Available online at www.sciencedirect.

com

Journal of Food Engineering 86 (2008) 19

www.elsevier.com/locate/jfoodeng

Modeling heat and mass transfer during drying of green coee

beans using prolate spheroidal geometry
W.N. Hernandez-Daz, I.I. Ruiz-Lopez, M.A. Salgado-Cervantes, G.C. Rodrguez-Jimenes,
M.A. Garca-Alvarado *
Chemical and Biochemical Engineering Department, Instituto Tecnologico de Veracruz, Av. Miguel Angel de Quevedo 2779, 91897 Veracruz, Ver., Mexico

Received 3 April 2007; received in revised form 19 August 2007; accepted 30 August 2007
Available online 25 September 2007

Abstract

A heat and mass transfer model in prolate spheroidal coordinates system was proposed to describe the drying of green coee beans.
The model describes the 3D moisture and temperature proles inside the bean. The results were integrated over volume in order to obtain
a drying kinetic equation for prolate spheroidal geometry. The average eective diusivity of water as function of temperature and mois-
ture was estimated at 45 and 60 C by slope method from experimental drying kinetic of green coee beans. The expression obtained
applied to drying kinetic equation reproduced approximately the experimental behavior.
2007 Elsevier Ltd. All rights reserved.

Keywords: Green coee drying; Diusivity; Prolate spheroidal geometry

1. Introduction There are two types of coee processing methods: dry

and wet. When the coee is wet processed, the cherries
There are critical points concerning quality and microbi- are pulped and fermented and only the beans are dried
ological safety in the drying of coee beans. Several (Suarez-Quiroz et al., 2004b). No reports for the drying
researchers have reported Ochratoxin A mycotoxin con- of green coee beans were found. Heat and mass transfer
tamination during drying (Bucheli, Kanchanomai, Meyer, of the coee cherries subject to the dry method of process-
& Pittet, 2000; Paulino de Morales & Luchese, 2003; ing have been reported (Perez-Alegra & Ciro-Velasquez,
Suarez-Quiroz et al., 2004a). Contamination risk is greater 2001; Sfredo, Finzer, & Limaverde, 2005; Varadharaju,
during sun drying due to the bean-soil contact and bean re- Karunanidhi, & Kailappan, 2001). Varadharaju et al.
humidication under high relative humidity conditions. (2001) and Sfredo et al. (2005) modeled the heat and mass
These problems could be eliminated with articial drying. transfer in bed dryers with analytical solutions of Ficks
However, articial drying is a high energy demanding oper- and Fouriers laws at constant air conditions; nevertheless,
ation, and it has been shown (Sfredo et al., 2002) that if the air conditions do not remain constant during the bed
beans temperature exceeds 45 C during drying, the cof- drying process. Perez-Alegra and Ciro-Velasquez (2001)
fees quality is harmed. Then, an adequate heat and mass recognized this variation and modeled the coee cherry
transfer representation of coee drying is required in order drying with a set of four dierential equations. Other
to state optimal operation conditions that guarantee mini- models, similar to Perez-Alegra and Ciro-Velasquez
mal energy consumption and maximizes beans quality. (2001), have been reported for bean dryers: Henderson
and Pabis (1961); Henderson and Pabis (1962); Spencer
(1969); Barre, Baughman, and Hamdy (1971); Bruce and
*
Corresponding author. Tel.: +52 229 934 5701. Giner (1993); Barrozo, Murata, and Costa (1998); Torrez,
E-mail address: [email protected] (M.A. Garca-Alvarado). Gustafsson, Schreil, and Martinez (1998); Herman-Lara,

0260-8774/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2007.08.025
2 W.N. Hernandez-Daz et al. / Journal of Food Engineering 86 (2008) 19

Nomenclature

a focal distance (m) Dimensionless groups

a0, a1, a2 parameters of Eq. (21) (m2 m3) j dimensionless conductivity
ab specic surface area between phases (m2 m3) P dimensionless group
aw water activity (m2 m3) h dimensionless latent heat
Abc surface area between phases (m2) W2 dimensionless temperature
Cp specic heat of coee (J g1 K1) Bim mass Biot number
e unit vector (m) Bi Biot number
Db water eective diusivity in coee (m2 s) g1 dimensionless density
h heat transfer coecient (W m1 K1) g2 dimensionless density and specic heat
kb thermal conductivity of coee (W m2 K1) /1 dimensionless diusivity
kc internal mass transfer coecient (m s1) W1 dimensionless moisture
Keq distribution constant between phases s Fourier number
LD characteristic path length for diusion (m) /2 Luikov number
mexp experimental slope of moisture kinetic (g Re Reynolds number
water (g dry matter)1 s1) Sc Schmidt number
nbc unit vector from b to c
N number of nodes in numerical solution Subscripts
p total or partial pressure (Pa) 0 indicate a reference value
R upper limit for u domain avg indicate average
t time (s) d indicate a dry matter basis
T temperature (K) e indicate at equilibrium
X water content (g water (g dry matter)1) i indicate at the interface
u spheroidal prolate coordinate b indicate solid phase (coee)
x, y cartesian coordinates c indicate gas phase (air)

Greek Symbols Superscripts

DHwv latent heat of water evaporation (J g1) i indicate a node in the prolate
u, B spheroidal prolate coordinates j indicate a node in the prolate
q density (kg m3) k indicate a node in the prolate

Salgado-Cervantes, and Garca-Alvarado (2005). This last 2. Mathematical formulation

mathematical representation (Herman-Lara et al., 2005) is
particularly useful because it is built from heat and mass 2.1. Model building
transfer properties of both phases involved, and the dis-
continuous phase (dried product) properties can be Heat and mass transfer during the drying of a product
deduced from its water eective diusivity and its thermal (phase b) in constant conditions of gas ow (phase c) is
conductivity. This representation avoids any assumption represented in dimensionless form by Ruiz-Lopez, Cordo-
of which is the heat or mass transfer controlling mecha- va, Rodrguez-Jimenes, and Garca-Alvarado (2004),
nism. The application of Herman-Lara et al. (2005) model
to coee beans drying requires the evaluation of water og1 W1 og
m1 1 r  /1 r g1 W1 m1 g1  1
eective diusivity and thermal conductivity in the beans os os
in a representative geometry. Thermal conductivity and  Bim
 nbc  /1 r g1 W1 m1 g1 X ci  X c 2
other thermal properties (like specic heat) have been X b0  X be
reported (Perez-Alegra & Ciro-Velasquez, 2001; Perez- og2 W2 og
m2 2 r /2 r W2  3
Alegra, Ciro, & Abud, 2001). Therefore in this study the os os
heat and mass transfer during green coee beans drying  nbc  jr W2i BiW2  Pnbc  hr g1 W1 m1 g1 4
at constant air conditions were modeled with rigorous
mathematical representation that approximates the coee where
beans shape: a half prolate spheroidal geometry. The 3D
Db0 t X b  X be Tb  Tc
results were integrated over volume in order to obtain a s W1 W2
drying kinetic equation valid for this geometry, and the L2D X b0  X be T b0  T c
water average diusivity was calculated by slope method k b0
/2
over experimental results. qb0 Cpb0 Db
 W.N. Hernandez-Daz et al. / Journal of Food Engineering 86 (2008) 19 3

Db DH wv Db0 qdb0 X b0  X be qdb at Abc

/1 P g1
Db0 k b0 T b0  T c qdb0 where
qb Cpb hc L D qdc k cc LD K eq kb
g2 Bi Bim j gu; 1 sinh2 uR sin2 1 11
qb0 Cpb 0 k b0 qdb0 Db0 k b0
T c0 X be DH wv Db and
m2 m1 h
T b0  T c0 X b0  X be DH wv0 Db0 LD aR 12
r LD r
In Fig. 1, the half prolate spheroidal coordinate system in a
and with the equilibrium conditions, coee bean is represented. This system is symmetric in B and
X ci f X bi at Abc 5 u. Therefore the problem can be solved only using a quarter
of the dominion, that is for 0 6 u 6 1, 0 6 B 6 p/2 and
T bi T ci at Abc 6 0 6 u 6 p/2. Then, the boundary conditions ((9) and (10))
Eqs. (1)(4) must be written in a coordinate system that must be written as
represents the heat and mass transfer of the medium shape
(coee beans). Half prolate geometry (Spiegel, 1968) was /1 oW1 Bim
 p X ci  X c 13
selected because of its close resemblance with the products gu; 1 ou K eq X b0  X be
3D form. Therefore Eqs. (1)(4) must be represented in j oW2 Ph oW1
prolate spheroidal coordinates (0 6 u 6 1, 0 6 B 6 p,  p p BiW2 14
gu; 1 ou gu; 1 ou
0 6 u 6 2p). This representation requires the three coordi-
nates. Therefore, some simplications are convenient. at u = 1, 0 6 B 6 p/2 and 0 6 u 6 p/2
Assuming constant density (g1 = 1), constant eective dif- /1 R oW1 Bim
fusivity (/1 = 1), and constant thermal properties (g2 = 1, X ci  X c 15
sinhuR sin1 ou X b0  X be
j = 1, h = 1). Eqs. (1)(4) can be written, based on spheroi-
jR oW2 PhR oW1
dal prolate geometry as BiW2 16
  sinhuR sin1 ou sinhuR sin1 ou
oW1 1 o oW1
sinhuR
os sinhuRgu; 1 ou ou at u = 0, 0 6 u 6 1, and 0 6 B 6 p/2.
2   The symmetry of the system produces,
R o oW1
sin1
sin1gu; 1 o1 o1 Wi u; 1; u Wi u; p  1; p  u for i 1 and 2
2 2
R o w1
2 2
7
sinh uR sin 1 ou2 2.2. Solution by RungeKutta method
 
oW2 /2 =g2 o oW2
sinhuR Eqs. (7) and (8) were solved by using the method of lines
os sinhuRgu; 1 ou ou
  described in detail in Ruiz-Lopez et al. (2004). They were
/2 R2 =g2 o oW2 rewritten in terms of nite dierences for the derivatives
sin1
sin1gu; 1 o1 o1 of spatial coordinates. Example: Eq. (7) was rewritten as
/2 R2 =g2 o2 W 2
8 Wi1;j;k Wi;j;k
2 2
sinh uR sin 1 ou2 dWi;j;k
1 sinhui1=2 R 1 Du 1

ds sinhui Rgui ; 1j Du
for 0 6 u 6 1, 0 6 B 6 p, 0 6 u 6 p (for dene a half of the i;j;k i1;j;k
W W
prolate spheroid). sinhui1=2 R 1 Du 1

/1 oW1 /1 R oW1 sinhui Rgui ; 1j Du
 nbc  p
eu  nbc  p
e1
gu;1 ou gu; 1 o1 Wi;j1;k W1i;j;k
R2 sin1j1=2 1
D1
/1 R oW1 Bim
 nbc  eu X ci  X c sin1j gui ; 1j D1
sinhuRsin1 ou K eq X bo  X be i;j;k i;j1;k
W1 W1
9 R2 sin1j1=2 D1

j oW2 jR oW2 sin1j gui ; 1j D1
 nbc  p eu  nbc  p e1
gu;1 ou gu; 1 o1 W1i;j;k1 Wi;j;k
R2 Du
1

jR oW2 Ph oW1
 nbc  eu nbc  p eu sinh2 u R sin 1 Du
i 2 j
sinhuRsin1 ou gu; 1 ou i;j;k i;j;k1
W1 W1
PhR oW1 PhR oW1 R2 Du
nbc  p e1 nbc  eu BiW2 17
gu; 1 o1 sinhuR sin1 ou sinh2 u R sin 1j Du
i 2

10 and Eq. (13) was rewritten as

4 W.N. Hernandez-Daz et al. / Journal of Food Engineering 86 (2008) 19

Fig. 1. Spheroidal prolate coordinate system representing a coee bean. (a) General representation. (b) A plane at u constant. (c) A plane at u constant.
(d) Two planes at B constant.

/1 W1N 1;j;k  W1N 1;j;k 3. Experimental and parameter estimation

 p
guN ; 1j 2Du
3.1. Experimental
Bim
X ci  X c 18
K eq X b0  X be Fermented washed coee beans were obtained from a
Eq. (17) is applied for i = 1, 2, . . ., N, B = 1, 2, . . ., N  1, local green coee producer (Huatusco, Veracruz, Mexico).
and u = 0, 1, . . ., N  1. Eq. (18) was substituted in Eq. The beans size was measured with a Vernier. The dimen-
(17) at i = N. The same scheme was applied to Eqs. (8), sions were characterized in agreement with coordinate sys-
(14), (15) and (16). The resulting set of N(N  1)(N  1) tem used, that is, the focal distance a was calculated in
ordinary dierential equations was solved by the fourth order that the half prolate spheroidal coordinates approx-
order RungeKutta method. imate the coee beans shape with R = 1. A typical green
coee bean represented in prolate spheroidal coordinates
2.3. Average moisture equation is shown in Fig. 1. The beans shrunk approximately 25%
during drying and therefore an average size was reported.
Moisture 3D proles obtained from the model solution The beans initial moisture content was evaluated according
were integrated over volume in order to obtain a macro- to the AOAC (1990) method No. 22.013 at 60 C and
scopic drying kinetic equation for average moisture content 13.3 kPa in a vacuum stove. The beans bulk density was
in coee beans. The resulting volume, calculated in agree- calculated by water displacement in a graduate tube. The
ment with the volume integral for curvilinear coordinate specic surface area was calculated with the numbers of
systems (Spiegel, 1968) is beans in the drying bed and with surface area and volume
R p=2 R p=2 R 1 equations for a prolate spheroid. The results are listed in
W1;i;j;k a3 gu; f sinhuR sinf du d1 du Table 1.
W1avg 0 R p=20 R p=20 R 1
a3 gu; f sinhuR sinf du d1 du A plant pilot xed bed dryer with transversal air ow of
0 0 0
02 m s1, 80 C of maximum temperature and with a
19
0.0024 m2 transversal section drying chamber, was used
Eq. (19) was solved by 3D trapezoidal rule. for the experiments. A monolayer of the characterized fer-
 W.N. Hernandez-Daz et al. / Journal of Food Engineering 86 (2008) 19 5

Table 1 where the water activity was obtained from experimental

Experimental variables reports for coee (Iglesias & Chirife, 1982) and tted by
Treatment Variable linear regression to the following equation:
1 Xb0 = 0.6 kg (kg dry matter)1, LD = 0.0041 m,  
Xc = 0.019 kg (kg dry air)1, Tb0 = 34 C, Tc = 61 C, aw 1  exp a0 T ab1 X ab2 21
ab = 891 m2 m3, Xeq = 0.0003 kg (kg dry matter)1
The parameters for Eq. (21) and other thermophysical
2 Xb0 = 1.2 kg (kg dry matter)1, LD = 0.0036 m,
Xc = 0.020 kg (kg dry air)1, Tb0 = 27 C, Tc = 60 C
properties of coee are listed in Table 2. The latent heat
ab = 901 m2 m3, Xeq = 0.0004 kg (kg dry matter)1 of water evaporation showed in Table 2 was assumed equal
to that of pure water. Eective diusivity was calculated
3 Xb0 = 0.6 kg (kg dry matter)1, LD = 0.0051 m,
Xc = 0.018 kg (kg dry air)1, Tb0 = 34 C, Tc = 41 C from drying experiments as described in the following
ab = 883 m2 m3, Xeq = 0.001 kg (kg dry matter)1 section.
4 Xb0 = 1.2 kg (kg dry matter)1, LD = 0.0043 m,
Xc = 0.019 kg (kg dry air)1, Tb0 = 27 C, Tc = 43 C (rst 4. Results
hour),
52 C (the rest of the process) The numerical solution of Eqs. (7)(16), resulted in a 3D
ab = 896 m2 m3, Xeq = 0.0005 kg (kg dry matter)1 description of the dimensionless moisture and temperature
14 P = 0.13, Bi = 6, Bim > 1000, qb0 = 983 kg m3, R = 1 proles inside the coee bean during drying. Figs. 25 show
the dimensionless moisture and temperature proles
obtained at u = p/2 in the plane x y (x = a cos h (u) cos (f),
mented coee beans, 1820 beans with maximum size devi- y = a sin h (u) sin (f) sin (u)) for two dierent Fourier num-
ations of 20%, was placed in the drying chamber and dried bers (0.15 and 0.4). The moisture proles observed in Figs.
with a cross ow of air at 1.6 m s1, so that, the drying was 2 and 4 were expected (the geometric center of the bean has
carried out at constant air conditions. The drying experi- the highest moisture content), but these results allowed us
ments were developed at the conditions listed in Table 1, to predict the moisture at the surface, where the bean is
with two replicates for treatment. The moisture evolution more susceptible to mycotoxin production. Another eect
was calculated from weight losses during drying. The tem- that can be predicted, is the fact that the bean corners
perature evolution was measured with a K type thermocou- dry faster, making them more susceptible to thermo-
ple inserted into a coee bean. mechanical stress and to suer cracks and deformations,
as was reported by Lima, Queiroz, and Nebra (2002). Figs.
3.2. Parameter estimation 3 and 5 show that a small zone on the at face of the bean,
near the corner, is the rst to equilibrate its temperature
External heat and mass transfer properties were calcu- with the drying air. This is not predictable by intuition
lated with reported correlations for uid ow around par- due to the eect of complex phenomenon that occurs in
ticles (Geankoplis, 1978). Interface condition (Eq. (5)) is the interface, as described by Eqs. (18) and (20). All of
represented by these eects can be predicted from a single solution of
Eqs. (11)(20) by its dimensionless nature. The moisture
aw pw =p 18 and temperature as a function of time can be calculated
X ci ; 20
1  aw pw =p 29 with the corresponding values of water average diusivity.

Table 2
Thermophysical properties used for solving Eqs. (11)(20)
Properties Value or expression Reference
aw a0 = 100.103, a1 = 17.75, a2 = 0.88 Fitting Eq. (21) to Iglesias and Chirife (1982) data
kb (W m1 s1) 0:49  0:443e0:206X b Perez-Alegra et al. (2001)
kc (Wm1 s1) 8.4044  105T + 4.63  105, T in K Geankoplis (1978)
DHw0 (J kg1) 2.501  106 Geankoplis (1978)
C pc (J kg1 K1) 1000.0 Geankoplis (1978)
X
C pb (J kg1 K1) 1652:2 5835 1Xb b Perez-Alegra et al. (2001)
C pwv (J kg1 K1) 1608.92 Geankoplis (1978)
C pw (J kg1 K1) 4185  1=2 Geankoplis (1978)
1:86107 T 3=2 1 1
Dwc pr2 XD;wc
wc Mw Mc ChapmanEnskog equation
M = molecular weight Geankoplis (1978)
r2wc = Collision average diameter (in Tables) Tables in Bird et al. (1960)
XD,wc = collision integral (in Tables)
DHw (J kg1) 2.501  106  2.26  103T  1.7T2, T in C Geankoplis (1978)
qc (kg m3) 352.961/T, T in K Ideal gas law
lc (Pas) 5.87  106 + 4.25  108T, T in K Geankoplis (1978)
6 W.N. Hernandez-Daz et al. / Journal of Food Engineering 86 (2008) 19

Fig. 2. Dimensionless moisture proles inside the half prolate spheroid at Fig. 5. Dimensionless temperature proles inside the half prolate spheroid
u = p/2 and s = 0.15. at u = p/2 and s = 0.4.

Moisture and temperature proles have the use dis-

cussed in the above paragraph. However, in order to apply
bed dryer equations, like those proposed by Perez-Alegra
and Ciro-Velasquez (2001) or Herman-Lara et al. (2005),
it is required a drying kinetic equation. Since the earliest
reports on drying of solids (Sherwood, 1929a; Sherwood,
1929b; Sherwood, 1932) until the most recent studies (Hay-
aloglu, Karabulut, Alpaslan, & Kelbaliyev, 2007; Jam-
radloedluk, Nathakaranakule, Soponronnarit, &
Prachayawarakorn, 2007; Kashaninejad, Mortazavi,
Safekordi, & Tabil, 2007), it is a common practice repre-
sent the drying kinetic with an exponential equation
obtained from an integrated analytical solution of Ficks
second law equation at a given geometry, or a simple
empirical exponential relation, like Pelegs model. In the
case of 1D geometry the integrated analytical solution of
Fig. 3. Dimensionless temperature proles inside the half prolate spheroid
Ficks equation is a convergent exponential series. This ser-
at u = p/2 and s = 0.15. ies with only the rst term is
 2 
X bavg  X be 8 p Db
2 exp  t 22
X b0  X be p 4 L2D
Eq. (22) is the limit of Sherwood equation valid for drying
of solids with 1D geometry and for W1avg < 0.7 (Sherwood,
1929a; Sherwood, 1932).
In the case of 3D prolate spheroidal geometry there is
not reported analytical solution, but from the numerical
solution may be used for the deduction of an integrated
equation like (22). Simulated integrated moisture (W1avg)
is plotted in Fig. 6. This plot is unique for dierent temper-
atures and air velocities because the Bim is greater than
1000 in all cases (Table 1), and therefore the drying is dif-
fusion controlled and the temperature only aects the
water diusivity that is implicit in the Fourier number
(s). A dierent eect can be observed in the case of temper-
ature (Fig. 7). In this case, like the Biot number is lower
than one (Table 2), two dierent behaviors were obtained.
Fig. 4. Dimensionless moisture proles inside the half prolate spheroid at The integrated moisture kinetic (Fig. 6) shows a zone of
u = p/2and s = 0.4. linear behavior in semi-log scale. This linear zone is the
 W.N. Hernandez-Daz et al. / Journal of Food Engineering 86 (2008) 19 7

1D rectangular geometry reported by Herman-Lara et al.

(2005)
p2 Db
k cb 27
4L2D =LD
Therefore Eq. (25) is a dierential equation that represents
the average moisture kinetic during drying of a half prolate
spheroid. This procedure is common in beans drying design
(Henderson & Pabis, 1961; Henderson & Pabis, 1962;
Spencer, 1969; Barre et al., 1971; Bruce & Giner, 1993;
Barrozo et al., 1998; Torrez et al., 1998; Perez-Alegra &
Ciro-Velasquez, 2001), but in the case of Eq. (26) the geom-
etry is taking in account.
Fig. 6. Dimensionless average moisture simulation of a half prolate
spheroid at 45 C and 60 C. 4.1. Water diusivity estimation

From Eq. (23) the eective water diusivity was calcu-

lated with experimental kinetics of treatments 13 of Table
1 (treatment four was used for validation), and using the
slope method. The slope method is a very common way
to estimate eective diusivity during drying and is
described with detail in Zogzas and Maroulis (1996). The
experimental slope (mexp) was calculated two by two points
in each one of the drying kinetics, and tacked as
12:3Db
mexp  28
L2D
All of the diusivities obtained were tted to an Arrhenius
type model with exponential dependence for moisture. Due
to the bean size, the thermocouples gave inconsistent mea-
Fig. 7. Dimensionless central temperature simulation of a half prolate surements. Then the temperature was calculated from the
spheroid at 45 C and 60 C. data of Fig. 7. The result for m2 s1 (and T in K) was
 
66856:4 X bavg
Db exp 2:7085  1:74 29
RT X b0
limit of a kinetic equation, similar to (22), valid for a half
prolate spheroid. The slope of this line is 12.3 and the ori- The parameters of Eq. (29) have statistic signicance
gin intercept is 0.78. Therefore the integrated equation is (p < 0.05) with a non-linear test developed in Matlab 7.0.
  The moisture dependence in Eq. (29) indicates that the
X bavg  X be Db t
0:78 exp 12:3 2 23 water diusivity is aected by the structural changes during
X b0  X be LD drying instead of the absolute moisture of the bean. Eq.
Taking the derivative (29) evaluated with Xbavg/Xb0 = 0.4 and T = 57 C gave a
result of Db = 7.90  1010 m2 s1, which is in the order
dX bavg 12:3Db
 X bavg  X be 24 of the reported ones for coee cherries by Sfredo et al.
dt L2D (2005).
Like drying was carried out as diusion controlled the In the case of heat transfer, Fig. 7 does not show a well
equilibrium moisture is equivalent to interface moisture. dened linear zone like Fig. 6, mainly because the process
Therefore, Eq. (24) may be expressed as is not conduction controlled (Bi < 1) and there is a simulta-
neous use of heat in conduction and water evaporation.
dX bavg
k cb ab X bavg  X bi 25 However, assuming that the process would be conduction
dt controlled, and if evaporation would not take place Eqs.
and therefore the macroscopic water mass transfer coe- (3) and (4) would be mathematically similar to Eqs. (1)
cient in coee beans is and (2), and therefore the internal heat transfer coecients
12:3Db (which are independent on surface Bi and surface water
k cb 26 evaporation) may be dened as
L2D ab
12:3k b
which is the analogy of internal mass transfer coecient for hb 30
a half prolate spheroid with respect to the coecient in a L2D ab
8 W.N. Hernandez-Daz et al. / Journal of Food Engineering 86 (2008) 19

In order to validate our results, the obtained heat and changed. The temperatures show the inconsistencies dis-
mass transfer coecients (hb and kcb) were introduced in cussed above, but the general tendency is predicted. These
the following coee bean xed bed system, which was results show that the internal heat and mass transfer coef-
solved under constant air conditions, cients (Eqs. (26) and (30)) approximately represent the
dqb X bavg phenomena involved, and therefore could be generalized
k cb qb ab X bavg  X bi 31 to other conditions, like multi-layer bed drying.
dt
dqb C pb T bavg
hb ab T bavg  T i 32 5. Conclusions
dt
k cc qc X ci  X c k cb qb X bavg  X i 33 The nite dierence equation for mass transfer 3D
hc T c  T i hb T i  T bavg  k cb qb X bavg  X i DH 34 model of Eqs. (7)(12) provides valuable information
about the moisture distribution proles inside the grain
The simulation results obtained with Eqs. (26), (29) and during drying. Thus, allowing us to predict the zones more
(30)(34) and the variables and conditions listed in Tables 1 susceptible to contamination or to mechanical stress and
and 2 are plotted in Figs. 8 and 9 jointly with experimental cracking. The integrated drying kinetic equation, along
data. The simulated behavior approximates the experimen- with the estimated water eective diusivity, reproduced
tal one, taking into account that simulated kinetics are not the experimental drying kinetics of a monolayer xed bed
individually tted, and the treatment 4 was not used for of coee beans. This model may be used in the optimal
diusivity calculation. Some deviations were observed but design of coee dryers.
it is important to note that Eqs. (31)(34) can reproduce
the tendency of drying curves when the air temperature is Acknowledgments

The authors wish to thank the Mexican Consejo Nac-

ional de Ciencia y Tecnologa (CONACyT) and the Direc-
cion General de Educacion Superior Tecnologica
(DGEST) for the nancial support given through the pro-
ject PCP Cafe and UR513.

References

AOAC, (1990). Ocial Methods of Analysis (method No. 22.013).

Washington, DC: Association of Ocial Analytical Chemist.
Barre, H. J., Baughman, G. R., & Hamdy, M. Y. (1971). Application of
the logarithmic model to cross-ow deep-bed grain drying. Transaction
of the ASAE, 14, 10611064.
Barrozo, M. A. S., Murata, V. V., & Costa, S. M. (1998). The drying of
soybean seed in countercurrent and concurrent moving bed dryers.
Fig. 8. Experimental drying kinetics for average moisture of coee beans Drying Technology, 16, 20332047.
at treatments 1 (s), 2 (h), 3 (+), and 4 () of Table 1, and simulated Bird, R. B., Steward, W. E., & Lightfoot, E. N. (1960). Transport
moisture kinetics (continuous line) with Eqs. (31)(34). phenomena. New York: John Wiley & Son.
Bruce, D. M., & Giner, S. A. (1993). Mathematical modeling of grain
drying in counter-ow beds: Investigation of crossover of air and grain
temperatures. Journal of Agricultural Engineering Research, 55,
143161.
Bucheli, P., Kanchanomai, C., Meyer, I., & Pittet, A. (2000). Development
of Ochratoxin A during Robusta (Coea canephjora) Coee Cherry
Dying. Journal of Food Chemistry, 48, 13581362.
Geankoplis, C. J. (1978). Transport processes and unit operations. Newton,
MA: Allyn and Bacon.
Hayaloglu, A. A., Karabulut, I., Alpaslan, M., & Kelbaliyev, G. (2007).
Mathematical modeling of drying characteristics of strained yoghurt in
a convective type tray-dryer. Journal of Food Engineering, 78, 109117.
Henderson, S. M., & Pabis, S. (1961). Grain drying theory. I. Temperature
eect on drying coecient. Journal of Agricultural Engineering
Research, 6, 169277.
Henderson, S. M., & Pabis, S. (1962). Grain drying theory. IV. The eect
of air ow rate on drying index. Journal of Agricultural Engineering
Research, 7, 8589.
Herman-Lara, E., Salgado-Cervantes, M. A., & Garca-Alvarado, M. A.
Fig. 9. Experimental drying kinetics for temperature of coee beans at (2005). Mathematical simulation of convection food batch drying with
treatments 1 (s), 2 (h), 3 (+), and 4 () of Table 1, and simulated assumption of plug ow and complete mixing of air. Journal of Food
temperature kinetics (continuous line) with Eqs. (31)(34). Engineering, 68, 321327.
 W.N. Hernandez-Daz et al. / Journal of Food Engineering 86 (2008) 19 9

Iglesias, H. A., & Chirife, J. (1982). Handbook of food isotherm: Water International Drying Simposium (IDS 2002), Beijing China. B, pp.
sorption parameters for food and food components. Food science and 13421351.
technology a series of monograph. New York: Academic Press. Sfredo, M. A., Finzer, J. R. D., & Limaverde, J. R. (2005). Heat and Mass
Jamradloedluk, J., Nathakaranakule, A., Soponronnarit, S., & Prachay- transfer in coee fruits drying. Jornal of Food Engineering, 70, 1525.
awarakorn, S. (2007). Inuences of drying medium and temperature on Sherwood, T. K. (1929a). The drying of solids-I. Industrial and Engineering
drying kinetics and quality attributes of durian chip. Journal of Food Research, 21, 1216.
Engineering, 78, 198205. Sherwood, T. K. (1929b). The drying of solids-II. Industrial and
Kashaninejad, M., Mortazavi, A., Safekordi, A., & Tabil, L. G. (2007). Engineering Research, 21, 976980.
Thin-layer drying characteristics and modeling of pistachio nuts. Sherwood, T. K. (1932). The drying of solids-IV. Application of diusion
Journal of Food Engineering, 78, 98108. equations. Industrial and Engineering Research, 24, 307310.
Lima, A. G. B., Queiroz, M. R., & Nebra, S. A. (2002). Spencer, H. B. (1969). A mathematical simulation of grain drying. Journal
Simultaneous moisture transport and shrinkage during drying of of Agricultural Engineering Research, 14, 226235.
solids with ellipsoidal conguration. Chemical Engineering Journal, Spiegel, M. R. (1968). Schaums outline of mathematical handbook of
86, 8593. formulas and tables. USA: McGraw-Hill.
Paulino de Morales, M. E., & Luchese, R. H. (2003). Ochratoxin A in Suarez-Quiroz, M. L., Gonzalez-Rios, O., Barel, M., Guyot, B., Schorr-
green coee: inuence of harvest and drying processing procedures. Galindo, S., & Guiraud, J. P. (2004a). Eect of chemical and
Journal of Agricultural Food Chemistry, 51, 58245828. environmental factors on Aspergillus ochraceus growth and toxigenesis
Perez-Alegra, L. R., & Ciro-Velasquez, H. J. (2001). Mathematical in green coee. Food Microbiology, 21, 629634.
simulation of parchment coee drying in a deep bed with airow Suarez-Quiroz, M. L., Gonzalez-Rios, O., Barel, M., Guyot, B., Schorr-
reversal. Transactions of ASAE, 44, 12291234. Galindo, S., & Guiraud, J. P. (2004b). Study of ochratoxin A
Perez-Alegra, L. R., Ciro, H. J., & Abud, L. C. (2001). Mathematical Producing strains in coee processing. International Journal of Food
simulation of parchment coee drying in a deep bed with airow Science Technology, 39, 501507.
reversal. Transactions of ASAE, 44, 17211726. Torrez, N., Gustafsson, M., Schreil, A., & Martinez, J. (1998). Modeling
Ruiz-Lopez, I. I., Cordova, A. V., Rodrguez-Jimenes, G. C., & Garca- and simulation of crossow moving bed grain dryers. Drying
Alvarado, M. A. (2004). Moisture and temperature evolution during Technology, 16, 19992015.
food drying: eect of variable properties. Journal of Food Engineering, Varadharaju, N., Karunanidhi, C., & Kailappan, R. (2001). Coee cherry
63, 117124. drying: a two layer model. Drying Technology, 19, 709715.
Sfredo M. A., Finzer J. R. D., & Limaverde J. R., (2002). Study of Zogzas, N. P., & Maroulis, Z. B. (1996). Eective moisture diusivity
the drying process of Arabica coee cherries using vibrated trays estimation from drying data. A comparison between various methods
driers in the ne drink attainment. In Proceedings of the 13th of analysis. Drying Technology, 14, 15431573.