Water absorption thermodynamics in single wood pellets modelled by multivariate near-infrared spectroscopy

Lestander, Torbjörn A.

doi:10.1515/hf.2008.071

Cited by 10 publications

(1 citation statement)

References 9 publications

(12 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The fraction containing RFAs is interesting because it contains valuable precursor chemicals and light-phase bio-oils that can be extracted, for example, in supercritical carbon dioxide and in the low-temperature range of thermochemical processes, such as torrefaction and pyrolysis. The RFA fraction is also a source of volatile organic compounds that can be problematic or hazardous during biomass storage and industrial processes. − These compounds are also associated with the phenomenon of auto-oxidation, which may result in self-heating and spontaneous ignition of stored biomass. − …”

Section: Introductionmentioning

confidence: 99%

Prediction of Resin and Fatty Acid Content of Biorefinery Feedstock by On-line Near-Infrared (NIR) Spectroscopy

Lestander

Samuelsson

2010

Energy Fuels

Self Cite

View full text Add to dashboard Cite

Extractives in biorefinery feedstock are a source of precursor chemicals and biofuel products. Resin and fatty acids (RFAs) in such extractives constitute an interesting fraction, which may contain both chemically attractive precursors and also problematic volatile organic compounds. On-line near-infrared (NIR) spectra were collected from a process stream, designed experimentally and involving softwood lignocelluloses; the data were regressed using partial least-squares to give RFA concentrations that varied between 0.1 and 0.5% (dry weight basis). At-line NIR models were also constructed using spectral data from pelletized samples from the process stream. In addition, off-line NIR modeling was conducted using softwoods with a wider RFA variation range (0.1-1.0% dry weight basis). All of the calibration models obtained exhibited good predictive abilities. The laboratory-based off-line NIR model explained 94.5% of the variation in concentrations and had a prediction error of 0.070% for the RFA content. The coefficient of variation (CV), representing the percentage of the ratio between the prediction error and the average concentration, was 17.8%. The on-line and at-line models explained 71.3-79.6% of the variation in the RFA concentrations and had prediction errors within the range of 0.026-0.041% and CVs of 13.7-18.1%. This was excellent in comparison to the ca. 10% error in accuracy when determining the RFA reference values. The results illustrate that on-line NIR spectroscopy provides a valid method for real-time predictions of RFA concentrations in biomaterials. This should facilitate better monitoring and process control as well as targeted pretreatments to obtain tailor-made biorefinery feedstock, thus adding value to the production process.

show abstract