The physics of the fiber glass drawing process has been extensively investigated, both numerically and experimentally. In particular, a two-dimensional axisymmetric model has been developed and validated with experimental data from both the literature and a dedicated experimental unit. The sensitivity of the physical model on the material properties, heat fluxes and viscosity law has been analyzed. Additionally, a simplified semi- analytical one-dimensional model has been derived by neglecting radial variations and assuming a purely extensional flow. The dif- ferent assumptions and/or simplifications on which this model is based have then been quantified. Finally, the one-dimensional model has been applied to two problems related to the industrial process, i.e., the identification of an optimal operating window
and the impact of temperature inhomogeneities at the bushing plate.
The physical model attempts to integrate all relevant physics. Nonetheless, several simplifications had to be made. In partic- ular, internal radiation has been neglected despite the fact that liquid glass is a participating medium, at least for some range of temperatures and wavelengths. The expected effect of internal radiation would be to reduce radial variations and modify the cool- ing rate in the region close to the tip. Then, the convective heat flux has been approximated, as commonly done, through a global empirical correlation. It has been shown that the use of this cor- relation is questionable, especially in regard of the non-uniform temperature distribution of the ambient air observed in practice. A better representation of convection, by additionally modeling the surrounding airflow, would also impact the cooling rate, and thus the axial stress. Another simplification of the model is the assumption that the glass melt behaves like a Newtonian fluid. While this assumption is valid at high temperature, i.e., in the initial region close to the tip, viscoelastic effects can be expected when the temperature decreases and the liquid glass approaches its transition point. Nevertheless, the impact of this simplification is supposed to be weak as viscoelastic effects would only be seen far downstream from the tip. Finally, the material properties have been shown to represent one of the major sources of uncertainty, as they are often difficult to determine for liquid glass at high temperature.
The semi-analytical one-dimensional model has been devel- oped based on several simplifications. A sensitivity analysis has shown that neglecting inertia and surface tension represents the major source of error in the model, while neglecting radial variations is an adequate approximation. Because of these simplifi- cations, the predictions of the simplified model are only qualitative. But despite its lower fidelity, the model provides a better insight into the physics of fiber drawing and reveals explicitly the relation- ship between the key parameters.
The application of the two models to different cases has led to several key results. First, it has been found that the initial region close to the tip is critical. In particular, the final stress depends pre- dominantly on the cooling rate in this region, i.e., a lower cooling rate leads to a lower stress. This critical result can be explained by the fact that the main contribution of the viscosity to the flu- idity takes place at high temperature. Another observation is that the surface heat transfer in this initial region is dominated by radi- ation, while convection becomes dominant further downstream. Then, it has been found that the best strategy to minimize the stress is to operate at a high tip temperature, a large drawing velocity and a small tip radius. This optimal strategy is however limited by different constraints, such as, for instance, the drawing instability that appears at high temperature. Finally, temperature inhomo- geneities have been shown to create large stress variations that could potentially play a key role in fiber breaking.
In light of the current shortcomings of the physical model, future work will focus on i) integrating viscoelastic rheology to better rep- resent the fiber stress in the transition region, ii) including internal radiation, iii) considering unsteady phenomena to characterize the fiber response to perturbations and to identify potential instabil- ities, and iv) improving the representation of the convective heat transfer by simulating the flow dynamics of the surrounding air. Ultimately, the physical model developed for a single isolated fiber will be extended to multiple interacting fibers.