Empirical modeling of chemical consumptions in kraft pulp bleaching

Abstract

The aim of this study was to investigate the opportunities for empirical modelling of oxygen delignification and ECF-bleaching of craft pulp. Previously published results for birch, eucalyptus and southern pine were modelled with linear multivariate regression. The outcome was ambiguous as adequate models were not obtained for all datasets.

In the modelling of oxygen delignification statistically significant models were obtained for both the keppa number and ISO-brightness. For one dataset, the obtained brightness model predicted the delignification better than the kappa number model. It was also shown that the HexA-free kappa number is generally a more suitable dependent than the kappa number.

In the modelling of the ECF-bleaching it was shown that satisfactory models can be obtained by utilizing initial pulp properties and chlorine dioxide charge as predictors. The accuracy was improved if important process parameters such as D₀-temperature could be incorporated. However, none of the attempts succeeded to include oxygen or hydrogen peroxide charges in the models. Based on the analysis of the bleaching models, it was suggested that the bleaching should preferably be modelled stage by stage, especially if it incorporates more than one bleaching chemical.

Nonlinear models were also tested, but without improving the model accuracies. This and other problems are discussed, and further investigations are recommended.