This Is AuburnElectronic Theses and Dissertations

Rapid prediction of wood constituent composition and energy traits using near-infrared spectroscopy




Wang, Qun

Type of Degree



Biosystems Engineering


There is increasing interest in producing energy from renewable biomass resources. These resources, however, tend to be highly variable in nature and this heterogeneity can complicate their effective utilization. Rapid prediction of energy and chemical characteristics of biomass could assist in optimizing its conversion to biofuel and bioenergy. Loblolly pine (Pinus taeda) is the most common woody biomass in the southern U.S. and the most promising renewable resource for bioenergy. Near infrared spectroscopy (NIR) has been of great interest as a rapid, cost-effective and nondestructive technique for quantitative and qualitative analyses. This research aims at developing NIR calibrations for rapid measurement of constituent and energy properties of biomass, which can be applicable to large-scale assessment of forest resource properties and consequently assist efficient process control of large-scale conversion of heterogeneous feedstock to energy outputs. In particular, the objectives of this research were: (1) develop NIR calibrations for prediction of lignin, extractives, ash, moisture content, and energy content (calorific value) of loblolly pine biomass; and (2) compare the NIR calibrations based on spectra from wood powder and chips and reveal the potential of calibrations based on a single spectrum per chip to predict the properties of loblolly pine. The calibrations based on spectra of wood powder, averaged spectra per chip (25) and single spectrum per chip for constituents and calorific value were established. Good calibration were obtained based on spectra from powder with coefficients of determination (R2) of 0.93 (SEC=0.28%) for lignin, 0.91(SEC=0.14%) for extractives, 0.85 (SEC=0.025) for ash, 0.96 (SEC=0.32) for moisture, and 0.91 (SEC=0.05 MJ/kg) for calorific value. The calibrations based on averaged spectra per chip also presented good correlation with an R2 ranging from 0.8 to 0.9 for composition and calorific value. Calibrations based on a single spectrum per chip gave an R2 of 0.81 (SEC=0.4%) for lignin, 0.84(SEC=0.18%) for extractives, 0.72 (SEC=0.03) for ash, 0.87 (SEC=0.33) for moisture, and 0.74 (SEC=0.34 MJ/kg) for calorific value. The results indicate that for all properties in the current study, the calibrations based on spectra from the powder gave the highest R2 and relatively lower standard error. Furthermore, good correlations between measured and predicted values were also acquired from the calibrations based on averaged spectra per chip, exhibited a slightly lower R2. Although not as good as powdered or average chip tests, relatively strong calibrations were possible for even the single spectrum treatment when predicting chip properties. With RPD values ranging from 1.3 to 2.9, the calibrations based on single spectrum per chip met the requirement for initial screening. The simplicity and rapidity of calibrations based on a single spectrum from a solid wood chip may outweigh the slightly greater precision achieved when analyzing ground, bulk samples. This study reveals that NIRS in combination with multivariate analysis has the potential to predict the bioenergy and chemical characteristics of biomass in an industrial conversion.