NORDIC BARISTA CUP &

SPECIALITY COFFEE ASSOCIATION OF EUROPE

GRINDING
RESEARCH
REPORT

AUTHOR – FRANCISCA LISTOV-SAABYE

GRINDING RESEARCH REPORT

INTRODUCTION

In 2013, the Speciality Coffee Association of Europe (SCAE) published the research paper
‘European Extraction Preferences in Brewed Coffee’. Our objective was to scientifically
establish European coffee taste preferences. It was the result of a dedicated programme
of field-based research conducted throughout Europe in 2011. It was the first publication
in a new research initiative and was both welcomed and valued by our members.

Continuing the association’s commitment to research based education, SCAE partnered

with the Nordic Barista Cup (NBC) in 2013 to continue this research initiative. The opportunity
to work together with the NBC and sensory scientist Francisca Listov-Saabye gave SCAE
an excellent second step in our research journey, targeting the murky subject of flat
versus conical burrs in coffee grinding.

This paper sees both the continuation of SCAE’s commitment to research based education
but also the beginning of a scientific journey which will bring our members an annual
deliverable of research based data to help our community drive excellence in speciality
coffee, which is based on fact rather than opinion.

Paul Stack

Chair SCAE Education Committee 2012-2014

June 2014

1
GRINDING RESEARCH REPORT

WHAT TO EXPECT……

Myth busting based on scientific research. A walk through of the entire grinding research
project which led up to the highlight of presenting the research at the Nordic Barista Cup
2013. An in-depth review of the challenge at SCAE in Nice where the coffee attendees did
NOT recognise the difference between the coffees. The grand finale with the disclosure
of the hard true facts from the live evaluation cupping at NBC 2013 - The attending coffee
professionals in Oslo could NOT taste the difference between flat and conical grinding burrs.

2
GRINDING RESEARCH REPORT

REPORT INDEX
BACKGROUND 4
Contact Information 4
The Project Manager 4
The NBC/SCAE Gold Cup Grinding Research 4

MATERIALS AND EQUIPMENT 4

The Coffee Lab 4
Equipment 5
Coffee 5
Water 5
Weight of Coffee 5

TABLE OF GRINDERS IN SENSORY PROFILING 6

The Sensory tests 7
Sensory Profiling 7
Training of judges 7

RESULTS 10
Triangle test – is there a difference? 10
Consumers 10
Coffee professionals 10
Conclusion triangle test Nice, France 13
The Sensory Profiling test 14
Sensory Profiling, University Panel 17
Particle Size Distribution 20
Discriminative Cupping Test at NBC 2013 22

CONCLUSION 23

FUTURE 24

THANKS TO THE RESEARCH PARTNERS AND SPONSORS 25

APPENDIX & REFERENCES 26

3
GRINDING RESEARCH REPORT

BACKGROUND
THE PROJECT MANAGER
Francisca Listov-Saabye has a master’s degree in Food Science, specialising in sensory science. She has been
gathering a broad knowledge in consumer and sensory science as well as new product development and food
production over 13 years in the food industry.

CONTACT INFORMATION
Francisca Listov-Saabye
francisca@listov-saabye.dk
+45 2634 4634
LinkedIn profile: dk.linkedin.com/in/franciscalistovsaabye/

THE NBC/SCAE GOLD CUP GRINDING

RESEARCH PROJECT
Back in 2011 during the competitions at the Nordic Barista Cup a discovery was
made: The temperature of the ground coffee seemed to increase throughout the
session. This assumption was the basis on which Francisca did a small scale trial at
the next NBC in 2012. The trial investigated the temperature increase in the ground
coffee beans during the espresso completion, where 250 espressos were produced
in 40 min. The discussions of the results lead to developing the present NBC/SCAE
Gold Cup grinding research project.

On coffee debate forums, in blogs and “through the grapevine” there was a saying,
an understanding, that there was a taste difference between a flat and a conical
burr grinder. We set out to investigate this. How do the grinder characteristics such
as burr size and type (flat or conical) influence the taste of coffee?

Hypothesis: There is a taste difference between a flat and a conical burr grinder!

MATERIALS & EQUIPMENT

THE COFFEE LAB
At Europa Tietgen, a café and canteen setting in Copenhagen, Francisca set up
her coffee lab collecting a lot of grinders and brewers.

The grinders were chosen according to variance in burr size and burr type as
well as being grinders targeted at consumer, prosumer and coffee professional
markets.

An investigation of the coffee bars in Copenhagen was conducted and additional

grinders were obtained in order to represent the reality of baristas etc.

In the coffee lab, trials were conducted looking into coffee retention, regrind,
setting of grinders and usability in general. See table of grinders.

4
GRINDING RESEARCH REPORT

EQUIPMENT
• Water and coffee was weighed using a G&G digital pocket scale, model LS2000H (sensitivity 0-1000 gram 0,05
gram, 1000-2000 gram 0,1 gram).
• The temperature of the brew as well as of the samples for %TDS measurement was measured using a
Testo Quicktemp 826-T3 thermometer or a Fluke 53-2 thermometer with a 9065 probe.
• The total dry solids (%TDS) was measured by a VST LAB Coffee II Refractometer.
• Coffee was brewed on either a Wilfa Svart Presisjon machine filter brewer or a programmable BUNN filter ICB brewer.
• Two different types of filters were used: Wilfa svart presisjon uses standard white paper filters size 4
from supermarket. These were washed with hot water prior to brewing. For the BUNN brewers, official
BUNN paper filters were used. These were used dry due to practicality and the side of the filter being vertical.

COFFEE
The coffee used throughout the research is Ethiopian TADE, sourced and roasted by Solberg & Hansen,
Norway. The roast profile is a light Nordic style roast. For each separate test during the research fresh roasted
batches were used, co-ordinating to ensure evaluation was within 5-14 days after roasting.

WATER
The brew water was filtered water. The filter BWT Bestmax 2XL was used.

At the SCAE World Coffee convention in Nice the filtered and controlled water at the fair was used.

At the Nordic Barista Cup in Oslo the clean Norwegian water was filtered by Everpure 7FC filter.

WEIGHT OF COFFEE
For the coffee retention trials, the coffee beans going into the grinder as well as the ground coffee coming out
of the grinder were weighed. For the range of grinders tested in this study the weight of coffee in hardly ever
resembles the weight of coffee out. Of any of the grinders!

5
GRINDING RESEARCH REPORT

TABLE OF GRINDERS IN SENSORY PROFILING

NUOVA LA
GRINDERS MAZZER MAHLKÖNIG FAEMA MAHLKÖNIG
SIMONELLI MARZOCCO
MODEL Mythos MD5 Robur Vario MC99 K30

APPEARANCE

BURR TYPE flat conical conical flat conical flat

BURR SIZE 75 mm 65 mm 71 mm 54 mm 68 mm 65 mm

BURR Titanium coated Stainless steel Stainless steel Ceramic Stainless steel Stainless steel
MATERIAL steel

MOTOR SPEED 1400 620* 500 1000 320 1420

(RPM)

HOPPER plastic glass plastic plastic plastic plastic

DOSER - + - - + -

RETENTION 0,7 ± 0,9 gram 0,3 ± 0,6 gram 1,4 ± 0,9 gram 0,5 ± 1,7 gram 0,6 ± 0,7 gram 0,5 ± 0,3 gram

STATIC
ELECTRICITY

SETTING OF Steeples, easy Steeples, Steeples, hard Steps, 10*22 Steeples, easy Steps, 20
GRINDER medium positions positions

*Calculated

6
GRINDING RESEARCH REPORT

THE SENSORY TESTS

Two different sensory test types has been used in the research project – discriminative and descriptive
testing respectively.

For the discriminative test, different triangle test was conducted. The respondent is to pick the odd sample out
of three cups of coffee.

For the descriptive test, two quantitative profiling trials are conducted.

SENSORY PROFILING
The main difference between classic coffee tasting, cupping, and sensory profiling is the use of a different
evaluation scale. Secondly, it’s the actual set up where cupping usually takes place in a social setting where
multiple judges evaluates 6 to 12 cups in one round whereas sensory science judges individually evaluates one
product at a time.

TRAINING OF THE JUDGES

For each of the descriptive tests, training sessions with the judges were carried out. For three sessions
of a couple of hours, the entire group was gathered around a table. The variance of the coffees in the test
were tasted blind. Based on a discussion about the coffees using physical references the judges agree upon
perceived attributes to describe the profile.

The panel leader, in this research project Francisca Listov-Saabye, brings physical references (fresh flowers,
pieces of lemon and oranges, liquorice powder, dark chocolate etc.) based on prior tasting of the coffees.

During the training sessions, the panel came up with other suggestions for the vocabulary, from which the
panel leader brought physical references for the next session (e.g. cut pipe tobacco, caramel sauce, fudge,
raisins and different types of roasted nuts). The final vocabulary representing the coffees from the different
grinders were determined and the use of the line scale was practised.

7
GRINDING RESEARCH REPORT

SMELL
HAZELNUTS NOTHING A LOT

LEMON NOTHING A LOT

CHOCOLATE NOTHING A LOT

FLORAL NOTHING A LOT

TOBACCO NOTHING A LOT

TASTE
SWEET NOTHING A LOT

ACIDIC NOTHING A LOT

BITTER NOTHING A LOT

SOUR NOTHING A LOT

TOBACCO NOTHING A LOT

LEMON NOTHING A LOT

CHOCOLATE NOTHING A LOT

LIQUORICE NOTHING A LOT

HAZELNUT NOTHING A LOT

ROASTED NOTHING A LOT

BURNT NOTHING A LOT

CARAMEL NOTHING A LOT

MOUTH FEEL NOTHING A LOT

Unmarked line scale used in the present grinder research project.

With a vertical mark, the judge scored the exact strength of the given sensory attribute ranging from nothing
to a lot. In the data processing, the sensory scientist measured the position of the mark corresponding to the
perceived intensity.

The results were prepared using multivariate statistics using the following programmes, Panel Check,
Unscrambler and Latentix.

8
 GRINDING RESEARCH REPORT

90g 85g 80g 75g 70g 65g 60g 55g

1.65%

1.60%

1.55%
STRONG STRONG STRONG
1.50%
UNDERDEVELOPED BITTER 50g

1.45%

1.40%

1.35%
45g

1.30%
UNDERDEVELOPED SCAE IDEAL BITTER
STRENGTH - Solubles Concentration

1.25%

1.20%
40g
1.15%
WEAK WEAK
1.10% UNDERDEVELOPED WEAK BITTER

1.05%
14% 15% 16% 17% 18% 19% 20% 21% 22% 23% 24% 25% 26% 27%
EXTRACTION - Solubles Yield

Figure 1. SCAE Coffee Brewing Control Chart

For the whole project the coffee is brewed to the same %TDS. This is to be able to compare different grinders.
The initial setting of a chosen grinder is determined based on different brews and a subjective tasting with the
aim of finding the optimum spot for the coffee. The target %TDS lies within the SCAE ideal square in the Coffee
Brewing Control Chart.

9
GRINDING RESEARCH REPORT

RESULTS
TRIANGLE TESTING - IS THERE A DIFFERENCE?
CONSUMERS

TEST Triangle (discriminative) #1

RESPONDENTS 50

SETTING Canteen, individual tables

GRINDERS La Marzocco MD5 (conical) Nuova Simonelli Mythos (flat)

BREW Wilfa Svart Presisjon

TDS 1,43% (21% extraction)

COFFEE Etiopian Tade, Solberg&Hansen

Are consumers able to taste the difference between coffee from a flat burr
grinder and coffee from a conical burr grinder? This was tested in a canteen setting
where 50 consumers did a triangle test (randomised between the six options,
see Appendix).

The consumers found NO SIGNIFICANT DIFFERENCE BETWEEN FILTER COFFEE

FROM FLAT AND CONICAL BURRS since only 9 consumers were able to pick the
odd sample out.( P<0,001, Meilgaard et al.)

COFFEE PROFESSIONALS

TEST Triangle (discriminative) #2

RESPONDENTS 164

SETTING Booth, Coffee Fair, SCAE, Nice/France

GRINDERS Mazzer Robur (conical)

Mahlkönig K30 (flat)

BREW Programmable Bunn filter ICB brewer

TDS 1,43% (21% extraction)

COFFEE Etiopian Tade, Solberg&Hansen

The triangle test was repeated by coffee professionals at the SCAE World of Coffee fair in Nice, June 2013.
Here 164 people from the coffee industry performed a triangle test. Fully randomised so that no one got
identical test standing next to each other. This was explained to the respondents coming to the booth.

10
 GRINDING RESEARCH REPORT

164
RESPONDENTS
IN NICE
FRANCE

Barista Own Coffee bar Other (student, sales MKT) Equipment

Roaster Import/export beans Research Trainer

Figure 2. 164 respondents in Nice, France

The triangle test participants in Nice is an extensive diversity of the multiple roles in the coffee community with a
predominance of people having everyday hands-on experience with the coffee and thus tasting coffee extensively.

A background questionnaire was obtained in order to get maximum information about the respondents. This
revealed the coffee professionals were familiar with the type of test brew, since 70 people had filter brew as
their preferred coffee drink.

80

70

60

50

40

30
NUMBER OF PEOPLE

20

10

0
aeropress filter drip/machine french press espresso cappuccino americano ice coffee

PREFERRED DRINK

Figure 3. Most preferred coffee drink for the respondents

The coffee professionals feel strongly about their coffee. With the question above regarding preferred coffee drink
you could tick a box if you use sugar and/or milk/cream in the coffee. Several people wrote NO! beside the boxes.

11
 GRINDING RESEARCH REPORT

ORIGIN OF
COFFEE
PROFESSIONALS

NICE
Northern Europe Southern Europe South America Other
Nordic Countries Eastern Europe Asia

Figure 4. Home of respondents in triangle test, Nice 2013

The Coffee Professionals participating in the research were primarily from different parts of Europe, with a
large number of the respondents familiar with the lighter roasting trend from the Nordic countries.

The respondents coming by the NBC booth in Nice to participate in the triangle test were also asked about their
history in the coffee business. This reveals that there is no correlation amongst time in the coffee business and
the ability to distinguish between flat and conical burrs regarding to taste.

45 Wrong
Right
40

35

30

25

20

15
NUMBER OF PEOPLE

10

0
not professional less than 1 year 2-4 years 5-9 years 10-15 years more than 15 years

YEARS IN COFFEE BUSINESS

Figure 5. Length of experience in coffee business

12
 GRINDING RESEARCH REPORT

Also no learning effect from coffee consumption was found since there is no correlation between daily coffee
intake and the ability to get the right answer.

90 Wrong
Right
80

70

60

50

40

30
NUMBER OF PEOPLE

20

10

0
less than 6 per week 1 cup per day 2-4 cups per day 5-9 cups per day more than 10 cups per day

COFFEE CONSUMPTION

Figure 6. Daily intake of coffee

CONCLUSION TRIANGLE TEST NICE, FRANCE

Since only 48 persons of the 164 coffee professional got the right “odd cup” of the three served it is significantly
concluded that COFFEE PROFESSIONALS COULD ACTUALLY NOT TASTE THE DIFFERENCE BETWEEN FLAT
AND CONICAL BURRS (48 out of 164 >p<0,001)

The poor result was somewhat unexpected to the research team, although no one assumed it was easy. The
result was an actual shock to the coffee community itself and was received with disbelief.

13
GRINDING RESEARCH REPORT

THE SENSORY PROFILING TEST

Since the taste differences between coffees from flat versus conical burrs are extremely small and not
possible to detect in a triangle test another approach was needed. The minor variances in the profile due to the
grinder used was to be described by a sensory panel.

An expert sensory panel consisting of people working professionally with coffee (roaster, previous coffee bar
owner, Cup of Excellence judge, barista and coffee consultant) was gathered in Copenhagen.

Since these people were accustomed to evaluating coffee by classic cupping a challenge was upon the
panel leader. Could Francisca Listov-Saabye “teach old dogs new tricks”? The training sessions began and the
line scale was introduced along with physical references. The panel agreed upon following sensory attributes
describing the coffee spectrum:

CATEGORY INDICATION IN PLOT ATTRIBUTES

SMELL S_xxxx Citrus, floral, chocolate, tobacco, caramel
Eg S_tobacco is smell of tobacco

TASTE T_xxxx Acidic, sweet, bitter, citrus, fruit, floral, roasted,

Eg T_citrus is taste of citrus liquorice, tobacco, caramel

MOUTHFEEL Mouth feel

The profile sessions took place at the NBC Coffee lab, at individually tables. Brewing area was out of sight. The
panel members were served coffee in individual order and at their own pace. They received no information
about the aim of the project so they had no background facts connected to the coffees and did not know what
differentiated the samples (that being the different grinders but the same coffee).

The results were analysed and displayed by multivariate statistics.

TEST PROFILING (DESCRIPTIVE) #1

JUDGES 6

SETTING NBC Coffee Lab, Copenhagen

5 GRINDERS Marzocco MD5 (conical)

Faema MC99 (conical)
Nuova Simonelli Mythos (flat)
Mahlkönig Vario (flat)
Mahlkönig K30 (flat)

REPLICATES 2

BREW Wilfa Svart Presisjon

TDS 1,43% (21% extraction)

14
 GRINDING RESEARCH REPORT

1 Correlation Loadings (X and Y)

conical
0.9
0.8
0.7
0.6
FAEMA
0.5 MARZOCCO

0.4
0.3 T_caramel
S_caramel
T_bitter
0.2 T_sweet
pan5
T_liquorice
Mouthfeel T_roasted
S_floral
0.1 pan4 Pan3 T-tobacco
pan6 T-floral
0 S-chocolate T-citrus S-tobacco
T_acidic
-0.1 pan2 S-citrus
T-fruit
-0.2 MYTHOS
VARIO
-0.3 Pan1
K30
-0.4
-0.5
-0.6
FACTOR-2 (27%, 3%)

-0.7
-0.8
flat
-0.9
-1
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

FACTOR-1 (8%, 49%)

Figure 7. PLS plot. Coffee expert panel profiling

JUDGES PAN1-PAN6 The sensory judges: Panel member no. 1-6

BURRS CONICAL, FLAT The location of the grinders having flat or conical properties

GRINDERS K30, FAEMA, MARZOCCO, Position of the grinder in the “sensory world of the Ethiopian coffee” position
VARIO, MYTHOS

ATTRIBUTES SENSORY ATTRIBUTES Location of sweet taste in the sensory profile

EG T_SWEET

The plot (figure 7) shows that all the sensory attributes (in red) describing the coffee are located in one side.
This is due to the fact that the judges (pan1 – pan6) themselves actually accounts for the biggest explanation of the
model-marked by the red dotted ellipse. The grinders have a much narrower variance- marked by the blue ellipse.

The position of the attributes and the panel members indicates that they were not able to use the scale in
the same way and that e.g. panel member no. 3 consistently graded the coffees higher intensities that panel
member 2 and 6.

Looking into the raw data it seems that the panel members, all being coffee experts or professionals, can’t
help to evaluate more than the plain attributes-one at a time. They will start to guess where the coffee is from,
how it is brewed and so on. They can’t put the coffee nerd aside!

15
 GRINDING RESEARCH REPORT

Is it all bad then, trying to get coffee people to use traditionally sensory science? No – there are some tendencies:

If a new multivariate model is computed to distinguish between the grinders there is a clear tendency towards
the conical grinders – marked by the red dotted ellipse in figure 8- are described by sweet taste whereas the flat
grinders –marked by the red dashed ellipse - are described by more acidic taste.

1
Correlation Loading (X and Y)

0.9
0.8
VARIO
0.7
0.6
0.5
0.4
0.3
MARZOCCO
0.2
0.1 1 4
MYTHOS
0
SWEET
2 8
9
10
13
14
ACIDIC
7 11
3 12
-0.1 5
6
-0.2
-0.3
-0.4
FAEMA
-0.5
-0.6
FACTOR-2 (25%, 1%)

K30
-0.7
-0.8
-0.9
-1
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
FACTOR-1 (25%, 1%)

1. T_roasted 4. Mouthfeel 7. T_bitter 10. T_floral 13. T_acidic

2. T_liquorice 5. T_caramel 8. S_caramel 11. T_citrus 14. S_chocolate
3. S_floral 6. T_tobacco 9. S_tobacco 12. T_fruit

Figure 8: PLS model flat versus conical grinders

These tendencies are worth looking into. Therefore a similar sensory profiling was conducted. This time by
a sensory panel consisting of food science researchers and master students. All located at the Food Science
Department of Copenhagen University. The panel members had prior experience with sensory profiling of
different food stuffs and are all familiar with drinking coffee.

16
GRINDING RESEARCH REPORT

SENSORY PROFILING, UNIVERSITY PANEL

Test Profiling (descriptive) #2

Judges 6

Setting Sensory Lab, Copenhagen University, FOOD

(Accredited)

6 Grinders Marzocco MD5 (conical)

Faema MC99 (conical)
Mazzer Robur (conical)
Nuova Simonelli Mythos (flat)
Mahlkönig Vario (flat)
Mahlkönig K30 (flat)

Replicates 3

Brew Wilfa Svart Presisjon

TDS 1,43% (21% extraction)

The sensory lab at the University is an accredited sensory evaluation lab consisting
of individual booths in a closed, neutral, sound proof and odour free environment.
The randomised samples were served to the judges from a hatch.

Again, the samples we served individually at each sensory judge’s own pace.
Serving was randomised across judges to eliminate carry-over effects.

During the training sessions the university panel came up with similar sensory
attributes to describe the coffees although not exactly the same words as the
sensory panel consisting of coffee professionals. Some words were added, such as
sour and burnt, and some words were specified, such as citrus becoming lemon.

CATEGORY INDICATION IN PLOT ATTRIBUTES

SMELL S_xxxx Hazelnut, lemon, chocolate, floral, tobacco, caramel,
Eg. S_lemon is smell of lemon

TASTE T_xxxx Sweet, acidic, bitter, sour, tobacco, lemon, chocolate,

Eg T_bitter is bitter taste liquorice, hazelnut, roasted, burnt, caramel

MOUTHFEEL M_astringency astringency

The sensory judges rinsed their palates in between samples with still and sparkling water.

17
 GRINDING RESEARCH REPORT

1.0
FAEMA

S_chocolate

0.5 T_hazelnuts

M_astringent
T_burnt S_floral
S_tobacco
T_bitter T_tobacco T_sweet

0.0 MAHLKÖNIG LA MARZOCCO SIMONELLI MYTHOS

K30 T_liquorice MAZZER ROBUR

S_hazelnuts

S_caramel
T_chocolate
T_sour

-0.5

T_roasted
S_lemon
T_lemon
T_caramel
PC2 (20.2%)

T_acidic

MAHLKÖNIG
VARIO
-1.0
-1.0 -0.5 0.0 0.5 1.0
PC1 (52.3%)

Figure 9. Principal component 1 versus principal component 2. For further explanation of multivariate statistics see references

Figure 9 shows the correlation between the grinders- written in red- and the sensory attributes – written in
black. Hence it shows that the Mahlkönig K30 grinder produces a coffee that has a higher intensity of bitter,
burnt and tobacco notes whereas opposite both the Simonelli Mythos and the Mazzer Robur grinder gives a
coffee described as more sweet, with both hazelnut smell and taste.

The Faema C99 grinder evaporates a more intense smell of chocolate compared to Mahlkönig Vario which has
a more roasted and acidic profile. The location of La Marzocco MD5 in the middle indicates that the grinder, in
this study, did not have a profound characteristics compared to the other five grinders.

18
 GRINDING RESEARCH REPORT

1.5

SIMONELLI
1

T_liquorice

FLAT
0.5 K30

T_chocolate
T_sweet

T_caramel
0 VARIO T_acidic 1
3 2 FAEMA
5 64
T_caramel
T_roasted

T-hazelnuts
S_lemon
-0.5 T_lemon
CONICAL
S_chocolate
PC#4 (13.073%)

-1 MAZZER Flora
MARZOCCO
S_Caramel

-1.5 -1 -0.5 0 0.5 1 1.5 2

PCPG#2 (24.408%)

1. T_burnt 2. T_bitter 3. S_hazelnut 4. M_astringent 5. T_tobacco 6. S_tobacco

Figure 10.

Vario, K30, Simonelli Flat grinders

Mazzer, Marzocco, Faema Conical grinders

T_chocolate, T_liquorice, The sensory attributes describing the coffee

T_hazelnuts, S_caramel etc.

The above figure 10 displays a model computed to differentiate between the flat and the conical grinders.
This reveals that the flat grinders in this study enhances the Ethiopian coffee’s liquorice and chocolate taste
notes where the conical grinders on the other hand enhances smells of flowers, chocolate and caramel.

19
GRINDING RESEARCH REPORT

PARTICLE SIZE DISTRIBUTION

The Nuova Simonelli R&D Lab in Italy kindly provided particle size distribution analysis on blind coded samples
from grinders in the study.

The technique for the individual analysis of the ground coffee uses Hemispherical Randomic Distributor with
compressed air injector. The vision system was developed by UNIVPM for Nuova Simonelli.

The samples were prepared at the NBC Coffee Lab in Copenhagen by finding the individual setting of each
grinder which produced a brewed coffee with identically %TDS. The results of the analysis did not reveal a
grouping in flat or conical grinders. It’s not a black and white picture. The overall view of the grinders in figure
5 are all based on the same beans: A light roasted Ethiopian coffee.

80 Mahlkönig K30
Mahlkönig Vario
Simonelli Mythos new
Simonelli Mythos old
60 Faema
La Marzocco
Mazzer Robur

40

20

0
215 275 362 512 1050

Figure 11. Overall view of particle size distribution (% of sample vs. particle size mean diameter [µm]

Looking into the overall picture of the grinders some resemblances are seen. The Mahlkönig Vario and
the new Simonelli Mythos have almost identical curves as do the Faema MC99 and Mahlköning K30 curves
show similarities.

So some grinders seem to resemble each other in particle size distribution despite of burr type being flat
or conical. This curve resemblances are not repeated in the sensory profile!

20
GRINDING RESEARCH REPORT

Mean distribution of volume

0.5

0.4

0.3

0.2

0.1

0.0
-1 0 1 2 3 4 5 6 7 8 9

Mean distribution of volume

0.5

0.4

0.3

0.2

0.1

0.0
-1 0 1 2 3 4 5 6 7 8 9

Figure 12. Mean particle size distribution

The above two plots of particle distribution are the same grinder (the new Simonelli Mythos), same setting but
light roast above and dark roast below which just confirms the complexity of the grinder topic.

21
 GRINDING RESEARCH REPORT

DISCRIMINATIVE CUPPING TEST AT THE NBC 2013

In the sensory profiling the academic panel at the university found that the attributes BURNT and SOUR were
significantly differentiating between the grinders. The cupping session during the Nordic Barista Cup in Oslo,
Norway in September 2013 was based on this finding. The 147 attendees were given a test to find the odd
sample out of three. Two different tests were conducted: One with respect to SOUR (“which sample differs from the
two others in sourness?”) and BITTER (“which sample differs from the two others with respect to bitterness?”)

Sample means & LSD 1 marzocco SIGNIFICANCE

11
2 faema ns
4
3 mazzer p<0.05
10 4 K30 p<0.01
5 vario p<0.0001
9 6 simonelli

8 15 2
3
7

6
6
5 4

5
1

3
2
3
SCORE

6
T_BITTER T_SOUR
ATTRIBUTES

Figure 13. The sensory attributes sour and bitter differentiate the grinders. University profiling

Test Triangle (discriminative) #3&4

Respondents 147

Setting Cupping tables, NBC Oslo 2013

Grinders Mazzer Robur (conical) Dalla Corte DC-one (flat)

Brew Programmable Bunn filter ICB brewer

TDS 1,43% (21% extraction)

Coffee Etiopian Tade, Solberg & Hansen

The NBC attendees found it slightly easier to discriminate the samples when referring to SOUR rather than
concentrating on BITTER taste. But only 20 of the 147 persons participating the grinding cupping got both
tests right. The attendees, being coffee professionals from all over the world, WERE NOT ABLE TO TASTE THE
DIFFERENCE BETWEEN FLAT AND CONICAL BURRS.

22
GRINDING RESEARCH REPORT

CONCLUSION
The NBC/SCAE Gold Cup Grinding Research project set out to find out if there really was a detectable taste
difference between coffees from a flat burr grinder compared to a conical burr grinder.

Nothing in grinders is black and white. More like a thousand shades of grey. The differences are so small that
it is hard to detect. And given a triangle test neither consumers nor coffee professionals were able to find the
difference. Not even when a descriptive word exists to focus the mind when tasting.

Grinders are so much more than the set of burrs. Not only does the type (flat or conical) matter, but also the
size of the burrs, the size and speed of the motor, the feeding angle into the grinder, the retention and regrind
as well as the physical appearance itself.

Grinders are individuals – having an affect on the coffee as well as the operating barista.

23
GRINDING RESEARCH REPORT

FUTURE
There are still a lot of scientific research opportunities waiting to be discovered in speciality coffee. Francisca
hopes to be part of the myth busting as well as the revelation of truths in the future.

Francisca’s dream is to get a PhD in coffee research. But partners and sponsors are needed in order to plan
and have a three year project approved.

Multiple ideas for smaller or bigger research projects are available however for cooperation and
ventures immediately:

• Can you actually taste the difference between washed and dry filters in a pour over coffee?
• What is the connection between degree of roasting and particle size distribution?
• What is the link concerning the age of the burrs and taste effect in the coffee?
• How does the speed of the grinder affect the taste of coffee and the particle size distribution?
• Sensory profiling of different origin of coffee
• How does the teeth pattern of the burrs affect the taste of the coffee?

Please don’t hesitate to contact Francisca Listov-Saabye.

24
GRINDING RESEARCH REPORT

THANKS TO THE RESEARCH PARTNERS & SPONSORS

The NBC/SCAE Gold Cup Grinding Research project is thankful for the partners and sponsors whose help we
could not do without. In accordance with the sponsorship the research partners receive this additionally report
with more details and unpublished scientific results. We appreciate your donation:

As well as the co-operation with the University of Copenhagen, SCIENCE; Faculty of Life Science Department
of Food Science.

Francisca would also like to acknowledge the people who help throughout the project – either guiding the way
into Speciality Coffee or otherwise professional sparring:

Björg Brend Jens H. Thomsen Michael Bom Frøst Morten Munchow

Tim Wendelboe Michael de Renouard Thomas Skov Troels Overdal Poulsen

Special thanks to Jens Nørgaards for his vision, support and guidance.

25
GRINDING RESEARCH REPORT

APPENDIX & REFERENCES

SET UP TRIANGLE TESTING:

AAB ABA BAA

BBA BAB AAB

Testing two different coffees A and B. In this study being from flat and conical burr grinder respectively.

REFERENCES SENSORY SCIENCE

Meilgaard, Civille and Carr. Sensory Evaluation Techniques, fourth ed. 2007, CRC Press

Lawless and Heyman. Sensory evaluation of food – principles and practices. 1998. Chapman & Hall

REFERENCES MULTIVARIATE STATISTICS

Martens, Wedøe and Martens. Manual for sensory data analysis in the Unscrambler. 1999.

Martens and Martens. Multivariate analysis of quality – an introduction. 2001. Wiley & sons

26
GRINDING RESEARCH REPORT

27