Saturday, February 20, 2016

Thoughts on Spitaels and Van Kerrebroeck et al, 2015

This post looks at a 2015 paper jointly led by Freek Spitaels (Gent U) and Simon Van Kerrebroeck (VU Brussels) who, with their co-authors, looked at the characteristics of bottled Cantillon gueuze of different vintages. The study was conducted in 2013 and the following gueuze samples were tested (years listed are bottling years): 2013 (one month in the bottle), 2011, 2010, 2008, 2004, and 1996. The authors also looked at 3 year old lambic to add to Spitaels's previous work on Cantillon lambic that stopped at 2 years old.

As mentioned above, Freek Spitaels has done some great work recently with Belgian lambic. I've seen two of his papers got some attention in the homebrewing and beer enthusiast world - one on the microbial succession of aging and fermenting lambic at Cantillon and the other following the same thing at a more industrial producer. For practicality, this paper looks at different blendings of gueuze which are of different ages rather than following one blending through time as it aged. The latter of course would be quite interesting - to properly follow the evolution of one blend. But 17 years of data collection is a pretty intense task. And it would be difficult to remain sufficiently consistent in a rapidly growing field such as those utilizing microbe biomarkers. And of course, the most important reason, it would not allow the comparative taste tests.

I quite like reading brewing scientific literature, though its not always that I find something that has a direct impact on me as a brewer/beer drinker. I'll often learn what is happening at given points and why the things that brewers do work. And that's not to say that I don't ever read a paper and think to myself that I'm going to do this new thing. And it's also not to devalue the papers that I don't come away from with a practical take away change in my brewing process. Those are still useful in understanding what is happening, which could help me down the line to change stuff  (and it is something that I am intellectually interested in anyway). But with this paper, I both learned about what is going on and took away clear useful information for both brewing and as a consumer.

I want to give a quick thanks to my friend an fellow homebrewer/trace metal ocean scientist Dave S who helped me by putting together a nicer figure than the janky plots you may have seen in the first couple hours of this post being up. Ok, so here are some of the highlights to me.

Viable microbes:

The authors found no culturable bacteria (by direct plating or enrichment) beyond 3 years in the bottle, but found culturable yeast in all the bottles. This was from directly plating dregs in the younger bottles but generally culturable microbes were only found after an enrichment process (30 mL filtered, then grown in media before plating). The authors term this viable but not culturable. The culturable/viable bacteria in younger bottles were generally Pediococcus damnosus and some acetic acid bacteria. Viable yeast were mainly Dekkera (Brettanomyces) and were mostly D. bruxellensis, with some D. custersianus, and  D. anomala, and with variability by year. Saccharomyces and Pichia isolates were found in 2013 gueuze after enrichment (not in direct plating). Starting with gueuze that had been bottled 2+ years, yeast were only cultured after enrichment. And in bottles that were 5 years old and older, only D. brux. was isolated.

The enrichment process may not be terribly dissimilar from a homebrewer pouring dregs into a small starter. Of course the lab is better set up to do this sort of thing with specific media, incubators, and general clean handling and culturing equipment. But the general idea remains the same - put the dregs in some liquid to try to get more viable microbes. So if, at home, you are trying to grow up some rather old dregs from an excellent and fairly old bottle of gueuze, you might find some yeast cells but you might also be better served by directing your efforts elsewhere. First off, if any viable microbes are present they (at best) will only be a piece of what produced that flavor and will only be a small group of what was active in the bottle. And as the number of viable cells goes down the likelihood of you contaminating your culture and having something you introduced take hold goes up. So perhaps you're culturing your house or something from the lip of the bottle rather than the dregs you're after. It says something that people who are well trained to grow microbes and who have a full lab to help them do so were not able to culture any bacteria out of bottles older than three years and were only able to culture yeast after enrichment from all but the youngest gueuze.

In younger gueuze the authors found a greater abundance of bacteria. In gueuze in the bottle for a few months there were about 1000 times more bacterial colony forming units (CFU, roughly equivalent to viable cells) than yeast CFU per mL. This contrasts 3 year old lambic, which has comparable levels or slightly more yeast than bacteria CFU. As shown by isolates from the bottles, as the gueuze aged in the bottle the balance of viable cells shifted toward yeast.

Flavor and aroma compounds:

In addition to the microbial data, the researchers also looked at changes in some flavor and aroma compounds. Lactic acid and ethyl lactate concentrations increased with aging while ethyl decanoate and isoamyl acetate decreased with age. Increasing lactic acid concentrations make sense with live lactic acid bacteria and carbohydrates in bottled gueuze. The lactic acid concentrations seem to level off around 2008, which also happens to be the point at which the researchers no longer find viable lactic acid bacteria. Ethyl esters can be synthesized by brett, so continued development of ethyl lactate in the bottle (especially as lactic acid concentrations increase) also make sense. And brett has been shown to lower isoamyl acetate levels. The tasting panel found the older gueuze to be less fruity, which matches with the ester data from these bottles compared to younger bottles.

In addition to the role of age, some of these trends may also be influenced by changes in lambic production over time. 17 years is long enough that there might be some minor changes in terms of preferred blends and also practices in barrel management as well as potentially brewing process and ingredients. And this range also covers the passing of the lead production role from father to son (Jean Van Roy started brewing on his own in the F season, 2002-2003).

Data from Spitaels et al., 2015. Comparison of aged gueuze samples. Figure by Dave S

Tasting panel:

On the beer consumer side, the big take-away was that the tasting panel preferred the 5 year old gueuze (as a reminder, they looked at 6 different gueuzes between the ages of 1 month and 17 years in the bottle). Speaking from personal experience, while certainly less thorough and not doing this side by side, I generally agree with this assessment. As far as aging goes I tend to prefer Cantillon on the younger side while I might prefer geuze from other producers with a bit more age. This could easily be influenced by choices in blending, and specifics of the brewing process in addition to the different native microbe community for each lambic brewer. But whatever the case, and although I definitely couldn't put a specific age as my optimal, I personally prefer Cantillon more in the range from release to a few years old rather than approaching or more than 10 years old.

Other notes:

Looking at the specifics of the three bottles tested per year, there is a lot of intra-year variability. I presume that all the bottles were taken from the same blending, so this is a surprising range covered by the bottles. Perhaps this has to do with small differences in packaging. And Cantillon got a new bottling line in December 2013 so all the beers from this study predate that and perhaps the new setup will limit this sort of variability in the future. But it was interesting to see that some bottles from the same year had up to 60% more lactic acid than others and up to twice as much acetic acid. Flavor- and aroma-active esters had similar differences, and in some cases bottles of one year had up to 4x higher levels of certain compounds than others. I guess this goes to show that each bottle of lambic is a unique product. I suspect this remains true even with identical treatment at filling as there will be differences depending on the variable integrity of corks and small changes in storage temperature.

The authors also discuss sugar composition in the aging gueuzes, with longer chain carbohydrates being consumed during aging (starting with the simplest remaining). This gives some additional idea of the sorts of sugars that brett and bacteria can ferment and the rate to which they do this when there is nothing else left.

The authors looked at DNA in the samples to give an idea of the microbes present that they might not have been able to isolate (but not whether or not they were still alive, as this technique wouldn't be able to determine that). This technique found a wider range of stuff, including some organisms that we don't think of making it very far into lambic fermentation like enterobacteria. Perhaps these cells got into the final product sometime during transfer or bottling. Again, this technique does not determine if cells are living and these things didn't show up in the culturable cells, but the presence of a more diverse range of stuff might indicate that some of the organisms whose DNA they detected played a role in bottle conditioning and/or flavor development.

Finally, with all the attention paid to fermentation and conditioning in barrels, I think it is cool to see a study highlighting changes that go on in the bottle. Especially with a beer such as lambic that ages well and that people may try to keep various vintages of for aging. The value of bottle conditioning and some extra time in the bottle is not unique to lambic. Some saison brewers are particular about the length of time and manner in which their beer ages in bottles before it is released. And I think keeping in mind the value of live beer in the bottle and the changes that continue to happen after carbonation has developed is helpful for the brewer who is interested in making excellent beers inspired by saison and lambic.

Wednesday, February 10, 2016

Thoughts on Johnson 1918 - a Belgian mashing system for low-strength beers

I thought it made sense to discuss George Maw Johnson's 1918 text about a Belgian mashing system suited for low strength beers to follow up on my last scientific/historical post on the Evans 1905 Brewing in France text, in which I also focused on mashing. Johnson had integrated well into the world of Belgian brewing and had gained respect from Belgian brewer-scientists and he was the inaugural editor for one of the leading Belgian brewing journals of the time. In addition to technical details, in this text he does a good job of setting the scene for the mashing procedure he covers with historical context and some driving forces behind this procedure. As with the Evans text and many other British texts of this era, the paper is followed by a discussion between other British brewing scientists/researchers of the time which makes for a great read.

Johnson starts by explaining the origins and duration of the Belgian tax law which centered on mash tun size. Many may have heard of this before but if you haven't, Belgium spent about 60 years basing brewing taxes on the size of brewer's mash tuns rather than something sensible like production/strength based like volumes and original gravities of beers, grain quantities, etc. The effect of this law was that it promoted using undersized mash tuns and overloading them with grain, and then devising mashing procedures to extract as much as possible from this mash of abnormal water:grist rations. This tax started in 1822 when Belgium was still under rule of the Netherlands and the law remained in place after Belgian independence in 1830, lasting until 1885. However, after this point it still remained an option for brewers, who for some time were allowed to choose between the mash tun law and a volume*OG-type law for their excise tax format.

The text is also written such that it is much more readable than modern scientific work and with other interesting historical bits of information hidden in there. Information like the original purpose of 'stuyk manden'  or brewer's baskets. And mechanical developments (mash extractors) that made this mashing method easier to conduct and at some point, allow brewers to overload their mash tuns even more. These extractors can still be seen at some lambic producers and I'll update this blog post with a picture of one if I am able to visit the right breweries in the coming months (for now check out the Timmerman's mash tun pictured in this Drink Belgian Beer post). The copper disks are extractors for pulling out wort.

Alright, on to the mashing system:
The system Johnson is describing is type of turbid mash. Johnson describes this as a 'modern turbid mash' and it seems pretty similar to what one might call a 'modern turbid mash' now as well. It might differ slightly from the 'Cantillon-type' turbid mash many might be familiar with, but in essence it is very similar. It is also similar in some ways to the mash described by Evans (and Johnson notes that this type of mashing was also employed in Northern France, so it makes sense that Evans describes something similar for Bière de Garde).

Goals: Perhaps goals isn't really the proper word here as it may not be the initial goals the Belgian brewers had in mind when they devised the system. Rather Johnson is discussing the reasons why he is recommending the mashing system/advantages he sees in it.

This mashing system is good for producing unfermentable wort for low strength beers so that they don't end up too thin. And in addition Johnson believed that the wort from such a mash both tasted better than a simple infusion mash of the same grist and that the wort was better for the yeast. While this sort of mashing can be well suited for mixed fermentation beers (as we know from lambic), Johnson describes the beers that formed the basis of this text as being consumed rather young.

The 'modern' turbid mashing process:

A simple schematic of the mashing process
-As a note before the steps, Johnson mentions in the text that the infusion temperatures may not seem to agree with the final rest temperatures stated. He explains that this is because of the thick iron mash tuns needing a lot of heat to warm up as well as a lot of heat loss from stirring and the mash tun design. So the temperatures really do work out this way, but applying it to modern equipment will need some adjusting (probably meaning use cooler water rather than less because of the mash already being on the thick side to start).

1) The mash starts with a cool (100-110 F / 37.8-43.3 C) and thick step with less than or equal to 1.5 barrels of water per quarter of malt. A quarter of malt is somewhere on the order of 336 lbs, so this means something like 0.5 quarts per pound (1.0 l/kg). When using raw wheat this step is carried out a bit cooler (note that this trend of 'if wheat the cooler' is true for many of these steps and Johnson mostly deals with a no-wheat scenario). Starting so cool helps to prevent a stuck mash, an especially important precaution when finely milled raw grain is used. The rakes are used for ~10 minutes and then the mash is allowed to rest for 30-45 minutes more. This allows the production of soluble proteins, swells the grain, and develops acid (both phosphate-system from phytins and microbially produced lactic acid).

2) Following this cool dough in/rest, 1-2 barrels of water per quarter (on the order ~0.37-0.74 quarts/lb, 0.77-1.5 l/kg) at 150-175 F / 65.6-79.4 C are added to bring the temperature up to 120-125 F / 48.9-51.7 C. Again, if raw wheat is used then this step is a bit cooler.

3) When the mash is up to temperature the first turbid portion is withdrawn. The rakes are not stopped for this turbid pull, which suggests that the turbid portion might be a bit chunkier than one might expect from seeing a modern turbid mash in action. Johnson reports that the volume and number of pulls varies (more raw grain means more pulls, if all malt maybe only one pull). The removal of ungelatinized starch from raw grain (or poorly malted grain) here helps to prevent stuck mashes as well. This turbid portion is heated to boiling in an extra kettle. If raw wheat is used, the turbid portion may be allowed to rest around 165-175 F (73.9-79.4 C) for about 20 minutes to allow some conversion to take place before boiling destroys the enzymes. After this the turbid portion is brought up to boiling and gently boiled until it is needed.

4) For all malt beers, an infusion of boiling water is used to raise the mash to 158 F / 70 C once the first turbid portion is withdrawn. The mash rests here from 45-60 minutes. Johnson notes that this saccharification temperature is higher than standard for English beers and he suggests that starches in Belgian malts may resist conversion at lower temperatures, possibly due to needing to be gelatinized.

5) The rest temperature yields a more dextrinous wort, which is drawn off to the boil kettle without a mash out.

6) After the mash is drained, the turbid portion is added back to yield a mix temperature around 167 F / 75 C. To hit this temperature cold water is added or the turbid portion is allowed to cool before it is added back. Johnson mentions that brewers were careful not to exceed this temperature. This mixture rests for 45-60 minutes and then it is transferred to the kettle. The previous wort in the kettle has still not been heated and the mix of the two runnings is about 160 F / 71.1 C, so enzymes remaining in the non-turbid runnings could act on any unconverted starches in the turbid wort.

7) The mash is now sparged normally with the sparge water being no hotter than 167 F / 75 C. It seems that the mash is mixed to some degree before sparging (suggesting this is more like batch sparging than fly sparging) and also that hulls are used to promote good flow.

Other topics of note:

Fermentation - Johnson lists typical attenuation levels for Belgian beers which are quite low. Much lower than the modern concept of Belgian beers as light-bodied, highly attenuated beers. Many of the beers which are the driest (saisons, lambics, etc.) would not fall into the category of 'typical Belgian beer' so they likely wouldn't be included in the sorts of attenuation ranges listed. But it is interesting to note that low attenuation was common for Belgian beers. Johnson lists typical apparent attenuation levels of around 66% and with 75% being too high. Again, keep in mind that this is speaking of beers drunk young and of more standard low-strength beer types.

Johnson also notes that pitching temperatures were on the warm side compared to English brewing, with yeast pitched at ~68 F / 20 C during winter.

There is an interesting discussion at the end of the text regarding yeast mutation and what makes a Belgian yeast different from a British one. There is possibly a bit of fanciful thinking and/or not clearly describing what is meant by some statements, but it is an interesting read either way.

Belgian sugar - In discussion with another brewing scientist, Johnson mentions that Belgian sugar producers felt that it was important to include non-fermentable material in their products. And therefore that the use of Belgian sugar in brewing may not have dried a beer out as much as we would suspect based on modern sugars. English sugars at the time of this text are discussed as being fully fermentable.

Storage: The beers brewed by the method that Johnson was describing were all consumed fully attenuated but young (fermentation was not stopped early so the real, albeit high, FG was reached). Johnson does note that some beers made by this method are susceptible to spoilage by wild yeasts. He also notes, as it pertains to how the beers kept, that lower strength beers from a given brewery which were brewed by this method tend to do better than higher strength beers.

Closing thoughts:

Johnson is pretty firm with saying that wort from the same malt tastes notably better with this turbid mashing system compared to a more normal infusion mash. Regarding the applicability of this sort of mashing to making highly attenuated beers - we know it works well for lambics. It might be interesting to try it in modern dry low strength beers like petite saisons, where a bit less attenuation might not be awful and where highly attenuative yeasts are being used. Also, with modern malts conversion would happen more quickly and therefore following this sort of procedure would yield a more fermentable wort now that with typical Belgian malt of the time of this article. I've generally not been a fan of high mashing temps for low fermentability in saisons or many mixed-fermentation beers (this is somewhat yeast-dependent) but the uniqueness of this procedure and the forgiving nature of lower attenuation in petite saisons makes this an attracting idea to me for at least a try. I think then, at least starting with beers that are intended to see a bit of aging with a mix of microbes, and perhaps some cleaner but really low OG beers, this sort of system is worth a try. And I plan to give it a go when I am back to brewing.