WOOD RESEARCH Volume 62, Number 3, 2017
Partson Tinarwo, Temesgen Zewotir, and Delia Nort

Modelling the effect of eucalyptus genotypes in the pulping process with generalised additive models and fractional polynomial approaches

The advent of lean (waste reduction), six sigma (process variation minimisation) and proper raw material selection are the essential challenges to achieve the required quality on the overall industrial processes. Accordingly a laboratory experiment for the dissolving wood pulping process was conducted on nine Eucalyptus genotypes to measure the change in lignin, viscosity and α-cellulose at each of the six pulp processing stages. The changes to these properties were modelled using the Generalised Additive Models (GAM) and Fractional Polynomial (FP) models. These models proved to be equally important in their unique ways and produced complementary results. The results revealed that Emearnsii genotype produced the best results for both α-cellulose and viscosity, while Enitens genotype was selected for the optimal lignin reduction. Egrandis genotype is the only genotype that proved to have adverse effects on the viscosity property.

WOOD RESEARCH Volume 63, Number 4, 2018
Partson Tinarwo, Temesgen Zewotir, and Delia North

Pre-hydrolysis pulping process optimization with multiple response surface modelling

Three properties of dissolving pulp namely lignin, viscosity and the α-cellulose were investigated. A laboratory experiment for the dissolving wood pulping process was conducted on nine Eucalyptus genotypes: Edunnii, Esmithii, Egrandis, Macarthurii, Emearnsii, Enitens, GCG438, GUA380 and GUW962. Repeated measurements were taken at each of the six processing stages for the changes in lignin, viscosity and the α-cellulose. A response surface approach was used to select the best genotype for each property and further application of desirability analysis to identify the genotype that simultaneously gives the best results for the three properties. The predictive models and associated statistical tests proved that all the nine genotypes were capable of producing the optimal results (>95.55% α-cellulose) although a few were at the thresholds of the feasible region. The optimisation process also revealed that the genotype Emearnsii possesses the most desirable properties for the α-96 cellulose product output and Enitens consistently produces results within the desired range. The use of simultaneous desirability functions indicated that the overall product quality characteristics for lignin, viscosity and the α-cellulose can be improved by steadily excluding the most resistant genotypes to lignin reduction, especially Edunnii and Esmithii.