(799g) Acicular Mullite Reaction Furnace Model Development

Acicular mullite (ACM, 3Al2O3:2SiO2) is a unique porous ceramic with an interlocking needle-like microstructure.   Its resistance to chemical attack, thermal stability, and mechanical robustness at high porosities has made it an attractive candidate for use in filtration and catalytic support applications.  When the mullite needles form the walls of a honeycomb structure, it can be used as a diesel particulate filter in which the soot laden exhaust gas flows into a set of inlet channels in the honeycomb structure, through the ACM walls where the soot is deposited, and then out through the exit channels of the honeycomb.  The soot is burned out of the walls during a regeneration phase of the filter cycle.

The acicular mullite material is formed in a furnace during a two-stage reaction. In the first stage, the furnace is heated to 750°C and SiF₄ gas is added to convert the honeycombed clay precursor material into fluorotopaz (Al₂SiO₄F₂).  In the second stage, the pressure is lowered to remove SiF₄ and the temperature is raised to the decomposition temperature of the fluorotopaz, which converts to acicular mullite.  TheSiF₄ is recovered and recycled, and the furnace is cooled down.  The topaz reaction is exothermic and the mullite reaction is endothermic.  Heat transfer within the furnace and ACM parts is one of the critical issues identified for this process.  It can affect the mullite needle size and quality, the batch cycle time, and furnace loading options.  The heat transfer components include radiation, conduction, and convection, with heat sources and sinks due to the heats of reaction. 

In order to better understand and control the heat transfer and chemistry within the furnace, a research project was initiated to develop a kinetic model for the gas-solid reactions and apply it to a computational fluid dynamics (CFD) model of the production furnace with over 1000 reacting parts.  The kinetic expressions incorporate SiF₄ partial pressure, which required the development of a SiF₄ material balance within the CFD model.  

This paper will summarize some of the methods used and the results obtained in the creation of a model for the precursor to topaz reaction stage, including model validation against data for a single part in a lab furnace and against production plant data.  The model does a fairly good job of predicting the initial rise in temperature in the center of the parts and in predicting the overall extent of reaction.