Estimating bitterness: Algorithms and state of the art - Page 3 - Home Brew Forums
Register Now For Free!

Home Brew Forums > Home Brewing Beer > Brew Science > Estimating bitterness: Algorithms and state of the art

Reply
 
Thread Tools
Old 03-24-2009, 05:05 PM   #21
olllllo
[]-O-[]
HBT_LIFETIMESUPPORTER.png
 
olllllo's Avatar
Recipes 
 
Apr 2006
Phoenix, Arizona
Posts: 13,330
Liked 140 Times on 114 Posts


I'm going to add this first hand account of a commercial brewery attempting to get lab tested IBU levels vs their calculations.

http://www.deschutesbrewery.com/blog...nd-statistics/

Quote:
But 95 IBU’s turned out to be so much wishing. Our first brewing of Hop Henge this year produced the following result. The very vigorous ferment, with a fermenter at capacity, blew our precious dry hops all over the floor depriving us of all the goodness therein. The result was a beer we calculated to, on paper, 243 IBUs! In the bottle, we only got 80 IBUs. Still, as I mentioned earlier, you apparently loved it. So, what did we do? We made another batch, added more hops, only filled the fermenter half-full and thought we would blow the doors off the beer (and your taste buds). The original calc’s on batch #2 were the same as batch #1, but without the blow-off on the floor we ended up with a massive 117 IBUs in the fermenter, as determined in our lab this time. We were excited at the possibilities and fastened our seat belts for the ride. The beer has now been centrifuged and removed from the dry hops (which then took our guys and gals 4 hours to remove from the fermenter!) and we sent it back to the lab for analysis looking to tell you of our herculean feat. Meanwhile we tasted the two batches side by side today and were impressed by the huge pucker-factor in the new Henge, much greater than batch #1. So, we sent it off to the lab to await the results. 87 IBUs! We still did not even hit our promised 95 IBUs.
__________________
Rabbit And Coyote Schwag
Rob - Phoenix Ambassador to Milwaukee
Where did your avatar go?
Ginger Beer for Moscow Mules Bacon Vodka
Twitter

 
Reply With Quote
Old 03-24-2009, 05:28 PM   #22
pjj2ba
Look under the recliner
HBT_LIFETIMESUPPORTER.png
 
pjj2ba's Avatar
Recipes 
 
Jul 2006
State College, Pennsylvania
Posts: 3,379
Liked 206 Times on 167 Posts


Quote:
Originally Posted by jkarp View Post
All were done with 30 minute variable and 10 minute fixed hop additions; the 30 minute quantity being adjusted to meet 40 IBUs, of course. My reasoning was, 10 and 30 min are the crossover points of aroma/flavor and flavor/bittering utilizations. That way hopefully some of each characteristic would be represented in the final beers.
This makes sense then. I think the more you boil the hops, the more you loose the individual characteristics. For my lager series, I'm looking for differences inflavor and aroma, with the bittering being held constant (hopefully).
__________________
On Tap: Doppelbock O'fest, Pale Ale, cider
Kegged and Aging/Lagering: CAP, Ger. Pils, OKZ (std Amer. lager), CZ Pils, Amer. Wheat, Rye IPA, Saison
Secondary:
Primary: Ger Pils, CAP
Brewing soon: Pale lager, Amer. wheat
Recently kicked : (
Pilsner Urquell Master Homebrewer
(1st NYC 2011, 2nd NYC 2012)
P U crowns winners in its inaugural master HB competition

 
Reply With Quote
Old 03-24-2009, 05:54 PM   #23
jkarp
 
jkarp's Avatar
Recipes 
 
Jun 2008
Elizabeth, CO
Posts: 2,112
Liked 45 Times on 35 Posts


It's fascinating stuff to play with, that's for sure. Looking forward to hearing your results pjj2ba.

 
Reply With Quote
Old 03-25-2009, 01:36 AM   #24
ssabin
Recipes 
 
Dec 2008
New Hudson, MI
Posts: 56

Well, it's obvious that there is a lot going on and that accurately predicting IBUs is difficult at best at this point in time. Still, I was curious as to which IBU predictor model was "best".

It would be nice to simply cut and paste the data I used behind these summaries, but that's not so easy to do here and keep the formatting (column alignment).

So to summarize, my take on the data:
1) there are a lot of holes in understanding how the data was obtained, what assumptions were made, what process was followed, etc.
2) there is a definite pecking order of algorithms in terms of accuracy at predicting IBUs in the final brew.
3) The top 3 algorithms are likely in a similar range of acceptability, while the worst seems to have a flawed model.

The results:

1. Mosher
Chi-squared statistic (lower is better, and is equal to sum of (Oi-Ai)^2/Ai where Oi is observation i, and Ai is actual value i): 11.8
Std Dev of Chi-square data (lower is better): 2.53
Average Error value (closer to zero is better): 4.33 IBU's (over-predicted)
Std Dev of errors (lower is better): 4.32 IBUs
Worst error (lower is better): 8 IBUs (over-predicted)

2. Tinseth
Chi-squared: 17.97
Std Dev of Chi-sq data: 3.64
Avg Error: 4.67 IBUs (over-predicted)
Std Dev of errors: 6.41 IBUs
Worst error: 14 IBUs (over-predicted)

3. Rager
Chisquared statistic: 25.36
Std Dev of Chi-squared data: 5.79
Avg Error: 7.00 IBUs (over-predicted)
Std Dev of errors: 5.73 IBUs
Worst error: 14 IBUs (over-predicted)

4. Garetz
Chisquared statistic: 30.8
Std Dev of chisquared data: 3.49
Avg Error: -9.17 IBUs (under-predicted)
Std Dev of Errors: 11.27 IBUs
Worst Error: 20 IBUs (under-predicted)

5. Daniels
Chisquared statistic: 114.43
Std Dev of chisquared data: 16.14
Avg Error: 22.33 IBUs (over-predicted)
Std Dev of errors: 15.68
Worst Error: 40 IBUs (over-predicted)

I also notice that a single data point (the 6 Simcoe) sticks out in the Tinseth model, meaning there might have been some issue with the process that wasn't captured in the prediction for that particular batch. At any rate, I think that based on this limited amoutn of data, the top 3 algorithms perform well enough to be useful.


 
Reply With Quote
Old 03-25-2009, 03:46 PM   #25
SpanishCastleAle
 
SpanishCastleAle's Avatar
Recipes 
 
Jan 2009
Central Florida
Posts: 4,345
Liked 36 Times on 36 Posts


What is the Mosher algorithm?

I used Daniel's method in my homebrew spreadsheet...no wonder I rely more on 'feel' than anything wrt hops. I trust my color prediction more than my bitterness prediction.

I'm gonna try Tinseth first and see how that goes but I'd like to try all of them...but I didn't see the Mosher method.
__________________
Early brewers were primarily women, mostly because it was deemed a woman's job. Mesopotamian men, of some 3,800 years ago, were obviously complete assclowns and had yet to realize the pleasure of brewing beer.- Beer Advocate

 
Reply With Quote
Old 03-25-2009, 04:10 PM   #26
SumnerH
Recipes 
 
Feb 2009
Alexandria, VA, USA
Posts: 2,057
Liked 31 Times on 27 Posts


I haven't written up Mosher or Daniels yet. I'll probably skip Daniels since it's not particularly accurate, but I'll get around to Mosher eventually.
__________________
On deck: Little Bo Pils, Bretta Off Dead (Brett pale)
Secondary: Oude Bruin, Red Sky at Morning (Sour brown ale)
On tap: Saison Duphunk (sour), Amarillo Slim (IPA), Earl White (ginger/bergamot wit)
Bottled: Number 8 (Belgian Strong Dark Ale), Eternale (Barleywine), Ancho Villa (Ancho/pasilla/chocolate/cinnamon RIS), Oak smoked porter (1/2 maple bourbon oaked, 1/2 apple brandy oaked)

 
Reply With Quote
Old 03-25-2009, 04:50 PM   #27
SpanishCastleAle
 
SpanishCastleAle's Avatar
Recipes 
 
Jan 2009
Central Florida
Posts: 4,345
Liked 36 Times on 36 Posts


I just need the equation so I can put it in a spreadsheet. I converted the algorithms you posted earlier back to a 'regular' equation so I can use them. Either will do.

My last programming was a Fortran class in the late 80's...never really used it and can't remember much of it. So I just use a spreadsheet.
__________________
Early brewers were primarily women, mostly because it was deemed a woman's job. Mesopotamian men, of some 3,800 years ago, were obviously complete assclowns and had yet to realize the pleasure of brewing beer.- Beer Advocate

 
Reply With Quote
Old 03-25-2009, 05:02 PM   #28
Mike M
 
Mike M's Avatar
Recipes 
 
Jul 2007
, Delaware
Posts: 505
Liked 18 Times on 13 Posts


I use Tinseth in my spreadsheet. I think I found everything right here: Hop Utilization Page

 
Reply With Quote
Old 03-25-2009, 06:41 PM   #29
Denny's Evil Concoctions
Grande Megalomaniac
HBT_LIFETIMESUPPORTER.png
 
Denny's Evil Concoctions's Avatar
Recipes 
 
Oct 2005
West Kelowna BC, Canada
Posts: 7,733
Liked 67 Times on 53 Posts


I wish beersmith had more than 3 choices for calculating IBU's
__________________
I may not be an expert, but I did stay at a Holiday Inn Express........ 6 months ago.

http://www.homebrewtalk.com/f39/denn...9/#post1766281

http://groups.homebrewtalk.com/Tapro...ook_Repository

 
Reply With Quote
Old 03-27-2009, 01:06 AM   #30
ssabin
Recipes 
 
Dec 2008
New Hudson, MI
Posts: 56

You've inspired me, DEC, to see if the average, or some combination of the top 3 algorithms provides an even "better" algorithm. But I won't be able to look at this until tomorrow.

For the record, I also use Tinseth in my spreadsheet. I won't change unless I can markedly improve the quality of the prediction.

 
Reply With Quote
Reply
Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Estimating ABV? Cape Brewing Cider Forum 2 09-24-2009 01:20 PM
estimating OG newbrewr4fun Extract Brewing 4 09-12-2009 02:44 PM
Estimating AA% Yuri_Rage Hops Growing 13 05-12-2008 05:04 PM
Estimating Gravity Sea All Grain & Partial Mash Brewing 3 06-24-2007 02:51 PM
Help estimating Alcohol % rjm Beginners Beer Brewing Forum 11 04-21-2006 06:37 PM


Forum Jump