Applied Thermal Engineering 29 (2009) 998–1004

Contents lists available at ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Experimental investigation of steam pressure coffee extraction in a stove-top

coffee maker
L. Navarini a, E. Nobile b, F. Pinto b,*, A. Scheri b, F. Suggi-Liverani a
a
illycaffè S.p.A., Via Flavia, 110, 34147 Trieste, Italy
b
DINMA-Sezione di Fisica Tecnica, Università di Trieste, Via A. Valerio, 10, 34127 Trieste, Italy

a r t i c l e i n f o a b s t r a c t

Article history: The most common household coffee-brewing method in Italy makes use of a stove-top coffee maker
Received 20 December 2007 known as moka. This device uses the steam pressure, produced by the water contained in an auto-
Accepted 13 May 2008 clave-type aluminum kettle heated by an external source, to force upwards water itself through a roasted
Available online 24 May 2008
and ground coffee bed contained in a funnel-shaped filter. Despite its well-established usage, the moka
has never been the subject of detailed analysis, which led to a series of unclear descriptions or misinter-
Keywords: pretations concerning its functioning, such as the consolidated misbelieve that standard atmosphere boil-
Coffee
ing point temperature is needed to drive the water out. The detailed measurement of the
Moka
Brewing
thermodynamics of the moka, described here, sheds light on its actual behaviour. It is shown that extrac-
tion commences at pretty low temperatures and depends on the initial amount of dry air in the kettle.
Remarks on the time decreasing value of the coffee bed permeability are also drawn. A correct under-
standing of the extraction phenomenon, together with considerations on the coffee chemistry, serves
the purpose of assessing possible ways to improve the quality of moka product.
! 2008 Elsevier Ltd. All rights reserved.

1. Introduction tana, cafetera de rosca, cafetera de fuego or italiana and its use is
spread almost like in Italy [5–7], and Portugal where it is known
The most popular household coffee-brewing method in Italy is as cafeteira italiana.
that performed by using an inexpensive stove-top coffee maker in- This ingenious device uses the steam pressure, produced by the
vented by the aluminum technologist Alfonso Bialetti in 1933 [1,2]. water contained in an autoclave-type aluminum kettle heated by
This coffee maker was industrially produced and commercialized, an external source (gas or electrical stove), to force upwards the
by the Inventors son Renato from 1946, with the trademark same water through a roasted and ground coffee bed contained
denomination of ‘‘Moka Express”, but, nowadays, it is simply in a funnel-shaped filter. The beverage is conveyed through appro-
known as moka. In its original version, moka consists of two octag- priate tubing into an upper vessel, screwed and sealed by a rubber
onal conoids, which can be regarded as the very epitome of Italian gasket to the base kettle. The end of the brewing operation is usu-
household hardware, and in this version has racked up sales of ally announced by noisy mixture of boiling water and its vapour
more than 105 million units since market launch [1], with an ac- flowing from the upper tube, to indicate water depletion [8,9].
tual production of 4 million pieces per year [3]. During the Undoubtedly, a relevant part of the success of the moka coffee
1970s, the moka attracted the attention of several designers which maker has been played by the word ‘‘Express” in its trademark
reinvented the shape without remarkably affecting the overall pro- denomination. In facts this word evokes the worldwide well-
portion, and by the 1980s stainless steel started to parallel alumi- known espresso coffee brew, which is prepared by very different
num as moka construction material [4]. coffee machine and it is also organoleptically very different from
Due to its low cost and easy-to-handle characteristics, moka is moka coffee brew.
used, albeit not extensively, also in others countries where is also Italian espresso is a beverage prepared on request from roasted
known as stove-top espresso or often misnamed mocha or moca. and ground coffee beans by means of hot (90 ± 5 "C) water pressure
An exception is represented by Spain where it is known as napoli- (9 ± 2 bar) applied for a short time (30 ± 5 s) to a compact roast and
ground coffee cake (6.5 ± 1.5 g) by a percolation machine, to obtain
a small cup of a concentrated foamy elixir [8].
* Corresponding author. Tel.: +39 040 558 3503.
Unfortunately, the main factors controlling the coffee extraction
E-mail addresses: Luciano.Navarini@illy.it (L. Navarini), nobile@units.it (E.
Nobile), fpinto@units.it (F. Pinto), alessio.scheri@libero.it (A. Scheri), suggif@illy.it in the moka, such as the thermodynamic relationship between
(F. Suggi-Liverani). water pressure and temperature, the Darcys law of linear filtration

1359-4311/$ - see front matter ! 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2008.05.014
 L. Navarini et al. / Applied Thermal Engineering 29 (2009) 998–1004 999

[9] as well as the physico-chemical nature of roasted and ground

coffee, led to a beverage sometimes partially characterized by
harsh bitter flavour often describe as ‘‘burnt”, and by lack of the
foam layer typical of true Italian espresso coffee brew [8]. Differ-
ently from espresso coffee machine, the thermal balance of moka
is somewhat flimsy, being affected by several variables not easy
to control [8]. It has been suggested that the main feature, shared
by moka and espresso, is the fact that water wets the grounds
once through, increasing the extraction yield by fresh solvent
power [8].
It is clear that, in order to objectively interpret the differences
between moka and espresso brewing methods, it is necessary to
study in detail the moka functioning.
The moka coffee extraction physics, inspired from a primordial
washing machine known as lisciveuse [1], in turn derived from
the steam engine of a couple of centuries ago [4], have not been
the subject of detailed studies.
It has to be stressed out that, a part of a plethora of surprisingly
unconceivable, physically uncorrect or simply vague functioning
descriptions published even on scientific literature [5,6,10,11],
the only one paper dealing more correctly, although not thor-
oughly, with the moka physics has been published 74 years after
the moka invention [9].
In this work, we attempt to fill this gap by performing detailed
measurements on a standard commercial moka, in order to better Fig. 1. Coffee maker parts: (a) topper pot; (b) downstream filter plate; (c) washer;
understand its underlying physics and functioning characteristics. (d) funnel-shaped filter; (e) lower tank.

In addition, the availability of an experimental database, consti-

tutes a necessary requirement for the development and validation
of a mathematical model of the device. This, in turn, can be partic- input voltage data into desired physical quantities, by program-
ularly useful for parametric analysis and/or optimization studies. ming virtual instruments (VIs).
The present work, to the authors’ best knowledge, is the first
experimental attempt to investigate in detail the moka physics in 2.3. Temperature measurements
order to put in evidence misinterpretations or myths, and to assess
how this physics affects the beverage quality. In order to better understand the phenomenon of the steam
pressure coffee extraction, a series of temperature sensors have
2. Experimental setup been installed. Four probes have been used to infer temperature
at different points within the lower tank, where part of the hot
Among the different types of stove-top coffee makers available water turns into vapour, whose pressure supports the extraction.
on the market, MOKA EXPRESS#, produced by Bialetti Industrie These probes, numbered TI [0–3], are Chromel/Alumel thermocou-
S.p.A., Omegna(VB), Italy, is the most largely used household de- ples with U (insulated) hot junction and 1.6 mm inconel sheath.
vice. It is a stove-top aluminum coffee maker, made in different They are mounted in pairs on 2 bolts with 8 M thread, and screwed
sizes, and its three cups version has been used in the experiments. on the lower tank. Six probes, numbered TE [4–9], have been used
The coffee maker is composed of a 220 cm3 capacity lower tank, a to measure the external temperature of the tank and the pot, in or-
50 cm3 capacity funnel-shaped filter, a washer, a downstream filter der to assess the heating behaviour, and collect the most available
plate, and a topper pot, as shown in Fig. 1. data. They have been realized with Chromel/Alumel cable type GG-
30-KK, and they have been fixed to the device with an epossidic
bicomponent resin. Two more Chromel/Alumel thermocouples
2.1. Operative conditions
with insulated hot junction have been used to measure the tem-
perature of the aqueous extract (coffee) in the little column of
The experiments have been conducted for a standard usage of
the top pot.
the three cups moka, which is considered to be a 150 g of water fill-
A sketch of the coffee maker with a schematic view of probes
ing of the tank, and a 15 g of coffee filling of the funnel. The coffee
positioning is presented in Fig. 2a.
employed is a 100% Coffea arabica L. blend with a medium roasting
degree (total weight loss 16%), coarsely ground powder for stove-
2.4. Pressure probe
top coffee makers. An electrical stove has been used to heat the
coffee maker for two different values of heating power. Two series
The vapour–air mixture pressure in the lower tank has been
of 10 experiments at 400 W and 600 W have been made.
monitored by means of a Wheatstone bridge-based sensor, pro-
duced by Kulite. The model used, XTEL-190-100D, is a 0–7 bar
2.2. Data acquisition system (100 PSI) pressure range transducer, operating in differential mode,
with temperature compensation between 80 "C and 275 "C.
In order to collect the data from the test rig, a National Instru-
ments (NI) SCXI-1300 General-purpose voltage module has been 2.5. Mass flow measure
used; it is connected to a SCXI-1102B channel amplifier, and
mounted on a SCXI-1000 chassis. The chassis is connected to a For detecting the water level inside the tank at different height,
PC through a NI PCI-6221 data acquisition (DAQ) device. The soft- eight resistive circuits have been used. Their electrical scheme is
ware used is LabView version 7, which allow to directly process the sketched in Fig. 2b. Each circuit is composed of a 9 V DC generator
1000 L. Navarini et al. / Applied Thermal Engineering 29 (2009) 998–1004

a a 140
Temp TI0 3
120 Temp TI1
2.5
100 Temp TI2

Temperature [°C]
Temp TI3

Pressure [bar]
2
80 Pressure
1.5
60

40 1

20 0.5

0 0
0 50 100 150 200 250 300 350
time [s]

b 2.4
120 Temp TI0
Temp TI1
110 Temp TI2

Temperature [°C]
Temp TI3 1.9

Pressure [bar]
100
Pressure

90
1.4
b 80

70
0.9
200 250 300 350
time [s]

Fig. 3. In-tank temperature histories: (a) whole experiment; (b) detailed view of
the late phase of extraction.

the pressure rise is due to thermodynamic equilibrium between

water and its vapor in saturation conditions [11]. While the first
of these common believes might, at a first sight, be justified by
Fig. 2. Coffee maker sketch: (a) probes positioning; (b) water level detector scheme. Fig. 3, where a sensible pressure rise is perceived at about 90 "C
of the water, the second is clearly disproved. TI0 probe is in contact
with the bottom of the tank and senses the temperature of the
and a 7.5 MX resistance, and it is connected to the DAQ system. In- water layer adjacent to the wall. TI1 and TI2 probes are immersed
side the tank, the circuits are made of copper wires insulated with in water for most of the extraction time and give almost equivalent
high temperature silicone. The system senses a discontinuous values for the water temperature, apart from slight oscillations due
resistance variation as the tip of the wire gets out of the water. to convective plumes. On the other hand, TI3 probe, which is posi-
tioned at the top of the tank, measures the temperature of the air–
2.6. Heater vapour mixture. This temperature is considerably lower than the
water temperature, which indicates lack of thermodynamic equi-
The heating source used is a common 600 W electric cooker, librium during the extraction process.
whose temperature has been monitored and kept as steady as pos- Fig. 4 shows the temperature of the water TI2 inside the tank
sible during the experiments. and the eight measurements of water flowed. It reveals that, de-
spite the first impression, even the first convincement is wrong.
3. Results and discussion In fact, extraction commences at lower temperatures.
It is interesting to observe that the moka behaviour can be split
3.1. In-tank thermodynamic behaviour into two phases. Up to approximately 120 g of water flowed, the
lower tank air–vapour mixture and the evaporating water can be
Fig. 3 shows the pressure and temperature histories in the tank, considered a closed system, whose pressure, increased by sensible
for a representative experiment with a heating power of 400 W. heat and water evaporation, drives the extraction of the coffee. We
The temperatures in this figure are those obtained from the four name this phase regular extraction phase. In this phase liquid–solid
probles TI [0–3] that, as indicated in Fig. 2a, are positioned at dif- extraction occurs.
ferent heights. When the water level in the tank reaches the end of the funnel,
It is an article of faith, among stove-top coffee maker users, to there is a short-cut between external ambient and internal air–va-
think that standard atmosphere boiling point temperature is pour mixture, which no more drives in-tank water out of the tank.
needed to drive the water out of the tank [6], and to think that At this point, the remaining water undergoes intense evaporation.
 L. Navarini et al. / Applied Thermal Engineering 29 (2009) 998–1004 1001

140 Strombolian phase, corresponding to a vapour–liquid–solid

3 extraction, is difficult to study because of its complex thermody-
Temp TI2
120 namics, while more detailed considerations can be drawn from
Water flowed 2.5 the regular extraction phase.
100 Pressure Measured water level data have been fitted with an exponential

Pressure [bar]
2 regression for each experiment, as illustrated in Fig. 4. The function
80 used is
1.5
60 m ¼ "a þ aebs ; ð1Þ
Temperature [°

40 1 where m is the water flowed in grams and s is the time elapsed

from the beginning of the experiment. The water flow, m, _ is easily
20 0.5 obtained by deriving (1). The mean correlation coefficient between
measured data and regression model for all the experiments is
0 0 0.9963 and 0.9948, for 400 W and 600 W heating power,
0 50 100 150 200 250 300 350 respectively.
time [s]
The mean in-tank water temperature, T w , has been calculated:
Fig. 4. In-tank temperature, pressure, and water flowed.
R
Tm_ ds
Tw ¼ ð2Þ
_ ds
m
and is reported in Table 1, together with the initial and final extrac-
We name this phase, announced by a well-known rattling sound,
tion temperatures. Table 1 shows that the initial in-tank extraction
strombolian phase, because of its typical volcano-like behaviour.
temperatures are clearly below the misbelieved value of 100 "C,
Fig. 5 depicts the different phases during extraction.
with great part of the water flowing at quite low temperatures. In
High temperature extraction fluids (vapour, water and their
Table 1, the initial in-tank water temperature is considered at
mixture) transit in the coffee bed is noxious for the quality of the
10 g of water flowed, which is the first value sensed by the water
extract because, under these conditions, such fluids are more effi-
level measurement apparatus. Whereas the final in-tank water tem-
cient in solubilizing less soluble compounds, generally conferring
perature is taken at 120 g of water flowed, considered as the begin-
bitterness and astringency [12], and/or in stripping least volatile
ning of the strombolian phase.
aroma compounds which are organoleptically unpleasant and de-
At the beginning of the heating process, the tank has 20 cm3 of
scribed as clove-like, smoky, burnt, medicinal/chemical [13]. This
space occupied by air, which we may consider, for simplicity, at
is witnessed by an extraction yield (defined as the percentage of
saturated conditions. During the extraction, the pressure contribu-
the brew total solids with respect to ground and roasted coffee
tion due to air can be deduced by applying ideal gas law and the
dose) which is generally higher in comparison with the other
regression model for water flow. Pressure due to dry air is calcu-
brewing methods (e.g. filter, espresso, plunger or ‘‘French Press”).
lated as follows:
In particular extraction yield ranging from 18% to 22% have been
proposed as the most acceptable, as far as brews quality is con- TI3ðsÞ pðair;0Þ V 0
pair ðsÞ ¼ ; ð3Þ
cerned. The coffee brews below 16% are considered to be under-ex- V air ðsÞ T0
tracted and those above 24% are considered to be over-extracted
where p(air,0), V0, and T0 are the initial partial pressure, volume and
[13]. Independently on the coffee used (quality and quantity), val-
temperature of dry air, respectively, TI3(s) is the temperature mea-
ues from 27.59 ± 0.28% [7] to 31.9% [8] have been reported. In a
sured by the higher in-tank temperature probe, and V(s) is the vol-
comparison between moka and espresso coffee extraction methods,
ume occupied by air at a certain time s, which depends on Eq. (1):
the beverage concentration range suggested to be optimal for qual-
ity (> 2% for moka and > 3% for espresso) has been obtained by moka m
V air ðsÞ ¼ þ V 0; ð4Þ
operating under conditions of coffee dose and powder particle size qw
distribution leading to an extraction yield higher than 30% [14] and
where qw is the water density.
outside the optimal range suggested by the same authors (18–
It has been already stressed out the absence of thermodynamic
25%).
equilibrium between liquid and vapour phases of water, which re-
sults in a temperature difference sensed by probes whether im-
mersed or not. Vapour conditions are driven by both evaporation
from liquid–vapour separation surface and convective heat trans-
fer with each surrounding surface. A question arise on whether va-
pour is at saturated or overheated conditions, which is not possible
to answer precisely. Nevertheless, vapour formation and heating
can be considered driven mainly by evaporation. So, in order to
estimate the pressure due to water vapour, saturated vapour at
air–vapour temperature, TI3, rather than in-tank water tempera-
ture, has been assumed. Values are obtained by means of IAPWS

Table 1
Temperatures of in-tank water during extraction

Heat flux Initial Final Mean

Mean Std. Mean Std. Mean Std.
400 W 68.7 2.7 117.2 1.2 94.3 1.6
Fig. 5. extraction phases: (a) beginning; (b) regular extraction; (c) strombolian 600 W 70.2 2.9 120.6 3.0 97.6 1.2
extraction.
1002 L. Navarini et al. / Applied Thermal Engineering 29 (2009) 998–1004

140
measured pressure 3 a 140
Temp TI2 3
120 computed vapour pressure
computed air pressure 120 Temp TC10
2.5 Water flowed 2.5
100 computed (air + vapour) pressure
100 Pressure
Water flowed [g]

Pressure [bar]
2

Pressure [bar]
80 2
80
1.5 1.5
60 60

Temperature [°
40 1 1
40

20 0.5 0.5
20

0 0 0 0
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
time [s] time [s]
Fig. 6. Pressure contribution of dry air and saturated vapour.
b 130 3
Temp TI2
120 Temp TC10
IF-97 tables. Fig. 6 depicts a representative experiment, and it 2.5
Water flowed
shows that very good agreement exists between our assumptions 110
Pressure
and measured pressure. This reveals the major contribution of 2

Pressure [bar]
100
dry air in leading the extraction, and it will be the subject of further
detailed analysis. 90 1.5
Temperature [°
3.2. Aqueous extract 80
1
70
The funnel-shaped filter has 50 cm3 capacity and is filled with
0.5
15 g of coffee. Coffee true density is 1190 kg/m3 [15], thus the cof- 60
fee bed filling ratio is 0.244. The first drop of aqueous extract is
sensed by TC10 probe after the funnel has been completely filled 50 0
250 260 270 280 290 300
by water, completing the imbibition phase. This happens when an time [s]
approximative amount of 40 g of water has flowed out of the lower
tank. This can be noticed in Fig. 7, when TC10 probe experience a Fig. 7. (a) Aqueous extract temperature at the exit of the coffee bed; (b) detailed
view.
sudden temperature variation due to the contact with the aqueous
extract. In the imbibition phase no pressure drop is sensed, partly
because the water flow is low, and partly because in this phase the Table 2
coffee matrix presents low resistance to water penetration. During Temperatures of aqueous extract
imbibition and extraction phases the coffee undergoes chemical
Heat flux Initial Final Mean
transformations due to the interaction with water, which substan-
Mean Std. Mean Std. Mean Std.
tially chance its properties [16,17]. The coffee bed water invasion,
during the imbibition phase, induces the solubilization of more sol- 400 W 63.0 2.0 95.8 2.9 78.8 1.5
uble and low molecular weight compounds, as well as more vola- 600 W 61.8 2.5 97.7 2.4 80.5 1.3

tile aromatics (low temperature/pressure extraction).

Simultaneously, there is the coffee bed particle swelling, due to
the swelling of water-insoluble polysaccharides present in the A preliminary granulometric analysis of the coffee cake, after
roasted coffee [16], and with the geometrical rearrangement of the brewing process, reveals an almost uniform distribution, with
the coffee particles due to upwards water flow [18]. As soon as a variation in both average and medians particle size 9% and 14%,
the coffee bed swelling and spatial rearrangement provoke the pro- respectively, along the water path. This suggests a linear decay
gressive decrease of the coffee bed porosity, the extraction pro- assumption for pressure. Taking into consideration both conduc-
ceeds at increasing temperatures/pressures, thus making the tive and advective terms in the transport phenomenon, it can be
decrease in coffee bed porosity and the solubilization of coffee shown that the temperature profile in the cake is slightly concave
compounds competitive phenomena. The process goes on up to but, for simplicity, in the transit through the coffee bed, which in
the starting of the strombolian phase, which marks the passage our case is 21 mm thick, pressure and temperature profiles can
from a closed thermodynamic system to an open one. be considered linear with a good approximation. In Fig. 8 pressure
In its passage through the coffee bed, water transfers part of its and its saturation temperature are compared to the aqueous ex-
heat to the bed itself. Aqueous extract temperatures are sensibly tract temperature through the coffee bed at the end of the regular
lower than that of the in-tank water. This is clearly visible in Fig. extraction phase, where high in-tank pressure is present. During
7, where the extracted coffee has a much lower temperature, regular extraction, water temperature is always lower than satura-
TC10, than that, TI2, of the water in the tank. tion temperature, thus the risk of local evaporation in the bed is
After 120 g of water flowed the strombolian phase begins, and avoided.
no accurate measurements of the extraction phenomenon can be
made. As highlighted in Fig. 7b, a limited zone in between 50 g 3.3. Mass flow analysis
and 120 g of water flowed has been considered. In Table 2 the ini-
tial, final, and mean extract temperatures for the restricted zone Applying Darcy’s law, in [9] Gianino derives the permeability of
are presented. the coffee bed from an integral balance, assuming constant ther-
 L. Navarini et al. / Applied Thermal Engineering 29 (2009) 998–1004 1003

120 2 4. Conclusions
Temp measured
115 In this paper, an experimental study of a stove-top coffee ma-
Temp saturation 1.8
ker, known as moka, has been described. Despite its quite simple
Pressure manufacture and functioning, it has been shown that the thermo-
Temparature [°C]

110
1.6

Pressure [bar]
dynamic behaviour of the moka device is complex in comparison to
other coffee-brewing methods.
105
1.4 The brewing process of moka has been divided into two phases.
In the regular extraction phase liquid–solid extraction occurs,
100 which presents time-varying temperature and water flow rate. In
1.2
this phase extraction is driven by increasing air–vapour pressure
95 above the water level in the lower tank of the device. The pressure
1 increase is due not only to time increasing flow rate, but also to a
90 non constant rheological behaviour of the coffee cake, whose per-
0 5 10 15 20 meability decreases with time as the coffee undergoes chemical
coffee bed thickness [mm]
reactions, which in turn decrease its porosity. Moreover, the stove
Fig. 8. Temperature profiles in the coffee bed at 120 g of water flowed. heating power, which is usually constant during the process, ex-
ceeds the actual requirement in the final stages of the extraction,
when a little fraction of water is still in the tank and consequently
its heat capacity diminish, resulting in pressure and flow rate aug-
3 300
mentation. An analysis of pressure contributions has highlighted
Water flow the role played by dry air in the overall phenomenon, which is
2.5 ∆P 250 not negligible as believed by many. The quantity of dry air can
∆ P [bar]

Permeability influence both temperature and flow rate, thus affecting final ex-
Permeability [mD]

2 tract quality, and it is meant to be the subject of further studies.

200
When water level reaches the end of the funnel, the short-cut be-
1.5 tween external ambient and air–vapour mixture, which no more
150 drives in-tank water out of the tank, causes an intense evaporation,
1 named strombolian phase. In this phase vapour–liquid–solid extrac-
100 tion occurs, with consequent extraction of soluble compounds
0.5 which are generally noxious for the quality of the final product.
The higher the pressure and temperature, the higher the extraction
50
of undesired components.
0
250 260 270 280 290 The detailed measurement of the thermodynamic behaviour of
time [s] the moka which, to the authors’ best knowledge, is the first solid
experimental attempt of investigation, serves the purpose of an
Fig. 9. Time-varying permeability profile.
intimate understanding of such a popular, yet mysterious, device,
which so much diverges from other coffee-brewing methods, in or-
mophysical properties for water and coffee powder. The value der to assess possible ways to improve the quality of its product.
Gianino finds is 2300 millidarcy1 [mD]. We will show that this is
a way too rough approximation. References
Again we will consider the limited zone in between the imbibi-
tion and strombolian phases (50 g and 120 g of water flowed), [1] D. Samarelli, Omegna, paese di pentole e caffettiere: La pentola a pressione
Lagostina e la Moka Express Bialetti, Quaderni della Biblioteca,
where sensible pressure data are obtained, and measurements of Amministrazione Comunale, Omegna (VB), Italy, 1990 (in Italian).
the water flow are possible. Darcy’s law states: [2] Bialetti, I. segreti del Caffè, DE & CO Immagine e Comunicazione, Milano, Italy,
1995. (in Italian).
j DP [3] J. Myron, The story of the bialetti moka express, 2007. <http://
q¼" ; ð5Þ
l L www.ineedcoffee.com>.
[4] S. Gronert, The 9090 Cafetière by Richard Sapper, Verlag form GmbH, Frankfurt
where q ¼ m=ð
_ qAÞ is the water volumetric specific flow rate, j is the am Main, 1997.
[5] A. Peters, Brewing makes the difference, in: Proceedings of the 14th ASIC
permeability of the coffee cake, l is the dynamic viscosity of water, Colloquium (San Francisco), ASIC, Paris, France, 1991, pp. 97–106.
DP is the pressure drop experienced during filtration, and L is the [6] P. Parras, M. Martinez-Tom, A. Jimenez, M. Murcia, Antioxidant capacity of
thickness of the bed. During filtration, aromatic substances solve coffees of several origins brewed following three different procedures, Food
Chem. 102 (2007) 582–592.
into water, thus changing its rheological properties. Nevertheless,
[7] I. Lopez-Galilea, M. Paz De Pena, C. Cid, Correlation of selected constituents
considering pure water as reference point, a time-varying value of with the total antioxidant capacity of coffee beverages: influence of the
j can be obtained. Fig. 9 depicts the time-varying permeability va- brewing procedure, J. Agric. Food Chem. 55 (15) (2007) 6110–6117.
[8] M. Petracco, Beverage preparation: brewing trends for the new millennium, in:
lue for a representative experiment, where the pressure drop has
R. Clarke, O. Vitzthum (Eds.), Coffee: Recent Developments, Blackwell Science,
been taken neglecting the head and the friction losses, which give Oxford, 2001.
minimal contribution. The value of j, as expected, gradually decays [9] C. Gianino, Experimental analysis of the Italian coffee pot ‘‘moka”, Am. J. Phys.
to an asymptotic condition, which confirms former experiments on 75 (1) (2007) 43–47.
[10] R. Clarke, R. Macrae (Eds.), Coffee, Technology, vol. 2, Elsevier Applied Science,
espresso machines, described in [19]. From Fig. 9 it is evident that London, 1989.
typical ranges for j in the regular extraction region is 70–400 mD, [11] A. Varlamov, G. Balestrino, La fisica di un buon caffè, Il Nuovo Saggiatore 17
which is more than 10 times lower than the value obtained in [9]. (3–4) (2001) 59–66. (in Italian).
[12] M. Petracco, in: A. Illy, R. Viani (Eds.), Espresso Coffee. The Science of Quality,
second ed., Elsevier Academic Press, Oxford, 2005.
[13] T. Lingle, Coffee brewing control chart, in: T. Lingle (Ed.), The Coffee Cuppers’
Handbook. A Systematic Guide to the Sensory Evaluation of Coffee’s Flavour,
1
1 millidarcy = 1 & 10"15 m2. Specialty Coffee Association of America, Long Beach, 2001.
1004 L. Navarini et al. / Applied Thermal Engineering 29 (2009) 998–1004

[14] C. Severini, S. Giuliani, G. Pinnavaia, Survey on the different methods of coffee [17] O. Fond, Effect of water and coffee acidity on extraction. Dynamics of coffee
extraction, Industrie delle Bevande 22 (1994) 227–230. (in Italian). bed compaction in espresso type extraction, in: Proceedings of the 16th ASIC
[15] P. Singh, R. Singh, S. Bhamidipati, S. Singh, P. Barone, Thermophysical Colloquium, Kyoto (Japan), Association Scientifique Internationale du Café,
properties of fresh and roasted coffee powders, J. Food Process Eng. 20 (1) Paris (France), 1995, pp. 413–421.
(1997) 3150. [18] L. Odello, Moka: l’altro volto del caffé made in Italy, L’Assaggio 19 (2007) 43–
[16] D. Rivetti, L. Navarini, R. Cappuccio, A. Abatangelo, F. Suggi Liverani, Effect of 49. (in Italian).
water composition and water treatment on espresso coffee percolation, in: [19] G. Baldini, Filtrazione non lineare di un fluido attraverso un mezzo poroso
Electronic Proceedings (CD-ROM) of the 19th ASIC Colloquium, Trieste (Italy), deformabile, Thesis, University of Florence, 1992 (in Italian).
Association Scientifique Internationale du Café, Paris (France), 2001.