Viscosity models for New Zealand black liquor at low solids concentrations

Abstract

Listen

Translate article

AbstractAvailability of accurate models for prediction of the viscosity of black liquor (BL) from the chemical pulping of pine will facilitate its online monitoring and control and subsequently the optimisation of combustion in a recovery boiler. New Zealand (NZ) BL viscosity data are limited, and no predictive model is available. The viscosities of the NZ BL samples at solids concentrations (SCs) < 50% were obtained at temperature of 25–85 °C and shear rate up to ∼2000 s−1. The samples showed Newtonian behaviour. Existing models from the literature and a binomial model developed in this work were used to fit the viscosity data as a function of SC and temperature. Accuracies of these models were examined for both the log‐transformed and the untransformed viscosity data using coefficient of correlation (R) and maximum absolute relative error (MARE) (between the actual and predicted viscosities), respectively, as indices. Although the existing models fit NZ BL viscosity data well when they were log‐transformed, they performed poorly when not transformed. Conversely, the new binomial model gave accurate predictions with both the log‐transformed and untransformed viscosity data (R = 0.9997; MARE = 5.7%). It is concluded that at low SCs, the viscosity of Newtonian BL can be accurately predicted using the new binomial model. Copyright © 2010 Curtin University of Technology and John Wiley & Sons, Ltd.