Neural fuzzy model applied to ethylene-glycol pulping of non-wood raw materials

We studied the influence of the operational variables (viz. ethylene-glycol concentrations of 50-70%, temperatures of 155-185 degrees C, times of 30-90 min and numbers of PFI beating revolutions of 500-1500) on pulp yield and various paper properties (breaking length, stretch, burst index, tear index and brightness) obtained in the ethylene-glycol pulping of vine shoots, cotton stalks, leucaena (Leucaena leucocephala) and tagasaste (Chamaecytisus proliferus). The fuzzy neural network models used reproduced the experimental results with errors less than 15% and smaller than those provided by second-order polynomial models in all cases. An ethylene-glycol concentration of 65% at 180 degrees C for 75 min and 1500 PFI beating revolutions were found to provide substantial savings in energy, chemicals and facility investments as a result of operating under milder conditions than the strongest ones studied in this work. Tagasaste was found to be the most suitable raw material among those tested as it provided the paper sheets with the highest breaking length (4644 m), stretch (2.87%), burst index (2.46 kN/g), tear index (0.33 m Nm(2)/g) and brightness (40.92%); its pulp yield was also high (62.88%), which reflects efficient use of this raw material.