Determining % Alcohol
- Thread starter Tuckahoe
- Start date
Help Support Homebrew Talk:
rycov
Well-Known Member
google search. ABV calculator. just type your readings and your temperature in there and it will tell you
remilard
Well-Known Member
You can't, you can only do it with lab methods.
You can estimate it using that information with an approximation borrowed from wine making: 0.131 * (OG - SG).
You can estimate it more accurately using some formulae due to Balling which were summarized in an HBD post by George Fix several years ago (should be enough info to google it).
You can estimate it very quickly in your head as (OG -1) x 100%
There are dozens of formulas out there. The aforementioned Balling formula multiplies the difference between the OG and FG (AE) by a factor which depends on the OG. For example if the OG were 12 °P and the AE 3 (75% apparent attenuation) the ABW is approximately 0.421(12-3) = 3.789 i.e. 3.8%. Divide this by 0.791 for ABW. The multiplier doesn't change much in the region around 12 °P (1.040) OG. Values calculated by this, or any of the other formulas, are approximations.
but that doesn't take the aggressiveness of the yeast into account at all. with that calc, windsor and wlp001 would be just as effective as nottingham.
![Craft A Brew - Safale S-04 Dry Yeast - Fermentis - English Ale Dry Yeast - For English and American Ales and Hard Apple Ciders - Ingredients for Home Brewing - Beer Making Supplies - [1 Pack]](https://m.media-amazon.com/images/I/41fVGNh6JfL._SL500_.jpg)
i just take the attenuation supplied by the average i find out on the internet, plug it into beersmith, and am extremely close ( or dead on )
remilard
Well-Known Member
Right, but all of the formulae you can plug OG and FG into and get a number don't take a number of things into account either. You can't tell what the alcohol content truly is without a lab. The question is to do you want to get out a spreadsheet and be close but wrong or do you want to do something in your head and be close but wrong?
I think you'll find that any two yeast will give fairly close attenuation in the same wort. The numbers reported by the yeast suppliers more reflects the fact that Belgian wort is more fermentable than English wort (for example) than it does that Belgian yeast is more attenuative than English yeast. Mash at 145 for 3 hours and add 3 lbs of sugar per 5 gallons and any yeast will get you 80%.
Yes, actually, it does, if by the agressiveness of yeast you mean their ability to ferment i.e variability in attained ADF. The Balling formula is
ABW = f*(OG - AE)
but AE = OG*(1-ADF) so
ABW = f*OG*ADF
ADF is expressed as a fraction here rather than a percentage.
I was too lazy to look up the formula for f last night but it is
f = 0.39661 + 0.0017091*OG + 1.0788E-5*OG*OG
with the OG in °P. This is from a curve fit to the Balling Table as published in DeClerck, VolII p 428.
What Ballings formula does not take into account is that different yeast strains produce different amounts of biomass in the course of a fermentation. As the nominal production is 5.5% of extract consumed, and the tables are based on that, it is clear that doubling or halving of the yeast mass relative to the nominal doesn't have that much of an effect. It also relies on Tabarie's principle which, in my brief experiments with it seems a bit iffy. But then as you don't really know what your OG was (evaporation loss in the fermenter, dilution/concentration by the sugar content of the starter etc.) you couldn't really expect to be able to calculate alcohol content accurately from the formula anyway. It is, given this, often the practice that the alcohol content is measured and an effective OG calculated from that by using Balling's formula in reverse. Of course the limitations apply in the reverse direction as well so we seem to have a sort brewing Heisenberg principle: you can take a good OG measurement and calculate approximate alcohol content from that or take a good alcohol measurement and calculate approximate OG but you don't seem to be able to get both accurately.
There is another Balling formula
ABW = g*(OG - TE) where g is another factor depending in OG
g = 0.48394 + 0.0024688*OG + 1.5609E-5*OG*OG
and TE is the true extract. This takes Tabarie out of the picture and so is doubtless more accurate but the variability in biomass produced per unit of extract consumed is still problematical and few homebrewers are willing to go to the trouble to determine TE though it isn't really that difficult to do and requires minimal equipment in addition to a hydrometer i.e. a beaker and a volumetric flask.
The Balling equations may be the best out there because they take OG into account whereas the others don't but there are still limitations. In my experience the OG/AE form of the Balling equation, based on a specific gravity reading in the fermentor just after filling, underestimates the alcohol content of the beer by a few tenths of a percent. With respect to back calculation the effective OG is somewhat higher than the measured OG.
I'll go with "When I record the gravity readings in my brewing software, it automatically calculates a number that is close but wrong." Since that's how I record the gravity numbers anyway, it's literally zero extra work
+1 to the OG/FG calculations being inaccurate though. Some may be aware that Southern Tier Brewing recently revised the ABV on all of their beers, with some of them changing quite a bit. There's been no announcement that I'm aware of, but the public presumption is that they were listing the numbers based on OG/FG readings, and then for whatever reason (an audit? mere curiosity?) they got them lab-tested and found they were significantly off.
The calculations based on OG/FG assume that the only chemical change in the wort/beer is the conversion of sugar to alcohol. But there's all sorts of other compounds being converted from one thing to another as well, and those are all going to affect the density too. The assumption is that this contribution is small compared to the sugar->alcohol conversion, and that is true... but it does mean the ABV numbers derived from simply measuring the before and after density are not going to be perfect.
If Southern Tier's revisions are anything to go by, then for average gravity beers the results given by reasonably accurate gravity readings should be pretty damn close; whereas for higher gravity beers, all bets are off. I am speculating here, but I would guess that's because the other compounds consumed/produced by fermentation start to contribute non-trivially to the density. But I really don't know.
Balling's assumptions were that 2.0665 g of extract produces
1 gram of ethanol
0.9565 grams of CO2
0.11 grams of yeast
Okay, sorry... my point still stands though: By necessity, no calculation based on density readings can possibly take EVERYTHING into account (even things like esters have to affect the density a little bit, I would think).
Your basic point is, of course valid. Production of lots of ethyl acetate,for example, would throw the readings off both because extract shunted to EtAc isn't going to ethanol and because the EtAc will effect the specific gravity of the solution. But these substances are in the mg/L whereas the yeast has appreciable mass. It is, therefore, the uncertainty in yeast production which is most significant.
Looking back on 15 of my own beers it looks as if TrueAlc = 0.837*BalllingAcc + 0.927%
Using this to correct calculated ABV's (but from the True extract formula) gives an average error, relative to measured ABV, of 0% (the correction came from a min mse fit) and an rms error of 0.24%
Without correction the average error (difference between calculated and measured ABVs) is 0.43% and the rms error 0.5%.
I'd expect the error from the apparent extract version of the formula to be larger.
Awesome data... I'm guessing the data set is too small to look at this, but did you notice bigger beers having more relative error? I'm just curious because, like I say, with the recent revision in the Souther Tier ABV numbers, some of the imperials changed dramatically. For example, their Imperial IPA (Unearthly) went from 11.0% to 9.5%! The more average gravity beers didn't tend to change by more than a couple tenths of a percent.
Yes, the data set really is too small. In the biggest beer I checked the Balling formula actually over-estimated the ABV whereas in nearly all other cases (with the exception of one Kölsch) it underestimates. A lot of this has to do with how I measure things. Any other home brewer or commercial brewery will have his own methods and, if he wants to get a "calibration curve" for his method he needs to get ground truth somehow. Somehow means sending a sample off somewhere to be analyzed. There are several labs that do this including Siebel and Analysis Laboratory (the latter for $41 for alcohol, color, bitterness, pH).
remilard
Well-Known Member
AJ, how do you measure alcohol at home. Via distillation? Do you own a GC? I seem to think there is a GC method but in practice it is usually done via a purpose built apparatus.
Via distillation - ASBC MOA Beer 4A. There is a GC based method (Beer 4D) and there is at least one integrated instrument package (Anton Paar) of which I am aware. You start out basic with their digital density meter (DMA 5000 M) then add on other sensors such as the beer alcohol sensor (Alcolyzer), pH sensor, color sensor, carbonation sensor and wind up with a package you put a beer bottle into and it comes out with a pretty complete report, including an original extract estimate calculated, presumably, using the Balling formula as described in earlier posts. I believe the Alcolyzer uses IR absorption to determine alcohol and I think it's just the sensor - all the brains are in the DMA 5000. Certainly any large brewery doing lots of tests would use an Alcolyzer. Distillation and GC would both be way too slow, labor intensive and expensive.
nefarious_1_
Well-Known Member
(OG - FG) * 131 = approximate %ABV.
You can buy an inexpensive alcoholmeter ( a hydrometer that measures in ABV) if the OG - FG system isn't good enough for you, although I don't quite understand why it wouldn't be. The alcoholmeter is designed to measure higher ABV liquids but the scale does go down to zero so it would at least give you a secondary check on the typical homebrewer formulas for determining ABV.
The main disadvantage of these WRT to measuring beer alcohol is that they will not work for beer itself. The alcohol must be separated into an aqueous solution by distillation.
As noted above, if one uses the separation technique of ASBC MOA Beer 4A which distills 100 mLof beer one has 100 mL of an aqueous solution of the same strength (by volume) as the beer. This is to weak for use of the Tralle hydrometer. However if one were to quadruple the sample size and collect 400 mL of distillate (if simply scaling the MOAs procedure by 4 200 mL of water would be added to the beer before commencement of distillation) and then distill that down to 100 mL one would have alcohol concentrations in the range of 16 - 20% for beers in the range 4-5%. This is high enough for a decent reading from the Tralle hydrometer which, when divided by 4, would give a fair estimate of the ABV of the beer. And you would have the residue for true extract determination as well.
As noted above, if one uses the separation technique of ASBC MOA Beer 4A which distills 100 mLof beer one has 100 mL of an aqueous solution of the same strength (by volume) as the beer. This is to weak for use of the Tralle hydrometer. However if one were to quadruple the sample size and collect 400 mL of distillate (if simply scaling the MOAs procedure by 4 200 mL of water would be added to the beer before commencement of distillation) and then distill that down to 100 mL one would have alcohol concentrations in the range of 16 - 20% for beers in the range 4-5%. This is high enough for a decent reading from the Tralle hydrometer which, when divided by 4, would give a fair estimate of the ABV of the beer. And you would have the residue for true extract determination as well.
remilard
Well-Known Member
The Balling formulae are better, only very slightly more difficult to implement in software, don't assume implicitly silly things like that beer has no residual extract etc.
So nothing wrong with OG-FG except that something much better is available. I was really happy with CRT monitors until I knew flat ones existed. Sorta like that.
RCCOLA
Well-Known Member
Am I the only one who has a hydrometer with the "potential alcohol" scale on it?
OG potential alcohol - FG potential alcohol
OG potential alcohol - FG potential alcohol
remilard
Well-Known Member
As per above, it's not correct if there is any residual extract and it is also making assumptions about how much of the extract is used to make alcohol, yeast mass, and c02. If there is residual extract, as in beer, you should further be making some assumption about that.
This is the science forum and the question was about how to determine how much alcohol there is in a beer. The obvious answer is, you can't without spending a lot of time and money (or even more money and someone' else's time). The secondary answer is that the Balling formulae are the best way to estimate it.
Best answer.
Mathematically, this is exactly correct if you assume
ABV = ( OG - SG) * 131
and
76% attenuation.
Anyway, it's good enough for me. I do take hydro readings and good notes on every brew, but the quick way remilard wrote above is technically pretty good.
That scale is designed for wine but I've seen it suggested that the way to compensate for apparent extract is to call this scale a "potential alcohol" scale and to estimate your beer's alcohol by subtracting the "potential alcohol" reading at the completion of fermentation from the original "potential alcohol". That accounts for the residual extract to some extent but I doubt it improves the accuracy very much.
remilard
Well-Known Member
I guess that is true but it is still crude enough that if you can't do it in your head (I can but I have a math degree most people I know can't figure out what 20% of a dinner bill is without a small computer), you might as well be implementing the Balling formulae in software.
When someone asks me how how much alcohol a beer has, I just do 100 x (OG - 1). That tells them what they want to know (is it more like 4, 6, 8, 10).
If the State of Texas comes a knockin', I'm not using a hydrometer. Personally I don't care about accuracy in between but if I did, why not go with the best empirical method instead of an obviously far cruder one?
