Evan Terrell, Isabel M Lima and Gillian O Bruni
USDA-ARS Southern Regional Research Center, New Orleans, LA 70124, United States of America; evan.terrell@usda.gov
Polysaccharide contamination (e.g., starch, dextran) is among the leading causes of decreased sucrose yields from sugarcane processing. Although there is a wide body of literature on management and mitigation strategies, there is little quantification of the associated economic impact of polysaccharides relating to recoverable sucrose losses. In this study, raw sugar manufacturing data was collected from daily reports shared by Louisiana processing facilities. Data was analyzed with machine-learning techniques (e.g., regularized regression, random forest, gradient boosting) to generate regression models for sugar per tonne-cane as a function of reported daily manufacturing parameters. The relative contributions toward sugar production associated with starch and dextran were quantified. These sugar per tonne-cane values were then interpreted as (unrealized) revenue based on raw sugar price in the U.S. Results suggest that dextran and starch values in the models contribute to roughly 0.1-0.2 kg of unrecovered sugar per tonne of sugarcane processed, relative to production amounts of 115-120 kg/t on average for the Louisiana industry. This equates to hypothetical losses of around 0.15%, resulting in unrealized revenue of up to roughly USD250,000 seasonally for an average sugar mill (or USD2,500 per operational day). These results are comparable with other technoeconomic analysis methods to quantify revenue loss from starch and dextran. Quantifying estimates for the economic impact of gums and starch during raw sugar manufacturing can impact processing strategies for sugar mills. Because these models suggest that unrealized revenue amounts are relatively low, special attention must be given to application costs of enzymatic processing aids like dextranases and amylases. Applying machine-learning techniques/big data principles, in conjunction with technoeconomic analysis, is ultimately another low-cost strategy that can be utilized by sugarcane processing facilities to better predict and improve sugar recovery and increase profitability.