A Spanish project sends 48 AI-sorted wine barrels into cellar trials

A research group at the University of Valladolid is working with artificial intelligence to sort oak wood for wine barrels by how much oxygen it lets through, in an effort to make barrel aging more predictable and consistent.

The work is part of Tecnobarricas, a collaborative project that brings together the UVaMOX research group at the university’s ETSIIAA school, the Cesefor Foundation, Ramón Bilbao wineries and other organizations linked to forestry and winemaking, according to information published Friday, October 30, by Plataforma AKIS.

The project focuses on one of the key variables in barrel aging: oxygen transfer through the wood. That measure, known as the oxygen transmission rate, or OTR, affects how wine evolves during aging and can influence its final quality. Oak barrels are central to the production of many wines, but their behavior can vary from one barrel to another because wood is a natural material with uneven structure.

Tecnobarricas aims to address that variability by classifying oak staves before barrels are built. The Cesefor Foundation has developed a system that uses machine vision and artificial intelligence to analyze the wood. An arch fitted with cameras captures images of the ends of each stave, and an algorithm studies the anatomical structure of the wood to predict how it will behave in relation to oxygen.

According to the project description, each stave can then be placed within seconds into one of three permeability categories: high, medium or low. That gives producers an objective, measurable way to assess the wood before it becomes part of a barrel, instead of relying mainly on visual inspection and the experience of coopers.

UVaMOX’s role is to test whether those AI-based classifications match what happens in real cellar conditions. The group specializes in oxygen management and control in winemaking. Its researchers are responsible for checking whether barrels grouped in the same category actually show similar behavior during aging and deliver repeatable results.

That scientific validation is a central part of the project. If the classification system proves reliable, it could give wineries a clearer idea of how a given barrel will affect the wine over time. It could also help reduce differences between barrels, a long-standing issue for producers trying to maintain a consistent aging profile across batches.

The project has already moved from the research stage into commercial-style trials. A first batch of 48 barrels made from wood sorted by OTR has been produced and distributed to wineries. Those barrels are being used to evaluate aging performance with Verdejo, Tempranillo and Garnacha.

The trials are meant to show more precisely how a controlled oxygenation profile affects the sensory development of wine. Researchers will be looking at whether wines aged in barrels built from similarly classified wood develop in a more uniform way than wines aged in conventional barrels.

For the beverage industry, the potential importance goes beyond a technical advance in barrel making. A more reliable way to match barrels to a desired oxygen profile could help wineries improve control over aging, reduce uncertainty in cellar management and make barrel purchases more strategic. It could also offer coopers a new tool to differentiate products using measured performance rather than only traditional selection methods.

The project also points to a broader shift in how data tools are being applied in sectors often defined by craft and tradition. In this case, artificial intelligence is not replacing human judgment in the cellar or workshop. Instead, it is being used to add a layer of measurable information to a process that has long depended on observation, experience and trial and error.

According to Plataforma AKIS, the work also has implications outside wine. Because oxygen transfer through wood matters in other barrel-aged drinks, the same approach could eventually be adapted for products such as whiskey or brandy. Any such expansion would depend on how well the system performs in wine and on further validation for other beverages, where aging conditions and quality targets differ.

The project also has a forestry component. By linking the physical structure of oak wood to a measurable enological outcome, the partners are trying to increase the value of forest resources and create a stronger connection between timber selection, barrel production and beverage quality.

For UVaMOX, the effort is presented as an example of applied research designed to move from the lab to the production floor. Its researchers are using their work on oxygen control in winemaking to test whether a digital classification system can support a more standardized barrel-aging process without changing the role oak still plays in shaping the style of the final wine.

Results from the 48-barrel trial are expected to determine whether the technology can be adopted more broadly by wineries and coopers. If the categories identified by the AI system correspond closely to real-world barrel performance, the method could offer the wine sector a practical way to manage one of the least predictable parts of aging.