Abstract—Most of the existing models for describing black liquor (BL) viscosity behaviours are applicable over limited ranges of process conditions, whereas BL exhibits varied viscosity behaviours, Newtonian and nonNewtonian, over a wide range of process conditions. These limited-range models, resulting from different bases, may suffer predictions continuity over such wide ranges of conditions. In this paper, attempt was made to jointly model the Newtonian and nonNewtonian viscosity behaviours of literature liquor using artificial neural network (ANN) paradigm. A generalized multilayer feedforward network with 7 hidden neurons and 1 output neuron, having
R2=1.0 and maximum absolute relative error of ~8% between the actual and predicted data was obtained. Although a model with a higher accuracy is desirable, the proposed single network seems to be a reasonable alternative to the use of the limited-range multiple models for the purposes of describing black liquor viscosity.
Index Terms—Artificial neural network, black liquor, Newtonian, NonNewtonian, viscosity.
S. B. Alabi is with the Department of Chemical and Petroleum Engineering, University of Uyo, Uyo, Nigeria (e-mail: sundayalabi@uniuyo.edu.ng).
C. J. Williamson is with the Department of Chemical and Process Engineering, University of Canterbury, Christchurch, New Zealand (e-mail: chris.williamson@canterbury.ac.nz).
[PDF]
Cite: S. B. Alabi and C. J. Williamson, "Neural Network-Based Model for Joint Prediction of the Newtonian and NonNewtonian Viscosities of Black Liquor," International Journal of Chemical Engineering and Applications vol. 6, no. 3, pp. 195-200, 2015.
