Clustering 115 Hop Varieties: Part 3

An Alternative Approach & Looking at Bittering and Flavor/Aroma Separately

TL;DR: I used a slightly different methodology to group clusters of hops. I adjusted each essential oil to be expressed as a measure of specific volume, instead of as a percentage of total oil and re-ran the distance and clustering algorithm. I also re-ran based on bittering and only the flavor/aroma components separately, so one can see 2 more groupings of hops to look at based on how you may be using hops in your beer. Scroll throughout for visualization and groupings, scroll to the bottom for superlatives and a distance metric download.

If you haven’t read, try: Part 1, Part 2.

Clustering based on specific volume of essential oils (no variables removed):

I have rethought my methodology for grouping the hops. The biggest issue I see with the previous iteration is the fact that each essential oil is presented as a percentage of total oil. The issue being that if hop A & B have the exact same percentage of an essential oil based on total oil, this does not correlate to how much of that oil will be added to a beer if used in an ounce-per-ounce basis.

Example: Hop A has a myrcene content of 30% (of total oil), Hop B has a myrcene content of 30% (of total oil). One might assume that on an ounce-per-ounce basis they have the exact same amount of myrcene, but when you take into account the total oil expressed in each hop (1.2 mL/100g for Hop A & 0.8 mL/100g for Hop B) and the fact that myrcene percentage is based on the total oil, these are different weight-per-weight basis. Hop A actually has a 0.36 mL/100g contribution, and Hop B has a 0.24 mL/100g contribution, a difference of 0.12%, which could be a substantial difference in a finished beer. Therefore, I decided to present each essential oil as a percentage of weight, which was derived by multiplying the essential oil percentage by the amount of total oil in each hop, giving me the amount in mL/100g.

Re-running the distance and clustering algorithm gave me these results. It best answers, which hops should be grouped together irrespective of how they are used in the beer making process? It could be used to see which hops to trial next based on absolute similarity (both bittering contribution, bittering characteristic, and flavor/aroma).

Table 1:

HopClustersSVLOW

(high-res link)

Groupings:

Group 1 Group 2 Group 3
Brewers Gold (US) Aramis Admiral
Centennial Challenger Ahtanum
Chinook Crystal Bitter Gold
Citra HBC394 Glacier Bramling Cross
HBC 682 Hallertau Mittlefruh
Brewers Gold (DE)
Mosaic HBC369 HBC 431 Bullion
Simcoe Helga Celeia
Warrior Horizon Cluster
Kohatu Comet
Liberty Newport
Mt. Hood
Northern Brewer (US)
Mt. Rainer
Northern Brewer (DE)
Opal Perle (US)
Pacifica Perle (DE)
Palisade Phoenix
Premiant Pilgrim
Saphir Pioneer
Summer Target
Tahoma Ultra
Triskel
Whitbread Golding
Vanguard
Group 4 Group 5 Group 6
Bravo Amarillo Bobek
Columbus Cascade Loral HBC291
Galaxy Santiam
Mandarina Bavaria
HBC 438 Sterling Rakau
Tomahawk Sylva Super Pride
Zeus Wai-iti Waimea
Group 7 Group 8 Group 9
Aurora Blanc
Ekuanot HBC366
East Kent Golding Chelan Ella
First Gold Dr. Rudi Herkules
Fuggle Galena Magnum (US)
Golding (US) Green Bullet Magnum (DE)
Golding (UK) Huell Melon Merkur
Hallertau (US) Nelson Sauvin Millenium
Hersbrucker Nugget (DE) Northdown
Motueka Pacific Gem Nugget (US)
Progress Pacific Jade Polaris
Saaz (CZ) Pride of Ringwood Sorachi Ace
Savinjski Golding Riwaka Summit
Select Southern Cross
Smaragd Wakatu
Sovereign
Spalt
Strisselspalt
Sussex
Tettnang (US)
Tettnang (DE)
Tradition
Willamette

Clustering based on bittering only variables: AA (alpha acid), BA (beta acid), & COH (cohumulone)

Next, using the same information from above, I decided to remove essential oils with a high volatility (i.e. are driven off easily in the boil) and keep variables associated with bittering levels and bittering characteristics. Total oil is removed as AA, BA, and COH are already measures of weight, not total oil. Total oil is also removed because it is comprised mostly of essential oils which are driven off during the boil, it may not help me describe any aspect of bitterness, and not better than AA, BA, & COH. Alpha acid is used because of the iso-alpha acids contribution to IBUs, and the fact that it is most likely the single most important hop measure understood by brewers. Beta acids are used because of their ability to impart bitterness when not volatilized; it may stabilize IBUs as iso-alpha acids diminish during storage of unused hops as well as in finished beer (oxidized beta acids provide bitterness).

I wanted to answer the question: What hop should I substitute or trial when wanting the same bittering characteristics of beer, (e.g. bittering levels, bittering harshness, and bitter “shelf life”)?

Table 2:

BitteringLOW

(high-res link)

Groupings:

Group 1 Group 2 Group 3
Aramis Admiral Bravo
Challenger Brewers Gold (US) Columbus
Hallertau (US) Brewers Gold (DE)
Ekuanot HBC366
Hallertau Mittlefruh Bullion HBC 682
Hersbrucker Cluster Herkules
Kohatu Comet Millenium
Liberty Pilgrim Polaris
Mt. Hood Pioneer Summit
Opal Sussex Tomahawk
Premiant Target Warrior
Saaz (CZ) Whitbread Golding Zeus
Saphir
Select
Smaragd
Spalt
Sterling
Strisselspalt
Summer
Sylva
Tettnang (DE)
Triskel
Wai-iti
Group 4 Group 5 Group 6
Amarillo Bitter Gold Magnum (US)
Crystal Chelan Magnum (DE)
Glacier Dr. Rudi Nelson Sauvin
Helga Ella Pacific Jade
Horizon Galaxy Sorachi Ace
Huell Melon Galena Super Pride
Mt. Rainer Green Bullet Waimea
Pacifica Newport
Palisade Pacific Gem
Santiam
Tahoma
Vanguard
Wakatu
Group 7 Group 8 Group 9
Aurora Bramling Cross Ahtanum
Blanc Celeia Bobek
Centennial East Kent Golding Cascade
Chinook First Gold
Mandarina Bavaria
Citra HBC394 Fuggle Motueka
HBC 431 Golding (US) Northdown
HBC 438 Golding (UK)
Northern Brewer (US)
Loral HBC291 Perle (US)
Northern Brewer (DE)
Merkur Perle (DE) Phoenix
Mosaic HBC369 Progress
Pride of Ringwood
Nugget (US) Savinjski Golding Riwaka
Nugget (DE) Sovereign Tradition
Rakau Tettnang (US) Ultra
Simcoe Willamette
Southern Cross

Clustering based on flavor/aroma only variables (B-Pinene, Farnsene, Humulene, Myrcene, etc.)

This next grouping is done the same as above, but after removing bittering variables (AA, BA, COH) and total oil. As all essential oils have already been described as a specific volume per weight of hops, I removed total oil as well. Total oil does not tell me anything about the character of the hop. It may help describe “pungency”, but the levels of essential oils present already take that into account with the new measure of specific volume of each oil.

Since each of these have a different boiling point, and are driven off at varying temperature ranges, I will assume flavor and aroma is best preserved after flameout additions, especially while cooling wort quickly. Basically, I want to say: depending on the temperature and brewing process, similarities in flavor and aroma may vary. Also, don’t forget these measured essential oils represent a small number of components that contribute to hop flavor and aroma in a finished beer (see my links section for good reads on aromatic compounds in hops).

Below shows the groupings based on essential oils alone. It answers the question: What hops should I substitute or trial when wanting the same flavor/aroma characteristics of beer, if using the same amount ounce-per-ounce, (e.g. when using as a flameout, whirlpool, or dry hop)?

Table 3:

FlavaromaLOW

(high-res link)

Groupings:

Group 1 Group 2 Group 3
Amarillo Columbus Bravo
Aurora Ekuanot HBC366
Brewers Gold (US)
Blanc Ella Centennial
Cascade Galaxy HBC 438
Horizon Merkur HBC 682
Motueka Tomahawk
Saaz (CZ) Zeus
Select
Southern Cross
Spalt
Tettnang (DE)
Wakatu
Willamette
Group 4 Group 5 Group 6
Bitter Gold Polaris Bobek
Bullion Chelan
Celeia Loral HBC291
Chinook
Mandarina Bavaria
Citra HBC394 Pioneer
Comet
Pride of Ringwood
Crystal Rakau
Galena Riwaka
Magnum (US) Super Pride
Millenium Waimea
Mosaic HBC369
Mt. Rainer
Newport
Nugget (US)
Palisade
Simcoe
Sorachi Ace
Summit
Tahoma
Triskel
Ultra
Warrior
Group 7 Group 8 Group 9
Admiral Dr. Rudi Santiam
Ahtanum Hallertau Mittlefruh Sterling
Aramis HBC 431 Sylva
Bramling Cross Herkules Wai-iti
Brewers Gold (DE) Liberty
Challenger Magnum (DE)
Cluster Mt. Hood
East Kent Golding Nelson Sauvin
First Gold Northdown
Fuggle
Northern Brewer (US)
Glacier
Northern Brewer (DE)
Golding (US) Pacific Gem
Golding (UK) Pacific Jade
Green Bullet Pacifica
Hallertau (US) Perle (US)
Helga Perle (DE)
Hersbrucker Phoenix
Huell Melon Premiant
Kohatu Summer
Nugget (DE)
Opal
Pilgrim
Progress
Saphir
Savinjski Golding
Smaragd
Sovereign
Strisselspalt
Sussex
Target
Tettnang (US)
Tradition
Vanguard
Whitbread Golding

Discussion:

So for all variables, using specific volume, there is quite a difference in clustering. On the surface, I see the following points of interest:

  • Polaris looks like an outlier much like HBC 682 was previously, and looking at them side by side, it is easy to see that Polaris is most unlike any other hop due to its huge total oil content (4-5mL/100g), which if used on an ounce per ounce basis for any aroma addition, may have a profound effect on a finished beer. I have never used it though and reading descriptions online make me think it is quite pungent.
  • Group 7 has a preponderance of noble, and noble-like hops.
  • Group 8 has a large percentage of hops from Oceania.
  • CTZ cluster as they should, regardless of small differences.

Looking at the Bittering groupings:

  • Considering it is only looking at three variables (AA, BA, & COH), this is probably the most straightforward way at looking at similar hops used for bittering alone. However, considering there is no literature explaining all components of hops and how they contribute to bitterness based on usage, your mileage may vary.
  • I haven’t evaluated enough beer to determine what makes sense in this. I like Magnum and Simcoe for their “clean” bitterness, but can’t necessarily tell you what hops I avoid. I just haven’t trialed enough hops with a strict objective focus, silly me just prefers to enjoy beer.

Looking at Flavor & Aroma only groupings:

  • Groups 3, 5, and 9 have small groups, and Polaris has her own.
  • Group 7 has a tighter density, suggesting these are more similar to each other than a Group 4, which is more spread out.
  • By cherrypicking, Amarillo and Cascade are closest to each other, which I wholeheartedly agree with. However, I don’t have enough experience with other hops to evaluate this well. It may be easy to choose a hop that is pretty expressive, like Citra, and ask does it’s sibling Mt. Rainer smell anything like Citra? Reading the descriptions tells me, no, absolutely not. I think these essential oils are just not the best way of describing flavor or aroma, and considering there are over 400 compounds in hops that , your mileage may vary greatly.
  • I need more variables to describe flavor and aroma such as citranellol, limolene, nerol, 3-mercaptohexanol, 3 mercaptoheyl acetate, 4-mercapto-4-methyl-pentan-2-one, etc.

Conclusion:

So, I am going to reconsider the hops I am going to use in my 4 SMaSH beers, using most distant, close sibling, and “most average” hop with the distances calculated using all variables. Haven’t purchased any hops yet so no issue here, I still have to get rid of tons of hops I have from 2016/2017 and am making a 100% whirlpool hopped brut IPA, using all in one go. Need to meditate on this and get feedback before taking the next step.

Superlatives, using all variables:

Most different: Polaris & Sussex

Closest sibling to one of the above: East Kent Golding (closest to Sussex)

Most average: Challenger, AGAIN.

Superlatives, for bittering:

Most different: Brewers Gold & Glacier (most likely due to their very different cohumulone content and the fact that the AA% is very different as well)

Closest sibling to one of the above: Tahoma (closest to Glacier, but not by much over Brewers Gold and Admiral looking at the distance metric)

Most Average: Northdown

Superlatives, for flavor and aroma:

Most different hops: Polaris & Sylva

Closest sibling to one of the above: Sterling (closest to Sylva)

Most Average: Challenger, AGAIN.

Appendix:

Download distance measures as used in this write-up here (xlsx). How to use: Lets say you want to find the closest hop to Mosaic (on the specific volume tab). Scroll right until you see Mosaic in the top row (header). Then, sort A-Z using the filter option on the Mosaic cell. Scroll back left all the way. The closest hop to Mosaic is Mosaic, so it will be listed in the first column at the top with a distance of 0, but the next hop down will be the closest to Mosaic, i.e. Citra. Feel free to do this with the bittering and flavor/aroma sheets as well. The number will show you just how “different” the hop is to the one you selected, so you can compare them numerically.

That is all for now, please let me know what you think in the comments!

2 thoughts on “Clustering 115 Hop Varieties: Part 3

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s