Customising Yeast Starters (using the Homebrew Dad calculator)

[PeeBee]

If creating yeast starters, you probably use a calculator. The calculators will likely be based on experimental data collected by Chris White (White Labs) and published in the book "Yeast" with co-author Jamil Zainacheff. Kai Troester (Braukaiser.com) repeated the work to make it more relevant to stir-plate usage. Read about it here (note this work is over ten years old): Estimating yeast growth - Braukaiser.com.

My preference is for a calculator based on the Braukaiser work known as the "HomebrewDad Yeast Calculator" and accessible from >here<. The (editable) Excel spreadsheet file can also be downloaded from this link. Some other work at "Beersmith.com" that influenced the calculator is >here (Part One)< and >here (Part Two)< .

The "HomebrewDad" >Website< isn't very active anymore.

I set about hacking the Excel spreadsheet a couple of years ago to make it more useful; allow setting a "Brew Date" rather than use today's date, and things like that. But the multiple steps disturbed me: The next "step" assumed you were adding the yeast cells generated in the previous step, but ignores the starter liquid they must be in. I wanted to correct this, but it interferes with calculations based on experimental work. I felt this might "cross-the-line" of what can be extrapolated from experimental data (plenty of folk do just that - extrapolate well outside the realms in which the data was created which most likely results in very dodgy conclusions - actually, any conclusions derived from experimental data are likely "dodgy" unless proven repeatable many times by different "authoritative" bodies, but where do you stop?).

This is what I've done to hopefully stay on the straight-and-narrow...

[EDIT: Where-upon I noticed I'd introduce a difficult to find glitch. Sorted now, but the weather has come out nice so I'm off out. So, the "what I've done" follow-up is a bit delayed. But this is one of the reasons I write all this stuff up: It shows up any cobblers I've previously missed and produces a better end-result to present to you lot, who may point out further cobblers I've missed ... and so on.]

[PeeBee]

After an interrupted start having to fix an error I'd created in the spreadsheet, plus a change to very pleasant weather that I had to go out and enjoy: A resumption! A reason for doing this is in my previous post, but another reason is all this digging is turning up some interesting stuff about yeast build calculators that I'm sure some others will find valuable. The alternative (if not already knowledgeable in yeast starters) is to approach yeast build calculators as objects of magic (as I guess I once did).

The author of this spreadsheet was very helpful in separating Kai Troaster's yeast growth work to an entirely separate page in the spreadsheet. This page has the yeast growth modeling, the "magic" if you like, and is the "engine" of the calculator which I don't want to interfere with. The main page (where you enter details) supplies details to the "engine" and receives the yeast growth information.

YEAST BUILD CORE ENGINE
First "revelation" (well, it was to me): If this is the "engine", it must be immensely complex and rely on all sorts of details? ... Not a bit of it! The "engine" (or "model" if you prefer) draws a line (a flat, straight, line, but it does have a downward kink in it) on a virtual graph that roughly follows the data gathered from experiments. Provided with the information to describe where along the line best describes the growing environment, the "engine" returns the cell growth expected. And it only wants TWO bits of information! The starter's "cell count" and the amount of food (malt extract) made available.

What? No volumes? That is what the rest of the spreadsheet deals with. The "growth" calculations are independent of volumes.

The "food supply" is malt extract, of which 25% is excluded as "not food". This reflects a typical attenuation rate of 75%. Note this provides a potential amount of food to calculate the potential growth of cells. It does not mean the starter must be attenuated by 75% to achieve the predicted growth rate! Most of the yeast growth has been achieved before 5 or 10% of the food supply is gone. I mention this because I'll reuse the term "attenuation" later, but that will be referring to actual attenuation.

VOLUMES
The "engine" isn't concerned with volumes, but we will need to tailor the starter to work for a particular volume. This is what the rest of the spreadsheet calculates and is what my changes interferes with. But first, the spreadsheet does establish some "goalposts" to keep things in sensible limits: The first is gravity (SG) of the starter media which is one of the initial (global) settings, and the second the "Inoculation Rate" (in millions of yeast cells per millilitre) which is a setting per starter step. The "goalposts" establish some sensible limits on the information fed to the "engine". In this calculator the limits aren't enforced ... ignore them at your peril!

The author has commented these "goalposts" to provide sensible values. And the "Inoculation Rate" turns red if you go outside its suggestions (control it with the "Starter Volume" option for the step).

STARTER STEPS
This starter calculator allows four steps to be accommodated. Breaking the starter down into "steps" avoids the results dictating enormous starters that may be sitting around for two or more days before showing any signs of being done. Multiple step starters allow planning a time predictable outcome around which you can plan your activities. Ideally, you'd aim for each step taking one day. But beware: "Steps" are basically calculating an outcome based on the results of the previous "step". So, if the "error" is plus or minus 15% (the estimated error for this calculator), the error will be twice that for the next step (30%), twice that for the third step (60%) and by the fourth ... well? 120%?

In reality it shouldn't be as bad as that! An under-estimation in the first step (or two) should start to be countered by an increasingly over-estimation in the subsequent steps. Or at least that is how it's been working for me.

And when do you know a step is "done"? If you can see activity in the starter, it's probably done (as many extra cells as predicted have been created). But this isn't a full-proof method, I have had starters ferment out when I'm not looking and so continue to look inactive and not yet started. More reason to keep each starter relatively small and "predictable".


Enough for now! Here's a taster clip of my current spreadsheet (download the original from the link above):

Spreadsheet_Clip1.JPG (73.29 KiB) Viewed 2775 times

[EDIT: Updated spreadsheet clip]

[PeeBee]

Before delving into the main body of this spreadsheet, I'll explain some of that snip above. It has some significant alterations to the original.

Firstly, ignore the dotted "patterns" and use of purple colour. That was me tracking alterations; they're gone now! [EDIT: Clip updated!)

"Starter Details". This gained two new cells; the date of "brew day", and when to create the starter in time for that day. The original worked from today's date which was awkward for planning in advance and for retaining a record for future reviews. They were typical older modifications.

"1st Starter Step". Ignore the "Crisp's Liquid Malt Extract ...", it was added when I was having trouble getting hold of dry malt extract a couple of years ago. I can safely remove that now. [EDIT: Clip updated; reference to "Crisp's Liquid Malt Extract ..." removed.]

What follows is the meat of the customisations:

The original single line of cells (in each "step") has been split in two and staggered. There are three additional calls entitled "With 'Carry Over' ". The reasons for these additions will be the subject of a full-on rant about "starter calculators" later (not just this one I'm customising, but all - as far I can tell - "starter calculators"). "Total Volume" includes the "Starter Volume" plus the previous starters volume (in this case, the "1st Step", the liquid yeast contents volume) - it does make a significant difference, as does the next: "Total Extract" is the preceding "Dry Malt Extract ..." plus any remaining from the previous step. "Attenuation" allows calculating the amount of extract from this step that will be available for the next step. "Attenuation" can be altered, the default of 33% was selected because we are probably all guilty of allowing a "step" to run too long (some insist on allowing a step to ferment out completely). Some of my steps will possibly only have attenuated <5%.

The original's documentation does refer to "attenuation", but that's "potential", not "actual".

The next three cells are the originals that have barely been touched. But the results in those three cells will be greatly impacted by the last three ("Total Volume", etc.). The last cell fixes a bug in the original. I know it's a bug because what it calculated and the attached comments don't match. "Growth Factor" now displays "1" for no growth, not zero (multiply by one and the result doesn't move).

[Cobnut]

I'm waiting for you to post your modified spreadsheet

[PeeBee]

I'm waiting for you to post your modified spreadsheet

Hang-on...

I've been fixing a glitch in a part I didn't use ("Late Harvest") and I've got to arrange something to host the spreadsheet (I'm pretty sure this forum doesn't accept attaching Excel files).


This malicious brain-dead forum software trashed the rest of this post, so I'll try again later.


[EDIT: Getting multiple pictures on a post in this forum can drive you nuts!!!]

[PeeBee]

Here's another "taster" continuing from the one above:

Spreadsheet_Clip-ii.JPG (114.38 KiB) Viewed 2745 times

I've "enforced" a point for decanting and put harvesting over-builds to be either one side ("early") or the other ("late") of this point. It allows for some planning of these functions. The "4th Step" can now optionally be the "Last Step" as illustrated here (where the "4th Step" is actually the "3rd" and last step: These changes are all to improve planning. Chilling (to settle yeast before decanting) can also be planned over a day or two, further reducing the starter's ability to "dictate" the "Brew Day".

I'll also post here a clip off the original spreadsheet covering the bits in my last two clips. I have bloated the spreadsheet a little! But notice how the "Total Cells" differs in my calculations. This isn't an error I've found in the Homebrew Dad calculator, it's a common error in all the calculators I've seen. It's been allowed to pass because it only affects a small percentage of predominantly British brewers (those brewing "cask conditioned" styles) and the reason I want to have a good rant about "yeast pitch calculators".

Spreadsheet_Original-i.JPG (94.94 KiB) Viewed 2745 times

[PeeBee]

From the illustrations above it should be clear that I've bloated the spreadsheet by quite a lot. But I have my reasons! Initially I thought it was me at fault for attempting to modify something I only had very limited knowledge to modify: Namely the "Yeast Growth Engine/Model". I've since found the engine is only concerned with two parameters: The number of yeast cells and amount of food initially in the starter. The volumes are important but independent of the "engine" as the engine will scale to whatever the volume is. And many of us have an idea of volumes, starting with playing in the water baths at infant school.

Now to that rant! Remember I'm not picking on the "Homebrew Dad" yeast calculator, all the calculators I've used do it, this "Homebrew Dad" yeast calculator just makes it clearer (more "open" and capable of more steps).

Look at the last illustration (original unmodified spreadsheet). In steps 2 and 3 (last step) the "Inoculation Rate" is "red" because it exceeds the recommended 100 million cells per millilitre. Yet in my version, dealing with the same information, the "Inoculation Rate" is well below the recommended "100" (in my variation the last step is the fourth, the third step is empty). Why is this? The common "calculators" assume each step starts afresh with the starter volume plus the yeast cells that have been miraculously extracted from the liquid of the previous starter or retail pack. I wasn't comfortable with this and include the liquid from the previous starter or retail pack. Result being, it's much harder to exceed the recommended "Inoculation Rate".

Most calculators don't let you see this going on, but they are probably doing it all the same.

And why should this matter? It doesn't! The result is the calculations over-estimate the likely result. This moderate over-pitching should result in cleaner tasting beer free from the peculiarities of the yeast. Excellent result if brought up on the tasteless keg cr&p of old. Bit of a downer if trying to reproduce "traditional" flavours in say "British" beers (apparently - ??? - yeast should be slightly stressed at the beginning of fermentation to develop all the flavours it is reputed to have).

Most (all?) yeast calculators have an inbuilt prejudice (unintentional or malicious) against "traditional" British beer.

I also calculate residual malt extract in the liquid carried over from the previous step which improves the "Total Cells" prediction for that step. This is all done without interfering with the yeast growth model. Rant over! Before I start on the often arbitrary "Pitch Rate" ...

[PeeBee]

I've sort of got this modified version done, I'll get a link to a downloadable Excel file shortly (should work in Google Sheets but need to check it).

The snip that follows is an actual used starter modified with "live" data (starter attenuation and decant volumes) which I followed to keep an eye on how this starter progressed.

YeastStarterCalculator.jpg (490.96 KiB) Viewed 2663 times

Some substantial "bloating" of the original spreadsheet, most of which is related to "real" volumes in use which affects "target pitch rates", especially at lower end of the pitch rate range.

Summarising the changes: The volume of the liquid yeast starter pack has been added to "Yeast Details". This volume is added to the 1st Starter Step to make its volume more accurate. (This was a Wyeast pack which has a volume of around 0.13L). "Starter Details" acquired the actual dates for starting the Starter in time for Brewday. I'm not sure these calculations are entirely satisfactory yet.

Then begins the starter steps: The Information for each step is now split over two lines to accommodate the carry-over of volume and available "food". The "attenuation" is the yeast starter's attenuation and will apply to the following step. Unused steps get shaded out for clarity.

"Decanting and Overbuild Harvest" provides a stricter regime for these steps. It applies at the penultimate or (optionally) final step. "Settling" before decanting is given its own calculations, from a few hours to a couple of days. Settling obviously increases the cells available after decanting but is a process of diminishing returns (more time settles more cells but increases the number that die meantime). The data is entirely guess-work but discourages "settling" for more than one or two days.

Finally there's the 4th step, or last step if previous steps have been left out (like the 3rd step in this example). Then some final statistics for the finished starter.

In reality!

Step 1 went to plan. Sign of fermentation had begun before being mixed into step 2 (after 24 hours). Attenuation was recorded as 2% (remember, we are growing cells not fermenting starter medium!).

Step 2 was a worry. Seemingly increasing sediment but no sign of fermenting. It was given an extra 10 hours, still no signs and dropped into a fridge. After 18 hours it was removed to give it time to warm before the third and final step. Brewday was delayed a day. Where upon the starter did warm up and it erupted into action! An amount was carefully decanted off and discarded and final step begun. Attenuation was 33% (note "attenuation" is informational only and there is no good reason to record it).

After a day Step 3 was clearly ready, but Brewday had been postponed a day. 24 hours later it was pitched into the new brew.

1909TrumanX.jpg (27.95 KiB) Viewed 2663 times

The starter calculator hadn't failed at all. But anxieties during testing had resulted in a pitch rate that could be estimated as 0.85million-cells/[ml * °P] (target was 0.75)... but still a good deal less than this calculator would have created. But the resulting fermentation is barely affected (<3 days, but seemingly more sedate?). Must try harder not to panic next time!

[PeeBee]

Link to this modified "Homebrew Dad yeast starter calculator" I've been fiddling with. I've been having trouble with a very difficult to follow problem where numerous cells suddenly show "#N/A" (where upon the spreadsheet was trashed and had to be reloaded from a good copy - "undo" unavailable!). Seemed to be something to do with reiterating calculations. At the moment it's "cured" by allowing the maximum iterations, reducing accuracy of final iterated solution, and rearranging my modifications on the "Dropdown Lists" page (no idea why the latter makes a difference).

Homebrew Dad v1.X Baseline 11_22 GOOD

Excel spreadsheet, it won't work in Google Sheets.

It's loaded up with details for my next big brew which is built on an "over-build" for a preceding smaller brew. If doing things like that be aware that it will compound errors (whatever calculator you use). My modifications are attempting not to over-estimate "pitch rates" at the lower end of the scale (say 0.75 million-cells per ml/°P, or "Ale" rates); don't bother with this modification if targeting "lager" or "hybrid" rates. I'm aiming to build low pitch rate starters of about 0.5 - 0.6 million-cells per ml/°P. Don't bother with this modified version if you only build "single step" starters either.

And most importantly: This is still "work-in-progress" and may do things it wasn't intended to do. It will likely change.

Credit to "Homebrew Dad" (Olan Suddeth) who may also be the author ... I presume so but I haven't confirmed that yet! Original version here: Homebrew Dad's Online Yeast Starter Calculator