Evolution of glass melt quality in space and time from the batch to the working end

Bartolomey, Simon; Conradt, Reinhard (Thesis advisor); Hand, Russell J. (Thesis advisor); Roos, Christian Hans-Georg (Thesis advisor)

Aachen (2017, 2019)
Dissertation / PhD Thesis

Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2017


The present thesis aims at the investigation of the correlation between the constituting raw materials of the batch and the subsequent evolution of melt quality in order to reduce the break rate of fibres during the continuous production of reinforcement fibres to the lowest possible limit. The evolution of melt quality may be mirrored by the evolution of redox state, which undergoes three consecutive stages, i.e. the initial stage during the melting of the batch, an intermediate stage during homogenization in the basin, and a final stage during which the redox state is finally adjusted to that of the glass product. The initial stage determines predominantly the resulting redox state of the glass and the so called glass quality respectively. The term glass quality is imprecise and unquantifiable and should be substituted by the more accurate term of defect concentration. It is postulated that the defect concentration in the melt is proportional to two quantities, namely the break rate and the width of the density distribution. Therefore, the break rate as the standard forming performance indicator in the fibre glass industry can also be used as a quantifying measure of defect concentration. The width of the density distribution and the median density were determined for a representative set of glass particles by utilization of the newly adapted and significantly improved density titration method. The samples were taken at the bushing in regular intervals during running production. For further analysis, corresponding to these glass samples, ready mixed batch samples were taken at the charger, considering the mean residence time of the melter, 24 h prior to the glass samples. The sampling took place in the period between May to December 2015.Even though the chemical oxide composition of the batch was constant for all samples, the segregation into a coarse and fine grain fraction revealed a change of mineral phase composition, which was initially detected in October 2015 and all following samples. This could be attributed to a change of an orthosite raw material quality. This change in the mineral composition of the batch shows neither in the measured width of the density distribution nor in the measured median density of the glass samples an indication fora significant correlation. Another working hypothesis states, that any technical process is subject to fluctuations, that can be described by the superposition of simple wave functions with discrete frequencies. Based on this working hypothesis the time series of the break rate, production data, and furnace data were analysed for systematic, periodically reoccurring events by time series decomposition with a one year period and the subsequent isolation of the obtained seasonal components. All decompositions were conducted on time series comprising > 3 periods. It was shown, that the seasonal component of the break rate follows a sine wave with the tendency to a higher break rate in summer and a lower break rate in winter. Additionally, an exceptionally high probability for filament breakage was registered in August. This probability correlates with the power input for the furnace and the break rate. To be specific, a reduction of gas power coincides with an elevated break rate. A deeper investigation of the causal interlinks is shown by the following context: in August the gas power input decrease leads to a temperature drop to an all-year low in the combustion chamber in the batch melting area. Thereby the batch to melt conversion is affected in a way that the initial number of defects, which subsequently are conveyed to the bushing causing breaks, is increased. This special correlation cannot readily be used to explain the long-wave seasonal fluctuations during the rest of the year, because it is deemed possible the power fluctuations depend on fluctuations of outside temperature. It is also possible, that the fluctuations of the break rate correlate with the climate conditions in the forming area. It was shown, that the air conditioning run sat maximum capacity in summer and struggles to provide constant climate conditions. This may also cause an elevated break rate. Finally, a recommendation based on the results is given.


  • Division of Materials Science and Engineering [520000]
  • Chair of Glass and Glass-ceramic [524210]