WHAT IS SENSORY ANALYSIS?

The use of people as instruments to

measure sensory response to stimuli.
Flavor defined by Morton Meilgaard

“the term used to describe the complex

interaction of taste, smell, and chemical
irritation of foods in the mouth that add
to its mouth-feel, such as carbonation, the
burn of chili peppers, or the coolness of
menthol”
Flavor is made up of:

Basic Tastes

Aromatics
• Olfactory stimulation
• Trigeminal nerve stim

Mouthfeels
• Chemical or physical sensations
• Astringent, dry/tannin, bite, burn, etc
What can we taste?
 Sweet
 Sour
 Salty
 Bitter
 Umami
 response to salts of glutamic
acids like MSG. Processed
meats, cheeses, and soy sauce
also contain glutamate

The “receptor map” is dead!

Hanig 1901, turned over by Collings, 1974
Can you smell sweet?
 Perception of flavor is a
combination of sensory stimuli
 Basic Tastes 5%
 Smell is 90%
 Touch 1% (trigem)
 Balance is vision, hearing, etc

No, you can’t!

But, Aroma can be an
Indicator of flavor
Trigeminal & Tactile Sensation
 Carbonation
 Related to level, partial pressure
 Pain
 Can increase perception of bitterness

 Fullness
 Function of residual solids
 Dextrins and Oligosaccharides (complex sugars)

 After feel
 Astringency
 Pain (chili peppers)
Aroma:
 The human nose is very sensitive
More sensitive than a gas
chromatograph!

 Perception changes based on

situation

 Smell with sniffs in short burst for

1-2 seconds. After this receptors
become saturated and can require
20 seconds +/- to reset

 Don’t forget to exhale

The Flavor Unit
 Concept introduced by Dr. Morton Meilgaard
 Recognized 100’s of compounds in beer
 There are different thresholds for chemicals we can taste.
 Example:
 Ethanol 50 grams / Liter (50,000+ ppm)
 S-Methyl Mercaptan 4 nanograms per Liter!!
10-9 grams per Liter
(.0000001 ppm!)
 Need a way to compare
 One flavor unit = the starting threshold that a compound can be detected by
the olfactory
 Ethanol at 50 grams/Liter = 1 flavor unit
 S-Methyl Mercaptan at 8 nanograms per Liter = 2 flavor units

Allows apples to apples comparison

BASIC TRAINING
What it comes down to is everyone who
touches your beer should know how to talk
about your beer!
WHO?
 Beer school for all new hires
 Brewer / Cellar / Lab
 Packaging
 Sales / Marketing
 Distributors
 Festivals / Tours / Promotions
HOW and WHEN?
 Schedule Panels as Meetings
 Keep Consistent - same time and location
 Goal is Regular Attendance

 Work it in so people look forward to it

 End of Meetings
 Shift Changes
 Break in the Day
BEER SCHOOL
WHAT MAKES YOUR BEERS SPECIAL

 What is in it?
 Appearance / Aroma
 Taste / Flavor Characteristics / Aftertaste
 Style / Alcohol % / IBU
 Availability
 Food Pairings
YOUR BEER?

 Know the aroma and flavor characteristics

of all your beers
 MALTS

 HOPS

 YEASTS

 SPICES

Whatever is in it have available for

reference
Vocabulary & Recognition
 Use reference standards
 i.e. spikes, aroma vials, malt, hops
 Repetition is the key to learning
 Don’t Rush It!
 Build vocabulary slowly to take pressure off
 Spikes
 These are your Lego Blocks!
Teaching What and Why
 Common off aroma / flavor checks
 Diacetyl Rests
 Aroma and tastings through-out
the brewing processes
 From Wort to Package
 Technical Feedback
 True to Brand / Consistency
 BUILDING BLOCKS
 No Pressure Panels
 Build up your own Attribute Library
 Everyone Participates / Good Motivator
 Lower threshold levels
 Different Styles
 Oxidation Stages
 Brand Recognition
Flavor Wheel
SETTING UP FOR SUCCESS
Teach the Basics and Share as Much as Possible

APPEARANCE (Clarity and Chill Haze)

AROMA
SERVE AT AN APPROPRIATE DRINKING TEMPERATURE
TASTE (Don’t DRINK!)
Scoring System
AFTERTASTE
We don’t spit beer!
BLIND TASTING (Fair and Good Practice)
LIMIT THE NUMBER OF BEERS TO AVOID TASTEBUD BURNOUT
PALATE CLEANSERS (Plain Crackers and Water)
LEARN / TEACH TO NOT FINISH THE WHOLE SAMPLE
MOTIVATION
 Change things up / variety
 Different types of Panels
 Aroma Vials, Spikes, Triangle Tests,
Preference
 Everyone learns differently
 Positive Rewards
 Feedback and Guidance
 SERIOUS BUSINESS BUT FUN!
TRAINING, VALIDATION AND
FEEDBACK OF A
PRODUCTION/MARKET
RELEASE PANEL

Cathy Haddock
Sensory Specialist Quality Assurance Dept. Sierra Nevada Brewing Co.
CBC Annual Meeting, 2011
Why is it important to have a trained and
validated panel?

 A trained sensory panel is a valuable instrument;

each taster being a unique tool in the toolbox.
• Can not rely on just 1 opinion. Everyone has their own
sensitivities.

 Example: Brew master who is blind to diacetyl is

the sole taster for release of product to market. Not
good!
What is a Production Release Panel?

Production Release Panel-a trained

panel that evaluates product to be
release to market.
 Panel can use various quality control
tests such as a go-no go, in/out
yes/no, pass/fail, quality ratings, etc.
format.
Training Process-
Production Release Panel

 If you are on the Production Release

Panel you must be trained and validated
on off flavor recognition and brand
attribute!!
What to train with??

Train using spiking compounds.

 Flavor Activ…quick, easy, good shelf life, but

costly, cost varies per compound.

 Seibel Training Kit….quick, easy, shorter shelf life

(2 months-refrigerated), 25 vials for $180.

 Sigma Aldrich Flavor and Fragrances Kit-some

dilution prep work. 22 vials for $473. No DMS or

Acetaldehyde. Good shelf life.
 Next Steps
 Identify what key flavors and off flavors are
important and train on those compounds.
 Recommend spiking and training with a flagship
brand, as well as any other brands if time and
expense is available.
A spiking in one brand, can come across different in another
brand.
Types of Training and Test Methods

Types of Training/Validation Methods:

 Off Flavor Recognition Testing-present panelist with spiked
beer samples for evaluation and identification. Recommend 6
samples in a session.
 Brand Attribute-training with pantry reference standards, one-
on-one and group exercises.
 Sensitivity Threshold Testing-provides training on how
compounds are perceived at various concentrations as well as
gain insight on panelist sensitivities.
Tasters report from test session-taste
panelists are known by a 4 digit code

Flavor Activ Friday-12-17-10

100.0% 100.0% 100.0% 100.0%
100.0%
Panel
90.0%
average=85.0% .
83.3% 83.3% 83.3%
80.0%

70.0% 66.7% 66.7% 66.7%

60.0%

50.0%

40.0%

30.0%

20.0%

10.0%

0.0%
Off Flavor Recognition- Spreadsheet

2009/
8/13/2010 8/27/2010 10/15/2010 10/22/2010 11/5/2010 11/19/2010 12/3/2010 12/17/2010 2010 avg 2009 avg 2010avg
62.6% 72.6% acetaldehyde 39.5% 55.8%
75.0% 81.0% acetic acid 91.8% 84.0%
100.00% 97.2% alkaline 76.9% 69.2%
75.0% 68.1% butyric acid 59.7% 63.7%
53.8% capryllic acid 80.4% #DIV/0!
87.50% 90.9% 80.0% catty 91.7% 76.5%
87.50% 88.9% 90.0% 77.8% chlorophenol 82.5% 80.9%
87.50% 100.0% 86.3% diacetyl 97.2% 97.5%
62.50% 75.4% dms 78.1% 59.2%
87.50% 100.0% 95.8% earthy 81.7% 80.4%
62.50% 54.3% ethyl butyrate 72.6% 68.0%
87.50% 90.0% 91.1% ethyl hexanoate 86.7% 83.4%
45.2% freshly cut grass 62.5% 8.2%
85.70% 63.3% geraniol 72.3% 74.5%
75.00% 44.4% 48.6% grainy 57.0% 42.2%
50.00% 45.5% 59.4% indole 58.6% 30.9%
87.50% 77.8% 76.9% isoamyl acetate 82.7% 81.3%
62.50% 81.8% 79.2% isovaleric acid 68.8% 74.3%
37.50% 44.4% 70.0% 49.1% leathery 59.7% 50.9%
100.00% 100.0% 100.0% lightstruck 94.4% 97.2%
37.50% 18.2% 37.5% mercaptan 43.0% 32.3%
75.00% 90.9% 89.6% metallic 82.2% 88.1%
37.50% 80.0% 53.3% musty 69.5% 62.9%
25.00% 37.5% 39.2% onion 36.1% 35.9%
62.50% 87.5% 80.0% 85.2% papery 86.9% 87.4%
62.6% 77.4% phenolic 61.4% 37.3%
37.50% 38.6% plastic 23.5% 33.5%
50.00% 54.5% 63.9% sour 64.2% 46.5%
50.00% 63.5% lactic acid 66.6% 66.6%
50.00% 68.7% ethyl acetate 65.0% 65.0%
92.9% methional 81.7% 81.7%
50.0% sulphitic 62.5% 58.4%
62.50% 68.75% 62.50% 70.53% 67.52% 75.92% 63.63% 85.00% 70.3% 64.2% 64.8%

easy 75-100%
mod 56-75 %
hard 0-55%
Charts allow visualization of taste
panel recognition abilities.

The beers that are bottle conditioned at 62 deg. F, whether normal or elevated yeast pitch, measured having higher
levels of diacetyl.
Brand Attribute Training

 Determine what are the key attributes in your

brand and associated descriptors.
 Find Reference Standards to demonstrate those
attributes.
Example: Fruit note in Sierra Nevada
Pale Ale is our yeast; make yeast cake and
serve side by side with the beer.
 As a Panel, discuss changes in those attributes
in different ages of beer.
Threshold Testing-
ASBC Method of Ascending Limits
• 6 triangle tests
• Go from lowest to highest concentration with threshold
in the middle.
• Test samples increase in concentration, usually 2-fold.
• Get Best Estimate Threshold of group; geometric
mean of the highest concentration missed and next
higher (adjacent) concentration.
• Get an individual and panel BET average.
• Take into account amount of compound in beer matrix.
 THRESHOLD OF ISOAMYL ACETATE
Procedure: ASBC Sensory Analysis - 9 (Ascending Method of Limits) Number of Assessors: 26
Medium: Miller High Life Dilution Factor: 2
Scale Steps: 6 Sample: Isoamyl Acetate Aldrich Cat. # 112674 98% Form of Test: Simple / Aroma Only

rving Concentration, ppm Individual Threshold

Name Description
rder BET, ppm Log 10
0.375 0.75 1.5 3 6 12
1 taster 1 N N Y Y Y Y 1.06 0.026
2 taster 2 Y Y Y Y Y Y 0.27 -0.576
3 taster 3 Y Y Y Y Y Y 0.27 -0.576
4 taster 4 N Y N Y Y Y 2.12 0.327
5 taster 5 N N N N Y Y 4.24 0.628 fruity, banana
6 taster 6 N N Y Y Y Y 1.06 0.026 banana
7 taster 7 Y N Y Y Y Y 1.06 0.026
8 taster 8 N N N Y Y Y 2.12 0.327
9 taster 9 N N N Y Y Y 2.12 0.327
10 taster 10 Y N N Y Y Y 2.12 0.327
11 taster 11 N N Y Y Y Y 1.06 0.026
12 taster 12 N N N Y Y Y 2.12 0.327
13 taster 13 Y Y Y Y Y Y 0.27 -0.576
14 taster 14 Y Y Y Y Y Y 0.27 -0.576
15 taster 15 N Y Y Y Y Y 0.53 -0.275
16 taster 16 N N Y Y Y Y 1.06 0.026
17 taster 17 N Y N Y Y Y 2.12 0.327
18 taster 18 N N Y Y Y Y 1.06 0.026 Isoamyl acetate
19 taster 19 N Y N N Y Y 4.24 0.628 fruity, banana
20 taster 20 N N N Y Y Y 2.12 0.327 fruity, pear drop
21 taster 21 N N N Y Y Y 2.12 0.327
22 taster 22 N Y Y Y Y Y 0.53 -0.275
23 taster 23 N N N N Y Y 4.24 0.628
24 taster 24 N Y Y Y Y Y 0.53 -0.275
25 taster 25 N N Y Y Y Y 1.06 0.026
26 taster 26 Y Y Y Y Statistical
Y Y Summary:
0.27 -0.576

Data collected at 2005 ASBC Group Best Estimate Threshold (BET), ppm = 1.09
workshop by Everett Boiling Sum Log 10 = 1.114
Average Log 10 = 0.056
Graph of Individual Tasters BET
More than Performance Feedback,
…say Thank You!!

It takes a time and commitment to be on panel. A

lot of training is involved.
Say “Thank you” with a ………
 Post-sensory treat/snack

 Rewards Program-gift certificates, prizes, etc.

(Frequent Tasters Reward Program)
 End of the year parties

 Positive Feedback-be a cheer leader

 Free Beer!!
 SENSORY STATISTICS:
TRANSFORM YOUR SUBJECTIVE
TASTE PANEL INTO AN
OBJECTIVE MEASUREMENT.
Beer’s job is to deliver 10 minutes of pleasure…
So, we need to:
 Understand your brand’s flavor profile: TTB (trueness to
brand). Know the difference between normal process
variation, process trends and/or anomalies. Consistency!
Identify and classify flavors and determine their desired
levels for a given brand. Then, pinpoint their origins from
grain to glass/processing.
 Provide specific- technical- actionable flavor interpretations
to help production correct flaws, identify trends
and better control the outcome of each brew.
 Keep it simple, thorough repeatable,

reproducible test method and data analysis.

 What do I do with my
production release data results?
 How bad is ‘too bad’? Sellable vs. non-sellable finished
product.
 What is normal/natural process variation?
 What is out of specification? What warrants investigation?
 How can you avoid flavor shift over time?
 Is the product defect enough to cause the consumer to
notice? Complain? Recall?!?
Need to Move away from grading. 1 out of 10 panelist (no go)
≠ 90% A. What about 3 of 20? 85% B- 3 diacetyl
comments?!?!
Move away from tests that generate limiting, unactionable,
un’mine’able data – scaling, ttt. Final score = 16- 76%
(P)robability control charts-
sleep well through the power of simple statistics!
The perfect tool for Sensory Finished Product Release Panel Analysis:
 A p-chart (probability chart) is an attributes (ttb or not ttb)
control chart that consists of points collected and plotted with the
control/natural process limits from data in subgroups of varying
sizes (different number of and panelists every panel).
 Think of the limits as the ‘voice’ of your process. P-charts monitors
normal variations: whether your process is stable and predictable
and determines whether a particular sample falls within the
normal variation or falls outside and needs to be examined
further. Can also monitor the effects of process improvement
theories (test brew validation) or spikes (panel validation).
 P-charts can easily be created using simple
excel add-ins that are pretty cheap and easy!
Production Release Ballot and Data Entry

Sample 1 2 3 4 5 6 7 8 9  Check it out-

V isual
A aroma
probability
F lavor charts in
M outhfeel
O verall
action….
Looks like your
normal data
COMMENTS

entry
spreadsheet.
Normal process variation.

p-Chart

0.450

0.400
UCL
0.350

0.300
Proportion

0.250

0.200

0.150

0.100

0.050 LCL

0.000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Observations (Sample Number)

Proportion p-bar UCL LCL

Out of Control data patterns- alert!
Your process is changing behavior!!
Is your panel ‘plugged in’? Validation via spiked sample.

aroma

taste/aftertaste

mouthfeel/body

overall flavor
clarity trend
Decision tree- FYI/Alert- trend.
Investigate, report, document, follow up.
overall flavor out of control-not true to brand.

diacetyl comments
Each ‘out of spec’ scenario needs to flow like this
… Closed Loop Corrective Action Plan!

 Identify Problem/Attribute (diacetyl

comments and red OOC in aroma and
overall)
 Assignable Cause (slow fermentation,
t-down too early)
 Corrective Action (flag tank, measure
diacetyl before t-down)
 Hold the gains (change SOGs, train on
new procedure)
 Taste Panel vs. Sensory Program
1. ID process, method, measurement owners
2. Initial Measurement Validation (MSA)
3. Implement and maintain applicable SOG’s
4. Training Protocol & Records
5. Establish appropriate charts (Pcharts)
6. On going Validation Method
7. Calibration
8. OOC/OOS Action Plan Development/Decision Trees
9. First Pass Analysis- using OOC/OOS Action Plan
10. Document First Pass Analysis in control chart
11. Cleaning the data set- Brewmaster Ready
12. Trends/Pareto Analysis and Recommendations
13. Present in Quality Meeting
14. Transition to Project/Investigation ideation
 P CHART ADD-IN WEBSITES
•http://www.isixsigma.com/library/content/n070223a.asp
•http://www.qimacros.com/qiwizard/pchart.html

•http://www.pqsystems.com/products/SPC/SQCpack

•www.XLSTAT.com

•All materials are sole property of New Belgium

Brewing Company and should not be copied,
reproduced or used without permission.