Simulation of tagasaste pulping using soda-anthraquinone

Area of Science:

• Biomass utilization
• Pulp and paper technology
• Artificial intelligence in chemical engineering

Background:

• Tagasaste (Chamaecytisus proliferus L.F.) is a potential lignocellulosic resource for pulp production.
• Optimizing pulping processes requires accurate prediction of operational variables' impact on pulp properties.
• Existing predictive models may lack the precision needed for complex pulping systems.

Purpose of the Study:

• To develop a neural network model for predicting tagasaste pulping outcomes.
• To assess the influence of operational variables (temperature, soda, AQ concentration, time, liquid/solid ratio) on pulp properties (brightness, traction, burst, tear indices).
• To compare the predictive accuracy of the neural network model against a polynomial model.

Main Methods:

• Utilized published experimental data from tagasaste pulping using soda and anthraquinone (AQ).
• Developed a predictive model based on neural network architecture.
• Employed a factorial experimental design for model validation and comparison.
• Evaluated model performance based on prediction precision for key paper sheet properties.

Main Results:

• The neural network model effectively predicted the effects of pulping operational variables.
• Key pulp properties such as brightness, traction index, burst index, and tear index were accurately forecasted.
• The neural network model demonstrated superior prediction precision compared to the polynomial model.

Conclusions:

• Neural network modeling provides a highly precise tool for predicting tagasaste pulping performance.
• This approach can aid in optimizing pulping conditions for desired paper sheet characteristics.
• The findings support the use of advanced computational models in biomass processing for enhanced efficiency.