Multivariate analysis of physicochemical quality parameters and production yield in sustainable sugar processing

This study investigates the multivariate relationships between physicochemical quality attributes of plantation white sugar and its production yield within the operational dimension of sustainable industrial processing. Daily data on five quality parameters ICUMSA color value, polarization, moisture content, sulphur dioxide (SO₂), and crystal size index (BJB) and production yield were collected over 15 consecutive days at a plantation white sugar processing unit in East Java, Indonesia. Three complementary analytical approaches were applied: Principal Component Analysis (PCA) for exploratory analysis of multivariate structure, Partial Least Squares (PLS) regression for assessing predictive relationships, and Statistical Process Control (SPC) using a Shewhart control chart for evaluating process stability. PCA revealed an interpretable multivariate structure in which ICUMSA, polarization, and SO₂ co-loaded on PC1 (33.88% of variance), while moisture content and SO₂ dominated PC2 (22.71%), together explaining 56.59% of the total variance. The two-component PLS model explained 22.27% of the variance in production yield on the calibration data (R²Y = 0.223), but leave-one-out cross-validation produced a negative Q² value (Q² = −0.976), indicating that the model lacks predictive capability for unseen observations. Variable Importance in Projection analysis identified SO₂ (VIP = 1.502) and ICUMSA (VIP = 1.143) as the most influential variables. SPC confirmed process stability, with all 15 daily yield observations falling within ±3σ control limits (LCL = 627.30 tons/day; UCL = 951.90 tons/day) and no non-random patterns detected. These findings indicate that output-side quality parameters alone are insufficient to predict daily production yield variability, suggesting that upstream process variables exert greater influence. Future research should expand data collection, integrate upstream process variables, and apply Life Cycle Assessment to extend the operational sustainability framework into the environmental dimension.