Effect of Wood and Non-Wood Pulp Fiber Morphology on Laboratory-Simulated Tissue Paper Creping Performance: Functional Properties and Explainable AI

Abstract Fiber morphology influences bonding, sheet structure, and the mechanical response of tissue webs during creping, but these relationships remain insufficiently quantified for high-yield nonwood pulps. In this work, alkaline peroxide mechanical pulps (APMP) from wheat straw, bamboo, miscanthus, switchgrass, and sorghum were blended with bleached eucalyptus kraft (BEK) and northern bleached softwood kraft (NBSK) pulps to produce 30 g/m2 laboratory handsheets. After simulating high-speed laboratory creping, their physical and mechanical properties were assessed. The results show that fiber morphology is a key factor: slender fibers (such as NBSK and bamboo APMP) yielded higher tensile strength, while less coarse fibers (such as BEK) enhanced softness. Nonwood APMP fibers consistently increased tissue bulk. The creping process boosted bulk by 80–150%, but also decreased tensile strength by 80–89%. An exploratory machine learning framework utilizing Random Forest regressors was developed to assess whether precursor sheet measurements could assist in predicting properties in this 12-furnish data set. Cross-validation and comparisons with linear regression indicated that the predictions were consistent with papermaking physics and softness forecasting. Since papermaking properties of APMP pulps, including softness, depend on multiple variables, such as the type of machine technology used to enhance softness, generalization remains difficult. Overall, strategic blending of 15–30% nonwood fibers provides a viable route to balance softness, strength, and bulk in tissue products, while feedstock and process-dependent standardized data sets are required before data-driven models can reliably assist furnish formulation.