Application of compositional data analysis to determine the effects of heating mode, moisture status and plant species on pyrolysates.

Pyrolysate gas mixtures are multivariate and relative in nature. Statistical techniques applied to these data generally ignore their relative nature. Published data for permanent gases (CO, CO₂, H₂, CH₄) and tars produced by pyrolysing 15 wildland fuels were reanalysed using compositional data analysis techniques. Mass and mole fractions were compositionally equivalent. Plant species, moisture status and heating mode effects on compositional balances formed from subsets of pyrolysates were tested. Plant species affected the amount of phenol, primary and secondary/tertiary tars relative to permanent gases and relative amounts of single- and multi-ring compounds. Plant moisture status affected the amount of CO relative to other permanent gases, of H₂ to CH₄ and tars to phenol. Heating mode and rate strongly influenced pyrolysate composition. Slow heating produced more primary tars relative to multi-ring tars than fast heating convective and combined radiant and convective heating modes. Slow heating produced relatively more compounds with fewer rings and fast heating produced relatively more multi-ring compounds. Compositional data analysis is a well-developed statistical methodology, providing models and methods equivalent to traditional ones, that accounts for the special constraining features of relative data. Future analysis of the compositional data related to wildland fire using compositional techniques is recommended. Compositional data techniques were applied to show that heating mode, plant species and moisture status affected the composition of pyrolysis products in an improved and statistically rigorous analysis. These methods should be used for other compositional data related to wildland fire. [ABSTRACT FROM AUTHOR]