Valorization of acai bio-residue as biomass for bioenergy: Determination of effective thermal conductivity by experimental approach, empirical correlations and artificial neural networks

Abstract

Current concerns about the depletion of fossil fuels and global warming led to new policies put into place for the use of agro-industrial residues as biomass for thermochemical conversion. Acai berry residues are large-scale agro-industrial by-products that can be used in the production of bioenergy. Biomass effective thermal conductivity is one of the parameters that affect efficiency in the bioenergy process. Once research about acai berry residues as biomass for bioenergy is in its initial stages, prior determination of thermal conductivity may help future optimization studies towards high process efficiency and low operating and capital costs. In this work, it is shown that a simple experimental approach, together with a simple neural model, may allow initial assessment regarding reactor throughput and process cost. This is the main contribution of this work. Results from physical characterization, thermogravimetric analysis, and differential scanning calorimetry of the biomass were reported and discussed. Effective thermal conductivity as a function of moisture content and heating rate was experimentally determined by the line heat source method and predicted by artificial neural networks, empirical correlations proposed by literature, and multiple linear regression analysis. Four values of biomass moisture content were studied, 30.97%, 25.94%, 18%, and 11.86% (wet basis) and two values of heating rate, 4.16 and 16.74 W⋅m−1. The experimental values lay in the range 0.136–0.325 W⋅m−1·K−1, typical for common biomass and insulation materials. Such low values are not interesting for reactor design purposes since the dynamics of the thermochemical conversion is less effective. This leads to lower reactor throughput, which must be compensated by increasing reactor effective volume, translating to high capital and process costs. The findings obtained in this work have shown that thermal conductivity can be simply improved by increasing the heating rate. The developed neural model, which was superior to empirical correlations and multiple linear regression, accurately predicted experimental values outside the database. Simulation results have shown that artificial neural networks are a potential tool to provide means for understanding the influence of biomass properties and process conditions on a reactor design parameter like thermal conductivity. A simple neural model may contribute to the optimization and design studies towards efficient thermochemical conversion processes of acai berry residues, which may help to provide a proper destination to those by-products.