Acoustic Chemometric At-Line Characterization for Monitoring Particle Size Fractions in Pneumatically Conveyed Biomass

Abstract

Reliable on-line/at-line prediction of particle size fractions of biomass material is a process monitoring concern owing to inherent variation in particle size and difficulty in sampling of pneumatically ducted biomass material. A feasibility study on application of acoustic chemometrics for at-line prediction of size fractions of biomass material has been performed, with all models subjected to independent test set validation. This study serves as a platform for at-line characterization of biomass samples from a sampling device extracting biomass samples from pneumatic conveying systems (based on theory of sampling). A prestudy using complex biomass/plastic pellets mixtures was used to test reliability and robustness of the experimental setup. Promising prediction results were achieved (slope, relative root mean square error of prediction (RMSEP_rel), and correlation coefficient (r²) were 1.08, 24.94%, and 0.90, respectively). The same experimental setup was adapted for quantitative prediction of coarse versus fine biomass mixtures (the main objective) with satisfactory results; slope = 0.96, r² = 0.97, RMSEP_(rel) = 11%. A case study was also performed showing the adverse effect of using nonrepresentative coarse versus fine biomass samples for calibration. It is concluded that acoustic chemometrics is a viable technique for at-line prediction of size fractions of representative biomass materials.

Keywords:

• Acoustic chemometrics
• biomass
• fine-fraction estimation
• partial least squares regression (PLS-R)
• process analytical technology (PAT)
• theory of sampling (TOS)

Notes

*weight/weight percentage.