Last time on the blog, I went over the Yeast repitching project and why it was important for us to track the health of our four main strains over 12 generations. We found some important information, especially relating to how wort oxygenation helps some strains more than others. The previous post also explained how the yeast repitching experiment was set up and what kind of data we gathered from it, so if you missed it, be sure to check it out.
Going over yeast health is only one aspect of what makes a quality product, however. Ultimately, whether it tastes good or not the most widely relevant aspect of beer quality, applying to everyone from brewers to casual drinkers. That’s why for every R&D project we run several stages of sensory evaluation to ensure the best end results for every new product and process.
With the yeast repitching project, we were not only tracking yeast health, but also potential flavour profile changes. To do this, we collected the beer from each strain in each generation, and tasted each sample in at least 2 separate occasions, always using anchors for reference. We used a 9-point intensity scale for specific flavours in the beers. For most panels we had 8 trained tasters. This number is not optimal for sensory science, but we have to make do with what we have! As the company grows, so will our sensory panel.
After the sensory sessions, we built spider graphs that let us compare one generation to the next. At the end of the project we compiled the data for all 12 generations* and plotted them in a spider graph, to get a quick visual impression of the flavour profile for each strain, and if there is any generation that stands out form the others. Take a look below for an example of our Old World Saison Blend (OWSB) flavour profile.
Flavour profile of multiple generations of Old World Saison Blend.
To look at the big picture, we then take the raw data to continue our analysis and pick up any changes that are of interest. The graph below charts all the individual flavour components separately for each strain, and each bar graph represents how the perceived sensory score for each one has changed throughout all generations.
This is a great way to look at the overall data and see if there are potential trends that we want further analyzed. From this we can see right away that on the “spicy” column (second from the right), OWSB consistently scores higher than the other strains. The 4-VG produced by the saison strains in the OWS blend is perceived mostly as “clovey” and would be scored in our descriptive analysis as spicy, which is a confirmation that our sensory panel is working, and we can pat ourselves on the back.
A whole lot of sensory data, letting us look for trends. Changes for each sensory parameter for Cali (CA; red), Foggy (FO; green), OWSB (OW; blue), and Vermont (VT; purple) over 12 generations.
The sensory panel did a good job of repeatedly detecting the 4-vinylguaiacol (clove) note from the Saison yeast. Also, how often do you get to plot "spicy" on a graph?
When we look at the other individual flavour components in this graph, we can also see some patterns that stand out. The perceived “floral” and “fruity ester” scores (middle two columns) are going down steadily with continuous repitching. We can then look more carefully at this specific data and plot the mean scores for perceived flavour component against the generation, to determine the presence and strength of any actual trend that might be present.
The figure below paints a clear picture of how these two flavour components are decreasing in all strains tested. Vermont is the most noticeable, with marked decreases in both perceived floral and perceived fruity ester. Beers made with Cali and Vermont had the floral flavour notes decrease the most, whereas OWSB beers saw the largest decrease in the fruity ester profile, with Vermont hot on its heels.
Enhance! A clearer visualization of our data: trend of reduction in floral and fruity ester flavour through multiple generations of yeast repitching.
The sensory tests show us the real world limitations of using yeast for many generations. Each strain, with their own particular flavours and fermentation quirks, will change in different aspects and at different rates. One thing this tells us is that it’s always good to carry out sensory evaluations, and some strains shouldn’t be repitched indefinitely if the goal is flavour consistency or intensity. It might be possible to adjust pitching rates downward as generation increases to enhance the fruity and floral aromas, as long as the yeast is healthy.
This project was also presented at the 2018 Brewers Summit in San Diego, CA. We have made the slides from this presentation available here.
Of course, sometimes there are sensory fails.
Iz Netto is the R&D Biologist at Escarpment Labs, where she develops and improves on products and lab protocols. In addition to having lead a variety of projects including repitched yeast health, diastaticus contamination detection, and creating a robust Lactobacillus blend, she also contributes to Quality Control testing in the lab. Iz is a professionally trained brewer and has participated in Escarpment Labs collaboration brews with some of her favourite breweries. She is a member of Pink Boots Society, a global network of women working in the brewing industry.
* For the nerds: data analysis was performed in Excel (spider graphs and data collection) and R (statistics and data wrangling, plotting was done in ggplot2).