Under Pressure: Analyzing The Results From Past exBEERiments On Pressurized Fermentation

Author: Marshall Schott

For many, there’s a romantic appeal to the idea of shortcuts in brewing, especially one that doesn’t feel like cheating. Fermenting under pressure hits different than, say, fining a batch with gelatin or mashing for 30 minutes. It has an air of science to it. For homebrewers who lack a dedicated fermentation chamber or simply hate waiting around for a lager to finish its obligatory cold stint, pressurized warm fermentation has been sold as the great equalizer, affording one a sense of comfort when doing something that is widely viewed as brewing blasphemy. No wonder it’s received so much attention.

Here at Brülosophy, we do our best to avoid covertly leaning on cognitive biases such as appeal to authority, which has resulted in no fewer than 8 xBmts looking into the impact pressurized fermentation has on a range of styles fermented at different temperatures. Following publication of each of those articles, some readers would reach out expressing validation while others would accuse us of any number of issues, no doubt because the findings didn’t confirm their beliefs and practice.

Because of this, I felt it was a good time to review all of the xBmts we’ve performed on pressurized fermentation over the years, the hope being to make the collective findings more meaningful than each individual result. And just to be sure – we are fully aware our xBmts were not performed in a lab with fancy gear for measuring minuscule chemical differences, but rather neurotic homebrewers made the beers and served them to regular drinkers for evaluation.

| The Framework |

Before diving in, a quick refresher on how we run xBmts. We rely on the triangle test, which involves serving two identical samples of beer and one different sample of beer to tasters who are blind to the variable, then asking them to identify the unique sample. Using a binomial proportions test, statistical significance is reached when the p-value is lower than 0.05.

In layman’s terms, significance is reached when a certain number more tasters than would be expected by random guessing identify the odd-beer-out, and when this happens, we can say with some level of confidence that the variable had a perceptible impact. When the p-value is higher than 0.05, it suggests the variable did not produce a perceptibly different beer in that particular xBmt, not that the variable in question doesn’t have an impact. This matters because we’re not asking if a variable leads to a measurable difference; rather, we’re asking the more practically important question, “Can regular beer drinkers perceive a difference?”

Over the years, we’ve received plenty of feedback from folks who aren’t terribly keen on how statistics works, so let’s clarify a couple of things up front. First, a triangle test requires picking the unique sample out of three options, meaning tasters have a 33.3% chance of getting it right by pure luck. Randomness being what it is, the actual distribution of guesses from a panel rarely splits perfectly evenly, which is expected. Also, quite a few people have gotten hung-up on the preference question, which we ask only those participants who made the correct selection on the triangle test, because one of the options is that they didn’t actually perceive a difference between the beers. Here’s the thing – the math assumes a baseline of lucky guesses and adjusts the bar for significance accordingly. It’s bad science to ignore real data just because a taster admits they got lucky. This isn’t politics.

| The exBEERiments |

To keep things simple and clean, I’ll be providing a brief overview of several past xBmts on the topic of pressurized fermentation then sharing the pertinent data with as little fluff as possible, meaning just the participant count, the number of correct responses that were expected in order to achieve significance, the number of actual correct responses, and the p-value.

American Pale Ale

Our first foray into this topic was all the way back in 2015 and was inspired by contributor Greg Foster’s desire to implement “professional” practices in his homebrewery.  Having heard that some of his favorite commercial brewers were fermenting under pressure for various reasons, namely the ability to ferment beer at warmer temperatures without the risk of associated off-flavors, he designed an xBmt comparing an American Pale Ale fermented under 6 psi to the same beer fermented under no pressure, with both being held at 68°F/20°C.

Participants: 13
Expected: 8
Actual: 6
p-value: 0.33 – not significant

This being one of our earlier xBmts, the participant count was lower than we aim for these days, and we’d yet to hone our triangle test approach, which may have played a role in these findings.

Saison

Following release of that first xBmt, we received several comments from readers claiming that pressurized fermentation is more important for yeast-forward beers than hoppy styles. Since Saison yeast produces notable esters and phenols, which many believe is enhanced by fermenting at warmer temperatures, Greg designed a follow-up xBmt comparing a Saison fermented under 8 psi to one fermented under no pressure where both were held at 72°F/22°C.

Participants: 21
Expected: 11
Actual: 6
p-value: 0.68 – not significant

If the claim that pressurized fermentation suppresses ester and phenol development is true, we would have expected the beer fermented under pressure to have noticeably less fermentation character than the one fermented at ambient pressure, which wasn’t the case. Of course, it’s possible there was a difference in actual ester and phenol concentrations, but that it wasn’t vast enough to be perceptible.

Munich Helles

While the first two xBmts were on ales, we shifted our focus to lagers in 2016. This was due in large part to all the talk about commercial breweries pumping out delicious lager styles in record time by fermenting warm under pressure. Never one to accept a single, or I suppose even two, data points as fact, Greg yet again designed an xBmt looking at the differences between a Munich Helles fermented with Saflager W-34/70 yeast under 8 psi and one fermented under ambient pressure, both of which were held at 82°F/27°C.

Participants: 20
Expected: 11
Actual: 9
p-value: 0.19 – not significant

Seeing as these beers were fermented 32°F/17°C warmer than recommended, it’s possible both possessed the undesirable ester characteristics cool fermentation temperatures are intended to keep at bay. We’ll never truly know, though curiously, neither tasters nor the brewer noted any obvious off-flavors in these beers.

Festbier

Over the following few years, pressurized fermentation of lagers really seemed to take off, as it gave brewers permission to break rules of fermentation that had been etched in stone. Contributor Mike Neville questioned if the reason for the prior non-significant results might be a function of the pressure being too low, so he designed an xBmt comparing a Festbier fermented with Saflager W-34/70 yeast under 12 psi to one fermented under no pressure, both being held at 68°F/20°C.

Participants: 20
Expected: 11
Actual: 8
p-value: 0.34 – not significant

Yet again, the findings failed to validate popular claims about being able to ferment lagers warm if it’s done under pressure. Even Mike, who by virtue of designing the xBmt and brewing the beers was quite biased, felt the beers were identical.

German Pils

Still fascinated by the idea that pressurized fermentation was a sort of cheat-code for turning lagers around quickly, Mike opted to repeat the prior xBmt for Homebrew Con 2022 where he would be able to collect much more data. However, for this one, he used a new lager yeast, Imperial Yeast L26 Pilgrimage, in a German Pils where one was fermented under 13 psi while the other was left at ambient pressure, with both being held at 68°F/20°C.

Participants: 57
Expected: 26
Actual: 24
p-value: 0.10 – not significant

While this was the closest we’d come, the data still failed to reach the threshold of significance, and even Mike only picked the odd-beer-out in 1 of his 5 semi-blind triangle test attempts.

Belgian Single

Similar to Saison, Belgian ales are known for their fruity and spicy fermentation character, which many believe can be modulated by both fermentation temperature and pressure. Contributor Martin Keen was interested in revisiting how pressurized fermentation affects such a beer, leading him to design an xBmt comparing a Belgian Single fermented under 15 psi to one fermented under no pressure when both were held at a balmy 78°F/26°C.

Participants: 24
Expected: 12
Actual: 6
p-value: 0.86 – not significant

Just 25% of tasters got it right, less than what’s expected from random guessing, which allows for the presumption that even the correct selections were likely lucky guesses. At just 2 out of 5 of his own semi-blind attempts, Martin commented that the beers were identical in every way to his palate.

German Helles Exportbier

In our fervent quest (obsession?) to better understand the effect pressurized fermentation has on beer, contributor Martin Keen decided to test it out again on a German Helles Exportbier, which is slightly stronger than past lager styles. For this xBmt, he fermented two batches with White Labs WLP808 Mythical Hammer yeast at 68°F/20°C, one under 12 psi and the other at ambient pressure.

Participants: 20
Expected: 11
Actual: 6
p-value: 0.70 – not significant

Like all of the previously discussed xBmts, Martin’s performance aligned with the blind tasters, in fact he guessed wrong in all 5 of this semi-blind triangle test attempts.

In addition to our standard sensory analysis, we worked with White Labs on this xBmt to get a better picture of any objective differences between these beers, and the findings were fascinating. Indeed, the batch fermented under pressure had lower levels of isoamyl acetate, but it wasn’t much lower at all – just 0.11 ppm, which is quite a bit below the sensory detection threshold of 1.4 ppm. What’s even more curious is that levels of the other esters measured were higher in the beer fermented under pressure, as was diacetyl, though all differences were lower than commonly accepted thresholds for perception. Put simply – tasters couldn’t tell these beers apart because the objectively measured chemical differences were essentially nill.

Festbier: Pressure & Temperature

Our latest, and arguably most audacious, xBmt on this topic was performed by contributor Alex Shanks-Abel who, rather than holding temperature constant, was curious how a Festbier fermented at 68°F/20°C under 15 psi would compare to a version fermented at 50°F/10°C under no pressure. In essence, this was a comparison of modern vs. traditional approaches.

Participants: 26
Expected: 14
Actual: 9
p-value: 0.52 – not significant

On its own, this finding would likely feel validating to those who swear by pressurized warm fermentation, as it shows a pale lager fermented at ale temperature under pressure is indistinguishable from one fermented the “proper” way. But of course, this is rendered nearly laughable when viewed in light of all of the prior xBmts on the same variable. Even Alex, who was wholly biased, picked the unique sample in just 2 of their semi-blind attempts, and they acknowledged those were just guesses.

The Brü Club xBmt: German Helles Exportbier

Prior to joining Brülosophy as an official contributor, Will Lovell performed an xBmt as a member of The Brü Club where he compared a German Helles Exportbier fermented at 66°F/19°C under 20 psi to one fermented at the same temperature under ambient pressure.

Participants: 17
Expected: 10
Actual: 9
p-value: 0.08 – not significant

Out of his 3 semi-blind triangle test attempts, Will chose the odd-beer-out just once, exactly what we’d expect if he was randomly guessing, which is what he said he did.

| The Significance Of Non-Significance |

That’s a total of 9 xBmts on the impact of pressurized fermentation with nary a statistically significant result. Of course, we’d never accept this as proof that pressurized fermentation does not have the impact many claim it does, and that folks lean on it to alleviate the anxiety associated with going against tradition. As those who can’t tolerate when our data doesn’t confirm their strongly held biases often claim, this could be a function of poor xBmt design, lack of adequate controls by the brewer, or the oft cited shitty palates of the tasters.

But then, what reasons do we have to accept that fermenting beer under pressure has any notably perceptible impact? Let’s review some academic literature…

An experiment I’ve seen referenced many times did find that fermenting under pressure reduced levels of isoamyl acetate a substantial amount (Souffriau et al., 2022), but what’s rarely referenced is the fact those researchers intentionally selected a Japanese sake yeast known to be a high producer of this banana flavored compound. Additionally, they were inspired by reports from brewers who, having increased their fermentation volume due to demand, noticed decreased esters in their beers, yet the xBmt relied on a scaled down, albeit rather complex, setup. In an earlier experiment (Knatchbull & Slaughter, 1987), when researchers compared a beer with a baseline concentration of 2.6 ppm of isoamyl acetate (fermented at ambient pressure) to versions fermented under 7 psi and 15 psi, isoamyl acetate concentrations decreased to 1.6 ppm and 1.4 ppm, respectively – that’s a max delta of 1.2 ppm, which is still below the accepted sensory detection threshold of 1.4 ppm. Furthermore, while vicinal diketones (VDKs) including diacetyl ended up being nearly absent in the beers fermented under 0 psi (control) and 7 psi, the one fermented under 15 psi finished with a final concentration of roughly 0.12 ppm, which is right on the edge of being perceptible.

Ultimately, if you’re making beer that you and your friends enjoy drinking, it really doesn’t matter whether you stick to tradition or buck historical trends. If you’re the type who chooses to ferment under pressure because it alleviates your worries of a potential ester bomb, and you’re okay with the increased risk of diacetyl production, then by all means, keep doing it. For me, given the combined xBmt results, objective lab data, and hundreds of anecdotal reports I’ve received from brewers on every level, the juice just isn’t worth the squeeze.

If you have any thoughts about this xBmt, please do not hesitate to share in the comments section below!

Support Brülosophy In Style!

Follow Brülosophy on:

FACEBOOK   |   TWITTER   |   INSTAGRAM

If you enjoy this stuff and feel compelled to support Brulosophy.com, please check out the Support page for details on how you can very easily do so. Thanks!