"The valorisation of brewery derived CO2 and distillery derived pot ale in relation to more sustainable food production."

In this short introductory presentation Robbie will focus on fermentation scale-up challenges including speed, efficiency, process safety, and product safety.

"Counterfeit Spirits Detection."

Illicit and counterfeit alcoholic spirits represent a global crisis, accounting for 25% to 40% of total worldwide consumption and causing thousands of annual deaths due to toxic adulterants like methanol. Beyond public health risks, these activities result in massive economic losses, including billions in lost revenue and tens of thousands of lost jobs within the European Union alone. To combat this, we are developing rapid detection methods by building an extensive database of thousands of spectra using five distinct analytical techniques, including NMR and NIR. By applying chemometric tools, including Principal Component Analysis (PCA) and Unified Manifold Approximation and Projection (UMAP) to this data, we have successfully identified "fingerprints" that differentiate authentic spirits from counterfeits with high accuracy. This work highlights the potential for Near-Infrared (NIR) spectroscopy to serve as a fast, robust field-testing tool for identifying suspect-bad products.

"Impact of (Very) High Gravity Fermentations on the Characteristics of Distilled Spirits and Carbon Footprints."

Understanding and controlling fermentation kinetics is critical for improving alcohol yield and consistency in spirit production. Under Very High Gravity (VHG) fermentations (>1.074 OG), full optimisation of the process under these stress conditions often requires extensive lab trials. This study presents a dual approach, combining predictive modelling with experimental optimisation to enhance fermentation outcomes and reduce wet-lab experimental work.

An artificial neural network (ANN) was developed to model fermentation kinetics using time-series data of cumulative CO₂ evolution. The model was trained on six original gravities under a set temperature curve and pitching rate, using a commercial active dried yeast (ADY) strain of Saccharomyces cerevisiae. A smoothed curve was generated by the trained ANN to extrapolate kinetic information and resulting in a predictive model with high accuracy (R² = 0.999) for unseen fermentations. From this format, the ANN was then optimised to predict the fermentation curves of “unseen” gravities.

In parallel, a Design of Experiments (DoE) framework was used to investigate the impact of yeast rehydration conditions on final ABV%. Variable osmotic pressures were applied to ADY via rehydration media at set temperatures, using two commercial S. cerevisiae strains and three pitching rates. The model explored the hypothesis that osmotic shock during rehydration could restore oxygen consumption capacity, improving stress tolerance. The resulting predictive model (RMSE = 0.256, R² = 0.96, p < .0001) enabled simultaneous testing of multiple variables, identifying pitching rate and yeast strain, as well as original gravity with rehydration media, as the most influential paired factors for ABV% optimisation under VHG conditions.

While linked in design, the independent results demonstrate the potential of integrating machine learning and statistical modelling with targeted experimental design. The talk will present the ANN architecture and kinetic modelling approach alongside DoE results, as well as the future potential for creating a stacked model, offering opportunities for early-phase optimisation and reduced experimental waste in VHG fermentation systems.

"Data-driven assessments of spirit quality and cask health."

The maturation of Scotch new-make spirit (NMS) in American oak casks imparts flavour and colour producing a liquid with golden hue and complex chemistry, while simultaneously depleting the cask of these elements. Once depleted, the cask is classified as exhausted and is required to undergo a process of rejuvenation. Due to the heterogeneous nature of oak, rejuvenation produces barrels with a broad spectrum of chemistry that can make it difficult to produce a consistent flavour outcome without informed blending decisions. The quantification of flavour compounds in the maturing spirit is therefore essential to ensure premium outcomes. The presentation will examine the use of UV-VIS spectra and machine learning models for the rapid assessment of flavour in grain whisky to aid the production of premium whisky products from rejuvenated casks.

" Exploring how Scotch malt whisky fermentation microbiota influence congener and sensory profile of spirit character."

The microbiome of whisky fermentations encompasses a number of different bacteria and fungi. Anecdotally, these microbes are crucial for flavour development in whisky-making however there is little research literature exploring how these microbes contribute to flavour in whisky. To help understand how fermentation microbiomes vary from distillery to distillery, the SWRI conducted a survey of 40 distilleries. Subsamples of fermentations were collected from each distillery and underwent DNA sequencing to profile their bacteriome via 16S sequencing. This presentation will cover the results of the fermentation bacteriome survey. Process information was also collected and was factored into results interpretation to try and determine how process may influence fermentation bacteriome.