Ultraviolet/visible absorbance trends for beverages under simulated rinse conditions and development of data-driven prediction model

Abstract

Water use during beverage production requires careful balancing of food safety, product quality, and environmental stewardship. Rinsing between flavor changes is one example where water can be used more efficiently. In this study, ultraviolet/visible (UV/Vis) absorbance properties of 20 beverages were characterized for undiluted product and diluted product (from 2 × to 5000 × dilution) to simulate rinsing. UV/Vis absorbance was a robust measurement and able to detect beverage components in rinse water. UV/Vis absorbance was also a sensitive measurement and able to detect most beverages at dilution levels of 1000 × or greater. UV/Vis absorbance was more sensitive than conductivity for most diluted products, and the sensitivity of UV/Vis absorbance was able to capture odor and taste thresholds for the beverages. A machine learning algorithm was used to develop a model to predict UV/Vis absorbance properties based on product ingredients and dilution. The results of this research suggest the possibility for in-line UV/Vis absorbance sensors to monitor and control rinsing and other processes during beverage production to optimize water use.