THE COFFEE ROASTER’S

COMPANION
 n
1 n

Table of Contents

Acknowledgments vii
Preface ix
Introduction xi
1 Why We Roast Coffee Beans 1
2 Green-Coffee Chemistry 2
Structure
Sugars
Lipids
Proteins
Alkaloids: Caffeine and Trigonelline
Moisture Content
Organic Acids
Gases and Aromatics
3 Green-Coffee Processing and Storage 4
Primary Processing Methods
Wet/Washed
Dry/Natural
Pulped/Natural
Green-Coffee Storage
Water Activity and Moisture Content
Seasonality
4 Physical Changes During Roasting 9
Color Changes
Classic Definitions of Roast Degree
Cinnamon
City
Full City
Viennese
French
Italian
Structural Changes
Inner-Bean Development
Bean Size, Density, and Weight Loss
5 Roasting Chemistry  15
Green-Coffee Storage and Consistency

Changes in Chemical Composition
Ambient Conditions

Development of Acids During Roasting
Chimney Cleaning

Aroma Development
Managing Different Batch Sizes

Maillard Reactions and Caramelization
12 Measuring Results  53

Caffeine Content and Roasting
All About Bean Probes
6 Heat Transfer in Coffee Roasting  19 Choosing a Probe

Convection, Conduction, and Radiation Installing a Probe

Heat Transfer and Temperature Gradient Weight Loss

Heat and Mass Transfer Within Coffee Beans Measuring Roast Degree

Heat Transfer and Moisture Verification of Development Using a Refractometer
7 Roasting Machine Designs  22 13 Sample Roasting  57
Classic Drum
14 Cupping  59
Indirectly Heated Drum
How to Cup
Fluid-Bed
Cupping Recommendations
Recirculation
The Phases of Cupping
8 Progression of a Roast  29 Dry Aroma, or Fragrance
The Illusion of the S Curve Wet Aroma
The Myth of the Drying Phase Tasting the Coffee When It Is Hot
The Middle (Nameless) Phase Tasting the Coffee When It Is Cool
First Crack How to Interpret Cupping Results
Second Crack
15 Roasting, Brewing, and Extraction  66
Development Time

Testing Roast Development
9 Planning a Roast Batch  34
Calibrating Extraction
Batch Size
Roasting for Espresso
Setting Airflow
Blending
Adjusting the Air–Fuel Ratio
16 Storing Roasted Coffee  70
Charge Temperature
Machine Design 17 Choosing Machinery  72
Batch Size Features to Consider when Selecting a Roaster
Bean Density Capacity
Bean Size Configuration
Bean Processing Method The Drum
Intended Roast Time Airflow
Determining Roast Time Gas Control
Drum RPM Drum Speed
Bean Moisture, Density, and Size Data-Logging Software
Automated Profiling Software
10 The Three Commandments of Roasting  42
Pollution-Control Devices
I. Thou Shalt Apply Adequate Energy at the Beginning of a Roast
II. The Bean Temperature Progression Shalt Always Decelerate
Parting Words  79
III. First Crack Shalt Begin at 75% to 80% of Total Roast Time
Glossary  80
11 Mastering Consistency  49
References  84
How to Warm Up a Roaster
Between-Batch Protocol Index  87
Other Tips to Improve Batch-to-Batch Consistency
About the Author  89
 6
To understand the Maillard reactions’ contribution to flavor, consider the
different effects of roasting and boiling on the flavor of meat: Roasting imparts n    n
aromatics, complexity, and depth of flavor absent in boiled meats. Maillard
reactions contribute similar roast-flavor traits and complexity to coffee beans.
During roasting, once a bean’s internal temperature is high enough to boil Heat Transfer
off most of its moisture, the temperature rises more rapidly, speeding Mail-
lard reactions. This is one reason aroma development accelerates at mid-roast. in Coffee Roasting
Maillard reactions become self-sustaining at above 320°F (160°C).
Unlike Maillard reactions, caramelization is a form of pyrolysis, or thermal
offee roasting machines transfer heat to beans by convection, conduction,
decomposition. Caramelization begins at approximately 340°F (171°C),19 as the
heat of roasting breaks apart molecules of sugar and produces hundreds of new
C and radiation. Each roasting machine transfers heat by a different mix of
compounds, including smaller, bitter, sour, and aromatic molecules and larger, these mechanisms. The following is an overview of how machine design affects
brown, flavorless molecules.19 Although most people associate the word “cara- heat transfer. I discuss roasting machine designs extensively in Chapter 7.
mel” with a very sweet dessert food, caramelization, ironically, decreases the Convection, Conduction, and Radiation
sweetness and increases the bitterness of a food or beverage. Lighter roasts
are sweeter, and darker roasts more bitter and caramelly, primarily because of “Classic” (my term) drum roasters, which apply heat directly to the drum, cook
caramelization. beans primarily by convection and secondarily by conduction. Radiant heating
from hot roasting-machine surfaces and between neighboring beans makes a
Caffeine Content and Roasting small contribution to heat transfer as well. In a personal communication with me,
Despite what almost everyone has heard, darker roasting does not decrease a representative of a well-known German manufacturer estimated heat transfer
the caffeine content of coffee beans. Caffeine levels are virtually unchanged in his company’s drum roasters to be 70% by convection and 30% by conduction.
by roasting,3 as caffeine is stable at typical roasting temperatures. Given that Indirectly heated drum roasters segregate the drum from the heat source to
beans lose mass during roasting, their proportion of caffeine by weight increases maintain a cooler drum during roasting. Convection contributes a higher pro-
during roasting. Therefore, assuming one brews coffee of all roast degrees with portion of the heat transfer in these machines.
a particular ratio of water to ground-coffee mass, rather than volume, darker Fluid-bed roasters have no drum, and they roast by keeping the beans aloft
roasts will yield brewed coffee with higher caffeine content. in a high-velocity stream of hot gases. Recirculation roasters, such as the Lor-
ing Smart RoasterTM, capture and reuse a proportion of the exhaust air from the
roasting process. Both of these roasting machine designs transfer heat almost
exclusively by convection.
At the beginning of a roast batch, charging the beans introduces a large
volume of room-temperature beans and air into the hot roaster, sending the
environmental temperature in the roaster plummeting. During the first few
minutes of a batch in a classic drum roaster, conduction from the hot drum
plays a significant role in transferring heat to the beans. As the air temperature
in the roaster rebounds after its initial plunge, convection comes to dominate
heat transfer. In such a machine the drum acts as a “heat-storage” device that
jump-starts development early in a batch. Convection-oriented machines call
for the use of hotter charge temperatures to
provide adequate heat transfer early in a roast
and compensate for lack of a heat-storing drum. Establishing a high DT
Heat Transfer and Temperature Gradient early in a roast and
minimizing it by the end
The first two-thirds or so of roasting is an endo-
thermic process, meaning the beans absorb of a roast is essential
energy, and heat is conducted from the outer to creating good inner-
bean to the inner bean. The temperature gra- bean development
dient, or “DT,” within the beans largely deter- and a uniform roast.
mines the rate of heat transfer. Simply put, a

19
18 T H E C O F F E E R O A S T E R ’ S C O M PA N I O N
 n 
7   n

Roasting 5
Machine
Designs
2

coffee-roasting machine is a specialized oven that transfers heat to coffee

A beans in a stream of hot gas while continually mixing the beans to ensure 6 1
they roast evenly. Several types of roasters are in use today in the specialty
coffee industry: classic drum roasters, indirectly heated drum roasters, fluid- Single Drum
bed roasters, recirculation roasters, and several others. Recirculation roasters
return a portion of the exhaust air to the burner chamber to assist in heat gen- 4 3
eration for roasting. I will use the term “single-pass” to refer to machines that
do not recirculate exhaust air. Each roaster design has distinct advantages and
disadvantages, though no new design has eclipsed the popularity of the classic
drum roaster, the design of which has not changed much in the past century.

Classic Drum
A classic drum roaster consists of a solid, rotating, cylindrical steel or iron drum
laid horizontally on its axis, with an open flame below the drum. The flame Classic drum roaster. Beans (brown arrows) enter the roasting drum (1) through the loading funnel
(2). After roasting, the beans cool in the cooling bin (3). Air (blue arrows) passes from the combus-
heats both the drum and the air to be drawn through the drum. A fan draws tion chamber (4) through the roasting drum and exhausts through the chimney (5) by way of the
hot gases from the burner chamber through the rotating beans and exhausts cyclone (6), which traps chaff.
the smoke, steam, and various by-products of roasting and combustion out of
the building through a vertical pipe, or “stack.” The drum’s rotation mixes the
beans while they absorb heat by conduction from direct contact with the hot Outer Drum
drum and convection from the air flowing through the drum. Single Drum
Inner Drum
At the completion of a roast, the machine operator opens the door to the drum,
dumping the beans into the cooling bin, which stirs the beans while a powerful
fan draws room-temperature air through the bean pile to cool it rapidly.
The best classic drum roasters have a double drum of two concentric layers
of metal separated by a gap several millimeters wide. In a double drum, direct
contact with the flame heats the outer drum, while the inner drum remains
cooler. A double drum decreases conductive heat transfer and limits the risk of
tipping, scorching, and facing. (Henceforth, these three are referred to in this
text as “bean-surface burning.”) If you buy a classic drum roaster, I strongly
suggest finding one that has a double drum.
Advantages: The single pass of the roasting gas provides a clean roasting
environment, and the drum serves as an effective heat-storage system, provid-
ing conductive heat transfer, especially during the first few minutes of a batch.
Disadvantage: Overheating the drum metal can easily lead to bean-surface
burning. Single drum (left) and double drum (right) Outer Drum
Inner Drum

22 ROASTING MACHINE DESIGNS 23

 Which Roast is More Developed?

n 
10   n
batch A batch B
Which Roast Is More Developed?

The Three 500

Commandments 400
of Roasting

Temperature (°F)
300
lease don’t take the word “commandment” too seriously. One may trans-
P gress some of these rules harmlessly on occasion. As with a certain other
200
list of commandments, however, if you make a habit of ignoring the rules, you
might end up in a bad place.
As a roaster and a consultant over the past nineteen years, I’ve had the
100
opportunity to cup and view the roast data for each of more than 20,000 batches
0:00 2:00 4:00 6:00 8:00 10:00 12:00
roasted on a variety of machines by various methods. About five years ago, I
spent several days poring over reams of roast data in an attempt to find the Time
common elements in the best batches I’d ever tasted.* To be clear, I’m not refer-
ring to “really good” batches. I focused only on the data from batches so special Batch A and batch B had identical charge temperature, drop temperature, and roast time. Given that
batch A’s bean temperature initially rose more of
Comparison quickly than batch B’s,
Temperature batch A is more developed.
Gradients
that I could “taste” them in my memory months or years after physically tasting
them. That effort yielded what I think of as the “commandments of roasting.”
A method graduated to a commandment only if it seemed to apply to a great
variety of coffees and roasting machines. I’ve been testing and refining the com- Comparison of Temperature Gradients
mandments for five years, and so far I’ve yet to find a situation in which coffee
outer bean A inner bean A outer bean B inner bean B
tastes better when a commandment is broken. I’ve also had opportunities to
500
test the commandments in reverse; the times I’ve tasted stellar roasts from
others and the roaster was kind enough to share the roast data with me, sure
400
enough, the profiles conformed to the commandments.
I can’t fully explain why these methods work. But I’m confident that if you “nameless phase”

Temperature (°F)
remain open-minded and apply these techniques carefully and completely, you 300
will be impressed by how much better your roasts taste.

I. Thou Shalt Apply Adequate Energy at the Beginning of a Roast 200

Applying sufficient heat at the beginning of a roast is essential to achieving T

T
optimal flavor and proper bean development. While one may begin a roast with 100

too little heat and still cook the bean centers adequately, the flavor of such cof-
fee may suffer because the operator must lengthen the roast time excessively to
compensate for the insufficient early heat transfer. 0:00 2:00 4:00 6:00 8:00 10:00 12:00

Time

This graph illustrates the importance of establishing a large DT early in a roast. In batch A, the
machine operator applied sufficient energy early in the roast, creating a large DT, which gave the
inner bean the impetus to smoothly “catch up” to the outer bean by the end of the roast. Batch B
began sluggishly, creating a smaller early DT. Relative to batch A, the operator applied more heat
* I compiled and evaluated my roast data by using a pencil, calculator, and spreadsheet. mid-roast to adequately cook the outer bean in a similar total roast time. However, the extra energy
These days one can analyze such data much more efficiently with the aid of computer was too little, too late for the inner bean’s temperature to match that of the outer bean, and batch B
software such as Cropster’s “Roast Ranger” application. was underdeveloped.

42 THE THREE COMMANDMENTS OF ROASTING 43

 11
III. First Crack Shalt Begin at 75% to 80% of Total Roast Time
Experience has taught me that the roast time from the onset of first crack* to n    n
the end of a roast should make up 20%–25% of total roast time. Put another
way, first crack should begin at between 75%–80% of total roast time. I’m confi-
dent that the optimal ratio is actually in a much narrower range, and the ratio Mastering
should vary slightly depending on roast degree desired, but I don’t have enough Consistency
data yet to back up those beliefs.
If first crack begins at earlier than 75% of total roast time, the coffee will
probably taste flat. If more than 80% of the total roast time elapses before first uch like the elusive “God shot” of espresso, most companies roast the occa-
crack begins, development will likely be insufficient. M sional great batch but can’t seem to reproduce it consistently. Variations
Most roasters seem to adjust a roast’s “development time” separately from in a roaster’s thermal energy, green-coffee temperature and moisture, ambient
the rest of the roast curve, but such an approach will often lead to baked flavors conditions, and chimney cleanliness all collude to make roasting inconsistent.
or underdevelopment. Instead of focusing on development time, I recommend I’ve designed the tips in this chapter to help you control or lessen the impacts of
that roasters adjust the last phase of a roast curve to ensure it is proportional these factors. Following these recommendations will help any roaster improve
to the entire roast curve. I hope roasters will find this suggested ratio useful consistency.
and that the conversation among roasters shifts from “development time” to
“development time ratio” or aIdeal
similar phrase.Time Range
First-Crack How to Warm Up a Roaster
At a cupping of some lovely Cup of Excellence coffees a few years ago, I noticed
that one of the samples was very underdeveloped and another was slightly
Ideal First-Crack Time Range underdeveloped. The other cups had varying degrees of good development. It
bean profile
dawned on me that those two cups had been brewed from, respectively, the first
500
and second batches roasted that day. I suggested to my cupping host the order
in which he had roasted the samples that morning. I had guessed the order
correctly.
Every roaster I’ve ever asked has admitted to having difficulty with the qual-
400 75%–80% ity of the first few batches of a roasting session. The problem is usually caused
by inadequate warming up of the roasting machine. Most machine operators
warm up a roaster to the charge temperature and then idle the machine at or
Temperature (°F)

finish near that temperature for some amount of time, usually 15–30 minutes, before
300 charging the first batch. This protocol guarantees that the first batch will roast
sluggishly compared with successive batches.
The problem is that temperature probes are poor indicators of a machine’s
thermal energy. (See “Charge Temperature” in Chapter 9.) As a cold roasting
machine warms up, although the temperature probes quickly indicate that the
200
air in the machine has reached roasting-level temperatures, the mass of the
machine is still much cooler than the air in the drum. If one charges a batch at
this point, the machine’s mass will behave akin to a heat sink and absorb heat
from the roasting process, decreasing the rate of heat transfer to the beans.
100 After several roast batches, the machine’s thermal energy will reach an equi-
0:00 2:00 4:00 6:00 8:00 10:00
librium range within which it will fluctuate for the remainder of the roasting
Time session.
The trick to normalizing the results of the first few batches of a roasting
First crack should ideally begin in the shaded zone.
session is to seemingly overheat the machine during the warm-up, before stabi-
lizing it at normal roasting temperatures. To my knowledge, there is no practi-
cal, precise way to measure a roaster’s thermal energy. However, the operator
* I consider the beginning of first crack to be the moment the operator hears more than one can apply some informed experimentation to establish a protocol that brings a
or two isolated pops.

48 T H E C O F F E E R O A S T E R ’ S C O M PA N I O N 49
 n 
17   n

Choosing Machinery

electing a roasting machine is a long-term commitment, and I hope read-

S ers do their homework before buying a machine. Most small roasters,
especially first-time buyers, don’t have the experience to evaluate machines
properly, so if that’s you, I recommend seeking expert advice before making
what is probably your company’s largest investment. You must choose carefully
because the majority of machines on the market today will limit your coffee’s
quality or consistency, though their sales representatives may neglect to tell
you that.

Features to Consider when Selecting a Roaster

Every roasting company has its unique list of needs and preferences when
choosing a roaster, such as aesthetics, machine footprint, cost, and so on. While
I can’t comment on those company-specific requirements, I offer the following
technical recommendations to help you choose a roaster.

Capacity
First, decide how much roasting capacity you need. Second, use a manufactur-
er’s stated capacity as a starting point and look up a machine’s BTU rating to
estimate what its realistic capacity might be. Finally, given that every machine
will have different heat-transfer efficiency, I recommend that you contact a few
users of a given machine to ask about their typical batch sizes and roast times. Single-walled steel drum
Using those three pieces of information, you should have a good sense of the
machine’s realistic capacity. a cast-iron drum (a small, newer roaster manufactured in Taiwan) and one
machine with a sheet-iron drum, but every other machine I’ve ever seen has
Configuration had a steel drum.
A roasting machine’s configuration probably has the greatest effect on the qual- Most roasting drums are made of carbon steel, but some manufacturers
ity of coffee that it can produce. As I’m sure you’ve gathered by now, I recom- have recently begun building machines with stainless-steel drums; this seems
mend single-pass roasters over recirculation roasters, despite the latter’s energy reasonable, but I don’t have enough experience with them to have an opinion
efficiency. I also recommend an indirectly heated drum, or a double drum, over about their performance. Stainless steel drums may develop hot spots more
a standard flame-on-drum design. A single-pass roaster with a double drum or easily than mild carbon steel ones, but that’s probably not a serious concern,
indirectly heated drum will maximize your chances of producing great coffee given the drum’s rotation and an adequate thickness.
and minimize potential flavor taints due to bean-surface burning or a smoky
roasting environment. Airflow
I’ve come across few roasters with inadequate airflow but several machines
The Drum with poor airflow adjustment mechanisms. Ideally, your exhaust fan’s RPM
If you buy a classic drum roaster with a flame-on-drum configuration, I rec- should be adjustable in minute, stepless increments. Subtler airflow adjust-
ommend choosing a machine with a carbon-steel drum. Contrary to popular ments will produce smoother roast profiles. Machines with two or three discreet
belief, most old, German “cast-iron roasters” have carbon-steel drums, not cast- airflow settings, usually controlled manually by a damper, are acceptable but
iron drums. Those machines and many others often have cast-iron faceplates, limiting. Not only are the settings usually too far apart, forcing the machine
drum spokes, and drum paddles, but steel drums. I have seen one machine with operator to compromise and choose a suboptimal setting, but the large shifts

72 C H O O S I N G M A C H I N E RY 73
 n 
1   n

Index

A E
acids, 2–3, 15–16, 31, 64, endothermic flash, 32
afterburner, 77–78
airflow, 35–36, 73–74 F
air–fuel ratio, 36–37 freezing of coffee beans
alkaloids, 2–3 green, 6
aroma development, 17 roasted, 71
automated profiling software, 76–77
G
B GrainPro bags, 6
baked flavors, 33 green coffee, 1–8, 15–17, 39, 41, 51
batch
planning, 34–41 H
size, 34–35, 52 heat shield, 24
between-batch protocol, 50 heat transfer, 17–21
blending, 68–69 heat-sink effect, 49

C M
caffeine, 3, 18 machinery. See: roasting
caramelization, 2, 17–18, 31 Maillard reactions, 9, 17–18, 31
charge temperature, 4, 37–39 manometer, 58
chemistry of roasting, 1–2, 15–21 moisture content, 7, 21
chlorogenic acid, 16
cleaning your roasting machine, 49, 51–52 O
color changes of beans during roasting, organic acids, 3
3, 9–12, 17–18, 31, 55, 57–58
commandments of roasting, 33, 42–48 P
conduction, 19 planning a roasting session. See: batch,
consistency between batches, 41–55 planning
convection, 19 pollution control, 77–78
crack probe, bean, 29–30, 32, 35, 37, 49–50, 52–
first, 9–10, 12, 14, 16, 21, 30, 32, 45–48, 54, 58, 75
53, 58, 67 processing methods, 4
second, 11–14, 33, 45 dry, (natural) 4
Cropster, 44–45, 75 pulped natural, 4
cupping, 59–65 wet, 4
purchasing a roasting machine, 72–78
D
development time, 29, 33, 48 R
drum radiation, 19
RPM, 40, 7 rate of rise (ROR), 32, 42–47, 65
types, 19, 22–26, 37–40, 72–73 refractometer, 55–56, 66–67
drying phase, 9, 29–31

87
roasting S
chemistry, 15 seasonality, of green coffees, 8
consistency, 49–52 software, roasting, 28, 42, 75–77
ambient conditions, 51 stack effect, 51
batch size, 52 storage
chimney cleaning, 51 of green coffee, 5–7
green storage, 51 of roasted coffee, 70–71
degree
cinnamon roast, 10 T
city roast, 10 temperature
French roast, 12 effect on bean storage, 5–7
full city roast, 11 gradient, 19–20, 43
Italian roast, 12 probes, 49, 53–54
measurement, 55 testing
Viennese roast, 11 of bean development, 13, 53–55, 65–68
development, 13, 31, 33, 39–40, 42–48, tools, 55, 66–68
54–55, 66–68 thermometric lag, 30
machine designs third-wave coffee, 10
classic drum, 22–26 trigonelline, 3
double-drum, 22–23
    fluid-bed, 26 W
indirectly heated, 24–26 warming up a roaster, 49
recirculation, 27–28 water activity, 7
single-drum, 23 weight loss of beans during roasting, 13–14,
machine 54–55
maintenance, 51–52
selection, 72–76
time, 39–40

88 T H E C O F F E E R O A S T E R ’ S C O M PA N I O N