Model-based optimization of perlite expansion via a Response Surface Method (RSM)

Conventional perlite expansion suffers certain well-known shortcomings compromising its viability and the adherence of expanded perlite to modern technical specifications for high-quality insulation materials. A new perlite expansion process has been designed and a vertical electrical furnace for perlite expansion has been constructed to overcome such drawbacks, with concurrent modeling studies (Angelopoulos et al., 2013 [1]). Having already accomplished the production of various expanded perlite grades for many varying applications,
the entire state space of product quality against key manipulated variables has also been explored, and ideal experimental condition ranges for process operation have also been identified (Angelopoulos et al., 2013 [2]). Response Surface Methodologies (RSM) have an long track record of contribution to substantial improvements in many advanced chemical and material processing technologies. Their fundamental principle is the systematic exploration and statistical correlation of input (conditional) and output (response) variables with respect to interactions (if any) of the former and their combined effect (if any) on the latter – thus on end product quality. This paper focuses on using parametric sensitivity analysis results from the published furnace dynamic model as well as original experimental furnace results towards extracting a new three-dimensional RSM model which achieves an accurate correlation of both pivotal experimental variables (furnace temperature, air flowrate) with their combined effect on a known end-product quality metric (grain expansion factor): optimal condition ranges can thus be derived via bivariate quadratic polynomial fits for any plant and any raw perlite feed composition.