Hydrolytic and soil degradation of cellulosic material (paper): optimization of parameters using ANN and RSM

Bandi Girish,Golluri Ricky Rakshith,Atanu Kumar Paul,Vinoth Kumar Raja,Gourhari Chakraborty
DOI: https://doi.org/10.1007/s00289-024-05319-5
IF: 2.843
2024-08-24
Polymer Bulletin
Abstract:This work reports the degradation of cellulosic material, i.e., paper, under different environmental conditions (soil and hydrolytic). Models were developed using central composite design (CCD) within the framework of response surface methodology (RSM) and artificial neural network (ANN) techniques. It is done as part of a solid waste management system to eliminate the waste produced by excessive paper usage and obtain optimized degradation conditions of paper. In view of the real environments where paper-based solid wastes are usually exposed, soil degradation and hydrolytic degradation of paper were investigated. The factors pH (4–10), compost ratio (2–5), CaCl 2 (5–15 ppm), and time (7–20 days) were independent factors that varied for the study, and degradation conversion was measured as the dependent variable for soil degradation of paper. The factors pH (4–10), CaCl 2 (2–5 ppm), tripotassium phosphate (TPP, 2–5 ppm), and time (6–16 days) were independent variables that varied for the study, and degradation conversion was the dependent variable for hydrolytic degradation of paper. The optimal conditions for soil degradation were determined to be a pH of 4, a compost ratio of 5, a salt (CaCl 2 ) addition of 5 ppm, and a duration of 20 days, resulting in the highest observed conversion rate. The coefficient of regression (R 2 ) for CCD is 87.24%, whereas it is 94.82% for ANN. The optimal conditions for achieving maximum conversion in hydrolytic degradation include a pH of 10, a CaCl 2 salt concentration of 5 ppm, a TPP salt concentration of 2 ppm, and a duration of 16 days. The coefficient of regression for CCD is 75.15%, while the coefficient of regression for ANN is 93.91%. In both deterioration scenarios, the data exhibited a higher degree of fit when modeled using ANN compared to CCD (RSM).
polymer science
What problem does this paper attempt to address?