Application of factorial design to the modelling of organosolv delignification of Miscanthus sinensis (elephant grass) with phenol and dilute acid solutions

The Battelle-Geneva method of fractionation of lignocellulosic materials was applied to the biomass of an annual cane-like plant, Miscanthus sinensis (elephant grass). A factorial second-order composite design was used to quantify the influences of three variables of the process (independent variables), (1) time, in the interval: 0.37-2.35 h; (2) phenol concentration, in the interval: 41.6-58.4% by weight; (3) hydrochloric acid charge, in the interval: 0.0098-0.0602 g HCl/g dry Miscanthus, on five dependent variables: (I) pulp yield, (2) residual lignin in pulps, (3) dissolved reducing sugars, (4) total polysaccharides in pulp, and (5) selectivity of the process, measured as the ratio pulp yield/residual lignin in pulp. Experimental results were fitted to second-order lineal-models by means of a multiple regression method Expressions obtained showed R-2 values varying between 0.80 and 0.99, obtaining fittings from acceptable to good depending on each case. The model predicted a maximum in the selectivity for the following values of independent variables: time, 1 h 48 min: phenol concentration 53.2%; acid charge, 0.0394g HCl/g dry Miscanthus. With regard to the other responses variations shown were, in general, logical according to the increase or decrease of the independent variables. (C) 1997 Elsevier Science Ltd.