Benchmarking Spectroscopic Techniques Combined with Machine Learning to Study Oak Barrels for Wine Ageing

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Biosensing

Background:

• Oak barrels are crucial for wine aging, imparting flavor through compounds like lignin and tannin.
• Repeated use depletes these compounds, necessitating barrel replacement, but current estimation methods are empirical.
• Accurate assessment of oak barrel lifespan is needed to avoid costs and maintain wine quality.

Purpose of the Study:

• To investigate non-invasive optical biosensing techniques for characterizing oak barrels used in wine aging.
• To develop a model for predicting oak barrel lifespan without structural damage.
• To benchmark spectroscopic methods for assessing changes in oak barrel chemical composition.

Main Methods:

• Employed spectroscopic diagnostic techniques: reflectance, fluorescence, and Raman scattering.
• Assessed changes in lignin and tannin presence in oakwood barrels after repeated wine refills.
• Utilized chemometric techniques, including Linear Discriminant Analysis and Partial Least Squares Discriminant Analysis.

Main Results:

• Reflectance spectroscopy proved to be a valuable tool for characterizing oak barrels.
• Reflectance spectroscopy achieved the most accurate classification performance among the tested methods.
• A Partial Least Squares Discriminant Analysis model achieved >99% accuracy in classifying oakwood samples based on barrel usage.

Conclusions:

• Reflectance spectroscopy offers a cost-effective and non-invasive method for assessing oak barrels in the wine industry.
• This technique can reliably determine the remaining lifespan of oak barrels, optimizing refill strategies.
• The study provides a foundation for applying advanced biosensing for barrel management and quality control.