CFD SIMULATION OF GLASS MELTING FURNACE

Abstract

Glass manufacturing is an energy intensive process. Glass melting and homogenizing in the melting furnace accounts for about 70-80% of the total thermal energy consumed. The quality of the final product depends partly on the degree of mixing obtained in the furnace.. A better understanding of the hydrodynamics and heat transfer in the glass melting furnace will help to conserve energy, reduce emissions and improve quality. Developments in the field of computational fluid dynamics have enabled complete simulations of the glass melting process which has helped to improve quality and reduce energy besides being an aid for operational troubleshooting. Mathematical models of Glass melter furnace divide the total domain into two or three parts: Combustion Space, Glass Melt and or Batch Blanket sub domain. The Combustion space sub-domain includes fuel combustion and heat transfer by radiation and convection to the batch and the free surface of the glass melt. The Glass Melt sub-domain includes processes of mixing, fining (used to remove trapped gas bubbles in the molten glass), air bubbling (to aid in mixing) of the molten glass material. The Batch Blanket domain includes processes like heating, chemical reactions and melting of the raw materials to transform the feed to molten glass. Radiation plays a dominant role in combustion space heat transfer. It is composed of three components i.e. gaseous species radiation, soot radiation and radiation from crown surface. Investigation of the variation in heat flux due to location is fundamental to the understanding of the furnace operation and critical for model validation and future design improvements. Hayes et al. (2001) carried out measurements of crown incident radiant heat flux along the furnace axial centerline and developed a numerical model. However the effect of soot was not quantified. A numerical model which takes into account the effect of soot in radiation heat transfer is developed and simulated. The error between model predictions (without soot model) and experimental results due to Hayes et al. (2001) is between -13 and 43 % and with the soot model it is between - 6 and +14 %. Variation of crown incident radiant heat flux along the width of the furnace is also studied both in the presence and absence of soot model. In both the cases a decrease is observed as one moves away from the firing end with the decrease being around 10 % without soot and around 25-30% with soot model.