temperature maxima between the thermocouples with constant distance (fixed positions) using the cross correlation function xcorr in MATLAB. For these experiments platinum crucibles were used and constant parameters were reached before switching from test 2 to test 3a and from test 3a to test 3b. 5. Results and discussion The direct electric heating of the glass melt led to the development of a hot spot between the electrodes in all cases (tests 2, 3a, 3b). Figure 3 shows the time dependence of temperatures in the stacked melts (TUI-1, TUI-1Co1) at three z-positions for the different test parameters. In test 2 buoyancy and internal Lorentz force caused the melt flow. In test 3a both the buoyancy and the external Lorentz force acted upwards in the centre of the crucible resulting in an acceleration of the flow. This can be seen in the detected temperature differences which significantly changed with the application of the external Lorentz forces (test 3a) compared to test 2. The external Lorentz force (test 3a) led to a decrease in the temperature difference {T (z=60 mm) T (z=0 mm)} to 20 K after 5 min in comparison with a temperature difference of 40 K in the case of only direct electric heating (test 2). This was the result of a faster increase in the temperature at the bottom and at (0,0,z=30 mm). The temperatures at (0, 0, z=60 mm) exhibited approximately the same dependence on time for tests 2 and 3a. With an external Lorentz force acting in the opposite direction of the buoyancy (test 3b) the temperatures T(z) in the centre of the crucible increased much more strongly and the vertical temperature differences diminished due to the deceleration or reversal of the flow (see Figure 3). Furthermore the wavelike shape of the curves at z=0 mm and z=30 mm became evident. This effect was reproducible and was used for the calculation of the flow velocities. On the other hand we found that during test 3b the temperature differences between the centre (0, 0, z) and the glass near the side wall increased too. The Figure 3. Temperature time dependence for tests 2, 3a and 3b (see Table 4) - duration of direct electric heating and external Lorentz force: 5 min; positions of the thermocouples in the melt (0, 0, z) glass at the side wall became colder compared to the melt in the centre. Therefore the thermal homogeneity in the glass melt decreased for test 3b. As shown in previous work (14) the hot spot in the Glass Technology: European Journal of Glass Science and Technology Part A Volume 49 Number 1 February

6 melt moved in the direction of the external Lorentz force, and the temperature distribution became independent of time when static experimental parameters were applied. Numerical simulation of the arrangement indicated an increase of the maximum velocity by a factor of approximately three in switching from test 2 to test 3a. (15) The visualised electromagnetic stirring effects are presented in Figure 4 which shows one half of the scanned disks after tests 1, 2 and 3a. In interpreting the images the undesired diffusion and processes during the cooling and corrosion of the electrodes have been neglected. The formation of the striae in the glass was caused by the cobalt oxide doped glass (top layer) and by colloidal platinum particles resulting from the corrosion of the electrodes. These features enabled visualisation of the flow pattern and the interpretation of mixing efficiency in the glass melt. The striae formation indicates that a torus-like flow pattern existed in the melt (two vortices in the x z plane). Further, the vortices were enlarged in the direction of the generated external Lorentz force in case of test 3a compared to test 2. The additional magnetic field (test 3a) was applied for 5 and 25 min. The comparison in Figure 4 shows a well distributed CoO-concentration over the volume of the glass melt after 25 min. This was in agreement with the measurement of the CoO-content using AAS. A concentration of 0 003±0 0007wt% CoO was detected at the top and the bottom of the glass. The formation of striae demonstrates that the homogenisation after short times was not satisfactory and that longer experimental times had to be employed (electromagnetic stirring >25 min). In order to clarify the direction of the flow in the centre of the crucible we investigated the temperature distribution in the glass TUI-F3 (see Table 3) when switching from test 2 to test 3a and from test 3a to test 3b. The fluctuations were used to calculate the velocity, assuming the straight movement of the melt from one thermocouple to the other (fixed distance 30mm) along with the characteristic running time which was obtained from cross correlation. In Figure 5 the time dependence of the temperature fluctuations during test 3b indicates that the glass was moving from z=60 to 30 mm and thus in the opposite direction to the buoyancy. The calculated velocity between these positions in the crucible amounts to 1 5 mm/s for test 3b and +8 3 mm/s for test 3a. (19) As a result the striae formation in the glass melt and the simultaneous measurement of the temperatures at three positions in the melt demonstrate that: two vortices were developed in the x z cross section of the melt; the thermal homogenisation was improved by the acceleration of the flow caused by buoyancy and external Lorentz forces; the reversion of the flow by contrary-directed external Lorentz forces significantly increased the temperature at all positions on the line (0, 0, z), and; the temperature instabilities that existed were used to calculate the velocities and the direction of the flow. For improvement of chemical homogenisation Figure 4. Half of scanned images (40 80 mm) of stacked glasses TUI-1 and TUI-1-Co1 after tests 1, 2 and 3a; duration of direct electric heating and acting of external Lorentz forces: 5, 25 min (the black line in test 3a (5min) is a crack) 38 Glass Technology: European Journal of Glass Science and Technology Part A Volume 49 Number 1 February 2008

7 Figure 5. Temperature fluctuations after switching from test 3a to test 3b using Fe 2 O 3 -doped glass melt TUI-F3 knowledge of the formation of vortices is a very important precondition. The torus-like flow pattern in the melt results in low velocity gradients at the free surface and the crucible wall. The thermal homogenisation and the scanned images show that an accelerated flow existed if the external Lorentz force and the buoyancy act upwards in the centre of the crucible. By the inversion of the Lorentz force direction (test 3b) the velocity in the centre of the crucible became negative which led to flow in the opposite direction. Because of the lower velocity the melt remained longer between the electrodes (in the heating zone ) and thus the temperature in this area of the melt increased. This resulted from the interaction of buoyancy and external Lorentz force and the constant electric input power. At the same time the glass at the side wall became colder and the total thermal homogeneity of the melt decreased for experiments with a steady downwards-directed Lorentz force (test 3b). A paper detailing these temperature measurements is in preparation. Thus the time dependent variation of the external Lorentz force enables several possibilities for controlling the melt flow and improving the chemical homogenisation process. For example the periodic switching of the Lorentz force direction may result in an oscillating flow and increase the stretching and folding of the fluid elements. The characterisation of this effects will be investigated in future experiments. 6. Summary The effects of external Lorentz forces on glass melt flow in crucibles were studied using a special experimental facility. The equipment enabled the direct electric heating of the glass melt via two electrodes and the generation of Lorentz forces using an external magnet system. The formation of striae caused by doped glass (CoO) and colloidal platinum particles was used to visualise the vortices in the glass melt and to indicate the possibility of an electromagnetically forced flow. From these results and the calculation of velocities we found that improved homogenisation occurred when external Lorentz forces were applied to give accelerated melt flow. With the inversion of the external Lorentz force direction a deceleration and even a reversal of the flow could be achieved which resulted in an increase of the temperatures between the electrodes. We conclude that for the investigated conditions the flow can be controlled by the external Lorentz force. Further experiments with doped glass melts are planned in order to verify the effects of time, alternating direction of the external Lorentz force, direct electric heating power on the flow pattern. Also the numerical simulations will be optimised by considering the time and the temperature dependence of the glass properties. Based on the understanding of the electromagnetic stirring effects and the glass melt flow in crucibles our work facilitates the industrial application of external Lorentz forces for the optimisation of glass melting. 7. Acknowledgement This work was supported by the Deutsche Forschungsgemeinschaft in the framework of the research group Magnetofluiddynamik at TU Ilmenau. The authors would like to thank D. Keil and M. Schulke for their help with the experiments and U. Lüdtke and C. Gießler for the numerical simulations. 8. Symbols Glass Technology: European Journal of Glass Science and Technology Part A Volume 49 Number 1 February B ext B int E E i f flux density of the externally generated magnetic field flux density of the internal magnetic field (eigenfield of the electrode current) electric field strength induced electric field strength frequency

8 f g buoyancy force density f L Lorentz force density F L Lorentz force g acceleration of gravity j electric current density j E electric current density in the electrode I E electric current in the electrode P dir direct electric heating power, input power of the electrodes P ind indirect electric heating power, input power of the furnace T temperature difference v velocity x,y,z coordinates β coefficient of volume thermal expansion σ electric conductivity µ magnetic permeability ρ density φ electric potential ϕ phase angle (between electric current density and magnetic flux density) T temperature Nabla operator Δ 9. References 1. Vogel, W. Glaschemie. Third Edition. Springer, Loch, H. & Krause, D. Mathematical Simulation in Glass Technology. Springer, Walkden, A. J. Improvements in or relating to the manufacture of glass. Patentschrift, GB , Fekolin, V. N. & Stupak, F. A. Application of the magneto-hydrodynamic effects to stirr glass melts. Steklo Keram., 1984, 41 (12), (In Russian.) 5. Högerl, K. & Frischat, G. H. Homogenisation of glass melts by bubbling. Proc. Int. Congr. on Glass: Bol. Soc. Esp. Ceram. Vidr., 1992, 31-C, Zhiqiang, Y. & Zhihao, Z. Basic flow pattern and its variation in different types of glass tank furnaces. Glass Sci. Technol., 1997, 70 (6), Cable, M. Model studies of the homogenising of laboratory glass melts. J. Non-Cryst. Solids, 1996, 196, Chrisman, M. G., Ganzala, G. W. & Tiede, R. L. Method and apparatus for mixing and homogenising molten glass. US patent , Sims, R. & Geslein, J. Glass feeder has wide collection chamber above a drainage opening for difficultly miscible bottom glass to reduce formation of cat scratches and other flaws in finished glass products. Patent DE , Moukarzel, C., Kuhn, W. S. & Clodic, D. Numerical precision of minimum residence time calculation for glass tanks. Glass Sci. Technol., 2003, 76, Hofmann, O. R. & Kaliski, H. Die elektromagnetische Kraftwirkung auf die Glasschmelze in Elektrodennähe. Silikattechnik, 1992, 42, Hofmann, O. R. Electromagnetic force in electric glass melting. Glass Sci. Technol., 2003, 76 (4), Choudhary, M. K. A modelling study of flow and heat transfer in the vicinity of an electrode. Proc. XVII Int. Congr. on Glass, Beijing, 1995, Hülsenberg, D., Halbedel, B., Conrad, G., Thess, A., Kolesnikov, J. & Lüdtke, U. Electromagnetic stirring of glass melts using Lorentz forces - experimental results. Glass Sci. Technol., 2004, 77 (4), Krieger, U., Halbedel, B., Hülsenberg, D., Lüdtke, U., Kolesnikov, Y. & Thess, A. Elektromagnetische Strömungsbeeinflussung in Glasschmelzen : Proc. 79. Glastechn. Tagung, Würzburg, Moreau, R. Magnetohydrodynamics. Kluwer Academic Publishers, Dordrecht, Němeček, M. Inductive mixing forces in ash-melts. Proc. Electric Melting of Glass, 1992, Frishfelds, V., Jakovics, A. & Nacke, B. Study of melting dynamics of oxides in inductor crucible. Proc. Heating by Electromagnetic Sources (HES-04), Padua, 2004, STUDYOFMELTING.pdf 19. Krieger, U. PhD thesis, TU Ilmenau, Shaker Verlag GmbH, Glass Technology: European Journal of Glass Science and Technology Part A Volume 49 Number 1 February 2008