Assesment of the effects of chemical and physical parameters in the fluidization of biomass and sand binary mixtures through statistical analysis

Empirical correlations of fluidization design parameters such as characteristic velocity are based on properties such as particle mean diameter, sphericity and true density. These properties are difficult to measure for particles with morphologies typical of lignocellulosic biomasses. Thus, these correlations generally show limited practical applicability. The aim of the present work was to verify through an innovative multiple linear regression if other physical and physicochemical properties would be more appropriate for characteristic fluidization velocities prediction. The methodology involved evaluating sugarcane straw and bagasse, eucalyptus wood, rice husk and apple tree branches in binary mixtures with sand in the proportions of 20, 40, 60 and 80% of biomass volume fraction. The multiple linear regression analysis indicated that the mixture bulk density and repose angle were the most relevant parameters in the prediction of characteristic fluidization velocities. Two correlations were developed employing these properties and attained suitable results: the adjusted R2 coefficient obtained for Umf prediction was 0.772 while for Ucf prediction was 0.855. In 90% of mixtures, the relative error found for Umf was smaller than 24.3% while for Ucf this error was smaller than 22.5%.