Comparison of polynomial and neural fuzzy models as applied to the ethanolamine pulping of vine shoots

Abstract

The influence of operational variables in the pulping of vine shoots by use of ethanolamine [ viz. temperature (155–185 °C), cooking time (30–90 min) and ethanolamine concentration (50–70% v/v)] on the properties of the resulting pulp ( viz. yield, kappa index, viscosity and drainability) was studied. A central composite factorial design was used in conjunction with the software BMDP and ANFIS Edit Matlab 6.5 to develop polynomial and fuzzy neural models that reproduced the experimental results of the dependent variables with errors less than 10%. Both types of models are therefore effective with a view to simulating the ethanolamine pulping process. Based on the proposed equations, the best choice is to use values of the operational valuables resulting in near-optimal pulp properties while saving energy and immobilized capital on industrial facilities by using lower temperatures and shorter processing times. One combination leading to near-optimal properties with reduced costs is using a temperature of 180 °C and an ethanolamine concentration of 60% for 60 min, to obtain pulp with a viscosity of 6.13% lower than the maximum value (932.8 ml/g) and a drainability of 5.49% lower than the maximum value (71 oSR).