See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/240915210

Evaluation of a simulation model in predicting the drying parameters for

deep-bed paddy drying

Article  in  Computers and Electronics in Agriculture · August 2009

DOI: 10.1016/j.compag.2009.04.007

CITATIONS READS

60 898

2 authors, including:

Dariush Zare
Shiraz University
52 PUBLICATIONS   936 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Evaluation of Corn Drying Process in an Ultrasound-Assisted Fluidized Bed Dryer View project

Effect of radiation leakage of microwave oven on viability and concentration of sperm in mature and prepuberty male rat View project

All content following this page was uploaded by Dariush Zare on 19 July 2020.

The user has requested enhancement of the downloaded file.

 Computers and Electronics in Agriculture 68 (2009) 78–87

Contents lists available at ScienceDirect

Computers and Electronics in Agriculture

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

Evaluation of a simulation model in predicting the drying parameters for

deep-bed paddy drying
Dariush Zare a,∗ , Guangnan Chen b
a
Agricultural Engineering Dept., Faculty of Agriculture, Shiraz University, Shiraz, Iran
b
Faculty of Engineering and Surveying, University of Southern Queensland, Toowoomba, QLD 4350, Australia

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

Article history: A simulation model for deep-bed batch drying of paddy was developed to predict the profiles of grain
Received 3 August 2008 moisture content, grain temperature, air temperature and air humidity during the drying process. In order
Received in revised form 23 April 2009 to evaluate the validity of this model, a laboratory-scale deep-bed batch dryer was designed and fabri-
Accepted 24 April 2009
cated. Comprehensive drying experiments were carried out in three replications under different drying
conditions with two independent drying variables, namely, drying air temperature (at two levels of 45 and
Keywords:
50 ◦ C) and air mass flow rate (at three levels of 0.1, 0.16 and 0.22 kg m−2 s−1 ). Good agreement was found
Paddy
between the simulation results and the experimental data. After validation of the model, the dryer per-
Deep-bed
Simulation model
formance was optimized by minimizing specific energy consumption under identical moisture removal
Energy optimization using the simulation model. Based on this criterion, the treatment with mass flow rate of 0.03 kg m−2 s−1
and air temperature of 35 ◦ C was found to be the optimal drying condition.
© 2009 Elsevier B.V. All rights reserved.

1. Introduction the control is performed by the feedforward controller which is

Rice is a major staple food source in the world. Improper dry-
ing can cause conspicuous losses to this product. Losses may be
reduced through appropriate controlling of the drying process.
Extensive characterization of grain drying process using experi-
mental approaches is tedious and complicated and sometimes even
impossible due to the large number of variables involved. Drying
behaviour can be affected by a large variety of factors including
drying temperature, air velocity and relative humidity. It is also
affected by grain properties such as grain density, permeability, and
porosity. On the other hand, the drying of moist porous materials
is a complicated process involving simultaneous coupled heat and
mass transfer phenomena (Chua et al., 2002). Since computer sim-
ulation is a reasonable and powerful tool for providing a view into
physics of drying process it can be applied for predicting of drying
parameters including grain temperature, grain moisture content,
air humidity and temperature. By predicting these parameters we
are able to design new high efficient dryer or improve existing grain
drying system. The computer simulation of grain drying can be
applied for optimizing and controlling of grain drying systems and
processes. A process simulation model is an essential element of a
feedforward–feedback automatic grain dryer control system. In a
properly designed feedforward–feedback control system, most of
∗ Corresponding author. Tel.: +98 9171087383.
E-mail addresses: dzare@shirazu.ac.ir (D. Zare), chengn@usq.edu.au (G. Chen).
done by process model (Pabis et al., 1998).
Deep-bed dryers, also known as fixed bed dryers, are one of
the most common types of agricultural dryers, designed for het-
erogeneous drying of grain in a deep layer (more than 20 cm deep)
where drying is faster at the inlet end of the dryer than that at
the exhaust end (Lopez et al., 1998). Deep-bed models for grain
drying simulation can be classified as logarithmic, heat and mass
balance and partial differential equation (PDE) (Tang et al., 2004;
Pabis et al., 1998). The PDE model is more detailed, accurate and
valid for cereal drying, while the others are less accurate owing to
more assumptions being made during model derivations (Tang et
al., 2004; Parry, 1985). Various simulation models have been devel-
oped for thin layer drying of paddy (Abe and Afzal, 1997; Agrawal
and Singh, 1977; Chen, 1998; Chen and Shei, 1996; Islam and Jindal,
1981; Jindal and Siebenmorgen, 1994). However, only limited sim-
ulation models have been presented for describing paddy drying in
deep-batch dryer which are not based on PDE models (Noomhorn
and Verma, 1986; Sitompul et al., 2001). In this study, a comprehen-
sive evaluation of the PDE model for drying of paddy in a deep-bed
batch dryer was conducted and as far as we know, no study has
been reported on this subject in the literature.
For evaluating the simulation model a laboratory-scale dryer
was fabricated and several tests were conducted in three repli-
cations for different combinations of drying air temperatures and
air mass flow rates. Once a grain drying model is validated, com-
puter simulation can be used for practical applications. One of the
most interesting applications of drying simulation is optimization
of dryer performance from view points of energy consumption,

0168-1699/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.compag.2009.04.007
 D. Zare, G. Chen / Computers and Electronics in Agriculture 68 (2009) 78–87 79

efficiency and moisture removal. Here, the dryer performance is

Nomenclature optimized based on specific energy consumption.

A inlet area of bin (m2 )

as specific surface area (m2 m−3 ) 2. Materials and methods
ca specific heat capacity of air (J kg−1 K−1 )
cv specific heat capacity of water vapor (J kg−1 K−1 ) 2.1. Development of computer simulation model
cw specific heat capacity of water (J kg−1 K−1 )
cd specific heat capacity of dry mass equal to 1109 for In deriving the mathematical non-equilibrium PDE model for
unshelled rice (J kg−1 K−1 ) batch drying of paddy, the following assumptions were made
cp specific heat capacity of paddy (J kg−1 K−1 ) (Mandas and Habte, 2002; Parry, 1985):
Et total energy supplied for the drying process (J)
FP power supplied to the fan per unit bin floor area 1. The volume shrinkage of the bed of grain was negligible during
(W m−2 ) the drying process.
G mass flow rate of air (kg m−2 s−1 ) 2. The temperature gradients within the individual kernels of grain
ha grain bed volumetric heat transfer coefficient were negligible.
(J m−3 K−1 s−1 ) 3. The heat transfer by conduction between kernels of grain was
hv latent heat of vaporization (J kg−1 ) negligible.
H absolute humidity of air (kg kg−1 ) 4. The bin walls were adiabatic, with negligible heat capacity.
HP heat power required for drying (W) 5. During any short time interval, the heat capacity of moist air and
L depth of grain bed (m) grain were constant.
m total number of layers 6. (∂T/∂t) and (∂H/∂t) were negligible compared with (∂T/∂x) and
M moisture content of grain (dry basis) (kg kg−1 ) (∂H/∂x), where x was the bed thickness in the direction of air
Mab moisture content of a batch at end of drying process flow.
(dry basis) (kg kg−1 )
Me equilibrium moisture content of grain (dry basis) By conducting the heat and mass balances, similar to Sharp’s
(kg kg−1 ) the following set of partial differential equations can be obtained
Mi predicted moisture content at ith layer (dry basis) (Sharp, 1982).
(kg kg−1 ) Grain moisture-mass balance:
MRD mean relative deviation (%)
∂H ∂M ∂H
n total number of measurements in each experiment G = −p −ε (1)
∂x ∂t ∂t
Of percentage perforation (decimal)
P atmospheric pressure (Pa) Heat balance (energy) balance of air:
Pvs saturated vapour pressure (Pa)
∂T ∂M ∂T
r0 equivalent particle radius (m) G(ca + cv H) = p cv (T − ) − ha (T − ) − a ε(ca + cv H)
∂x ∂t ∂t
RH relative humidity of air (decimal)
(2)
RE relative error at each time step (%)
t time (s)
T air temperature (◦ C) Heat transfer equation (energy balance of grain):
Tabs absolute air temperature (K)
Ta ambient air temperature (◦ C) p (cp + cw M)
∂
= ha (T − ) + hv
∂M
p (3)
Td drying air temperature (◦ C) ∂t ∂t
V air velocity (m s−1 ) There are four unknown drying variables, namely, M(x, t), (x,
w1 thermal–electrical energy weight (decimal) t), H(x, t) and T(x, t) in the above three equations. To complete the
x depth in bed from air inlet (m) system of equations, an appropriate thin layer equation is normally
Yj jth predicted parameter employed (Pabis et al., 1998). In this study, a modified empirical
Ŷj jth experimental parameter thin layer drying equation for paddy was used (Wang and Singh,
 grain temperature (K) 1978):
particle density (kg m−3 ) 
p
∂M
 t Y −1 
ε porosity (decimal)
= [M(t) − Me ] −XY (4)
a air viscosity (kg m−1 s−1 ) ∂t 60
Pg pressure drop for clean grain (Pa)
Pf pressure drop due to perforated floor (Pa) where
Pd pressure drop due to duct system (Pa)
X = 0.01579 + 0.0000176T + 0.078867(RH) (4-a)
P total static pressure drop (Pa)
t drying duration (s) Y = 0.6545 + 0.002425T + 0.078867(RH) (4-b)
f fan efficiency (decimal)
th–elec conversion efficiency of thermal to electrical energy Some additional information about properties of paddy was
(decimal) needed including specific heat capacity of moist grain, the
trans–dist transmission and distribution network efficiency equilibrium moisture content and the volumetric heat transfer
(decimal) coefficient.The specific heat capacity of grain was calculated by the
following relationship (Pabis et al., 1998):

cp = cd + 4186M (5)

The equilibrium moisture content (Me ) of paddy is independent

of the rice type. It is only a function of air temperature and relative
80 D. Zare, G. Chen / Computers and Electronics in Agriculture 68 (2009) 78–87

∂T ∂M
G(ca + cv H) = p cv (T − ) − ha (T − ) (12)
∂x ∂t
Regarding the stated boundary conditions, the predicted tem-
perature and humidity for the first layer at time t + ıt become:
t+ıt  t+ıt t
T = T , H  = H
t
(13)
1 1 1 1

The moisture content of the grain at time t + ıt for first layer can
Fig. 1. Schematic representation of the ith layer of dryer bed. be calculated as:
 t
t+ıt t ∂M 
humidity (Brooker et al., 1992a). The following empirical relation- M = M + · ıt (14)
1 1 ∂t 
ship is recommended for the Me of rough rice at temperatures below 1
50 ◦ C.
 A c where (∂M/∂t) is obtained from Eq. (4).
Me = 0.001 (6) The three parameters of temperature, humidity and grain mois-
B ture content were determined for the first layer at time t + ıt. For
determining the last parameter, i.e. the grain temperature, it is
 −23.438 t
Tabs required to calculate (∂/∂t) using the governing Eq. (3) directly,
A = −ln(1 − RH)Tabs , B = 2.667 × 10−7 1 − , t+ıt 1
641.7
and then   is determined from the following equation:
1
c =
1  t
4.0E5(Tabs )−2.1166 t+ıt ∂ 
 = 1t + · ıt (15)
1 ∂t 
The volumetric heat transfer coefficient for paddy in a bed was 1

calculated using Barker’s equation (Barker, 1965): At the end of this predictor step method, the four unknown
 2r G b parameters of T, H, M and  will be calculated for the first layer
0
ha = aca G · as (7) at time t + ıt. Next, the values of the exhaust air temperature and
a
humidity are calculated for the second layer using the finite differ-
ence method:
a, b = constants (a = 0.2755 and b = −0.34),  t+ıt
r0 = equivalent particle radius equal to 0.00457 m for long rough t+ıt t+ıt ∂T 
T = T + · ıx (16)
∂x 
rice (Brooker et al., 1992a), 2 1
as = specific surface area of long paddy equal to 2437 m2 m−3 1
(Fontana, 1983),  t+ıt
a = 1.691 × 10−5 + 4.984 × 10−8 T − 3.187 × 10−11 T2 + 1.319 × 10−14 T3 . t+ıt t+ıt ∂H 
H = H + · ıx (17)
2 1 ∂x 
1
Different numerical methods have been proposed for solving
the coupled partial differential equations (Turner and Mujumdar, In Eqs. (16) and (17), (∂H/∂x) and (∂T/∂X) are calculated using Eqs.
1997). In this study, an iterative-predictor finite difference method (11) and (12), respectively. At the end of this step, the inlet and outlet
was utilized to solve the set of partial differential equations. For air temperatures and absolute humidity as well as grain tempera-
solving this set of PDE’s, two boundary conditions and two initial tures and moisture contents are known for the first layer. The above
conditions must be known. In a deep-batch dryer that is filled with procedure is repeated until the depth and drying time reaches the
paddy having an initial temperature of  i and initial moisture con- desired value. In the simulation of a deep-bed batch drying process,
tent of Mi subjected to constant inlet air temperature Ti and absolute the bed height is divided into a number of thin layers of thickness
humidity Hi , the initial and boundary conditions would be: (x). For better moisture content prediction it is better to divide
the bed to more layers. In this study several values of x were tried
Initial conditions: and it was concluded that a 1 cm thick layer would be sufficient for
 the computer simulation to give stable and convergent results with
(x, 0) = i good accuracy. The computer program was developed and written
(8)
M(x, 0) = Mi in standard FORTRAN 90 language (Zare, 2006). The structure of the
program is shown in the flow chart (Fig. 2). For running the program,
Boundary conditions:
 a number of input data are required. These include: the initial grain
T (0, t) = Ti temperature and moisture content, the inlet air temperature and
(9)
H(0, t) = Hi absolute humidity, the inlet air flow rate, the desired monitoring
depth of grain and depth increment, and the desired drying time
A schematic representation of the ith layer of the dryer bed is and the time interval.
shown in Fig. 1. Let us assume that the dependent variables H, T, M
and  are known at time t for the first layer. Hence, we can write: 2.2. Drying experiments

M = M1t ,  = 1t , H = H1t , T = T1t (10) 2.2.1. Experimental setup

To evaluate the computer simulation, a deep-bed experimental
Keeping in mind that the contribution of the derivative terms
dryer was designed and constructed (Zare, 2006). The system con-
(∂T/∂t) and (∂H/∂t) in the governing Eqs. (1) and (2) are very small
sisted of an electric heating unit, a hot air conveying pipeline, air
for each thin layer compared to the other terms, so they can be
supply and distribution system and a data acquisition unit (Fig. 3).
neglected and Eqs. (1) and (2) reduced as follows:
The inlet air temperature to dryer plenum was controlled ther-
∂H ∂M mostatically (±0.5 ◦ C) and remained unchanged at the set point
G = −p (11)
∂x ∂t during each of the individual drying process. A backward-curved
 D. Zare, G. Chen / Computers and Electronics in Agriculture 68 (2009) 78–87
Fig. 2. Flow chart for the computer program of simulation.
centrifugal fan was used to provide the required air flow to the dry-
ing system. The mass flow rate of air was adjusted by an electrical
inverter capable of controlling the fan speed, with control frequency
accuracy of ±0.1 Hz. The overall dimensions of the drying chamber
were 40 cm × 40 cm × 40 cm, which gave a full capacity of about
30 kg of rough rice. A galvanized screen was incorporated below the
dryer bed to let the hot air in uniformly, as well as to support the
drying chamber. To minimize possible heat losses, the electric heat-
ing unit, the air flow duct, the plenum chamber and the dryer bed
all were insulated. Prior to drying, the dryer was filled with grain
to a uniform height of 25 cm. The grain used in the experiment was
a native variety of long grain paddy named Hashemi, which was
obtained from a local paddy field in north of Iran. All samples in the
present experiments were natural freshly harvested without any
artificial rewetting prior to the test. The fresh material was put in
a sealed package to avoid moisture evaporation and was stored at
5 ◦ C. The initial moisture content and temperature of grains were
24–25% dry basis and 25 ◦ C, respectively for all experiments.
2.2.2. Measured parameters
In order to validate the simulation model in different drying con-
81
paddy drying, were chosen (Brooker et al., 1992b; Shei and Chen,
2002). For better evaluation of the simulation model, the duration
of each drying test was fixed at 150 min because for this duration
the moisture gradients could be achieved along the depth of dryer
bed. Under each set of drying conditions, the tests were conducted
in triplicate to assure the repeatability of the data. During the exper-
iments, the absolute humidity of the inlet air remained unchanged
at ambient condition of 0.01 kg kg−1 .
In order to equalize the temperature and relative humidity
sensors with surrounding air and also to eliminate fluctuation of
measured parameters, sampling of air temperature and relative
humidity were performed after 30 min from the beginning of the
drying process. During the experiments, the air temperature and
relative humidity were measured at 5 min intervals at three levels
of 8–9, 17–18, and 24–25 cm from the inlet screen. For determining
grain moisture content, grain samples were collected every 10 min
from the beginning of drying using a sampling spear with a sliding
trap door. The grain sampling positions were at two depths of 8–9
and 17–18 cm. The average moisture content of the grain samples
was determined using a vacuum oven based on the ASAE standard
(ASAE, 2000). The mass flow rate of drying air was determined
using a vane anemometer, “Loutron A-M-4202”, with an accuracy
of ±0.1 m s−1 . The temperature was measured by a semi-conductor
type sensor (LM35A) having an accuracy of ±0.2 ◦ C at 25 ◦ C and non-
linearity of ±0.18 ◦ C. This sensor is rated to operate over the range
of −55 to 150 ◦ C. The temperature sensors were calibrated over the
operating temperature of 0–60 ◦ C. For calibration, the sensors were
immersed into an insulated water bath equipped with an electrical
heating unit and precise thermometer (Testo735-2, with accuracy
of ±0.2 ◦ C in span of −55 to 150 ◦ C). Using water bath, different
temperatures were made in the range of 0–60 ◦ C with 5 ◦ C inter-
val and corresponding voltage for each temperature was recorded.
Next, the best fit linear lines (R2 ≥ 0.985) between output voltages
of the sensors and temperatures were determined. Finally, the inter-
cept and slop of the calibration lines were obtained. For measuring
relative humidity, capacitance type relative humidity sensors (Hon-
eywell HIH-3610) were used. This type of sensor has an accuracy of
±2% and linearity of ±0.5% for the operating span of 0–100% RH at
25 ◦ C. In case of HIH-3610 sensor, the manufacture’s calibration was
accepted and the accuracy of the sensor was verified using Testo625
having the accuracy of ±2.5% RH.
As it was described the absolute humidity is one of the unknown
parameters used in solving the model. For changing the relative
humidity to the absolute humidity the following equation was used
(Brooker et al., 1992c):
RH · Pvs
H = 0.622 (18)
P − RH · Pvs
2.3. Validation criteria
The first aim of the study is to evaluate the validation of the
model in prediction of grain moisture content, air temperature, and
absolute air humidity under different drying conditions. To achieve
this aim the numerical results of the PDE model were compared
with the experimental results. For each test, the accuracy of the
model was determined by calculating the relative error at each time
step and also by the mean relative deviation defined as follows:
 
Yj − Ŷj 
RE = · 100 (19)
Ŷj
⎡ ⎤0.5
n  2
ditions, two independent factors of drying air temperature (at two 1  Yj − Ŷj
MRD = ⎣ ⎦ · 100 (20)
levels of 45 and 50 ◦ C) and mass flow rate (at three levels of 0.1, 0.16, n Ŷj
0.22 kg m−2 s−1 ), which are in the recommended range of industrial j=1
82 D. Zare, G. Chen / Computers and Electronics in Agriculture 68 (2009) 78–87
Fig. 3. The lab-scale deep-bed batch dryer.
where Yj and Ŷj indicate the jth predicted and experimental param-
eters, respectively, and n is the total number of measurements in
each experiment.
2.4. Optimization criterion
The other aim of the present study is to use the verified simula-
tion model for optimization of dryer performance from view points
of specific energy consumption. However, other factors such as the
fixed cost of dryer, drying duration or throughput capacity of dryer,
the quality of paddy, the labour cost and weather condition affect
the economy of the drying process as well as dryer performance.
Energy efficiency or specific energy consumption is generally
expressed as the amount of energy required to evaporate 1 kg of
water from the grain during the drying process. It depends on the
design of the dryer, the amount of moisture removed, weather con-
ditions during drying, the initial grain temperature and moisture
content, the depth of grain in dryer, the drying air temperature
and mass flow rate. Since minimizing drying cost involves minimiz-
ing specific energy consumption, in this study the specific energy
consumption was considered as optimization criterion.
Generally, two types of energy were involved in convective dry-
ing. First, the energy required to drive the fan for supplying air flow
to the dryer bed and second, the energy required to heat up the
coming air to the dryer chamber. Three parameters have effects
on the fan power which are static pressure due to grain resistance,
pressure drop due to air conveying ducts and bin floor. Among them
the grain and bin floor pressure drop are the most significant (Pabis
et al., 1998).
The power supplied to the fan per unit inlet area of the bin floor
(FP) is obtained by the following equation:
FP = f · P · V
where f is the fan efficiency, for the deep-bed drying a backward-
curved centrifugal fans are usually used and depending on volume
flow rate the fan efficiency varies. Based on characteristic curves for
a typical backward-curved centrifugal fan and assuming the flow
rate of 0.22 kg m−2 s−1 is 80% of full delivery of the fan, the corre-
sponding efficiencies of the fan for the mass flow rates of 0.01, 0.03,
0.06, 0.1, 0.16, and 0.22 kg m−2 s−1 are 7, 22, 42, 61, 71, and 50%,
respectively (ASHRAE, 1985).
V is the air velocity through the bin floor (m s−1 ), and total pres-
sure drop, P, is sum of the pressure drop for clean grain, duct
system and perforated floor.
P = Pg + Pd + Pf (22)
The pressure drop for clean grain, Pg , can be calculated from
the grain resistance equation for rough rice (ASAE, 2004):
2.57 · 104 V 2 L
Pg = (23)
ln(1 + 13.2V )
where L is the depth of grain bed (m).
Losses of pressure in ducts supplying air to grain drying sys-
tems results from friction, restriction to air flow, change in direction,
and enlargement or contraction of cross-sectional area of the flow
stream. Considering all this factors, the pressure drop due to air
conveying duct was measured using a manometer at each level of
mass flow rate.
The pressure drop due to perforated floor, Pf , can be calculated
by Brooker et al.’s equation as (Brooker et al., 1974):
 2
V
Pf = 1.076 (24)
εOf
where ε is porosity of grain (decimal) and, Of is percentage perfora-
tion (decimal). If the area of all opening is 10% or more of the total
surface area, the pressure drop through a perforated floor is con-
(21)
sidered negligible compared to the pressure drop through the grain
(Pabis et al., 1998). In the present study, Of is 0.3 for bin floor used
and ε is about 0.5 for long paddy. Therefore for all mass flow rates,
the pressure drop for bin floor can be neglected compared to grain
pressure drop.
 D. Zare, G. Chen / Computers and Electronics in Agriculture 68 (2009) 78–87
The heat power per unit area of dryer bin required for drying can
be calculated as:
HP = a · V · cp (Td − Ta )
where Td and Ta are drying air and ambient temperature, respec-
tively.
Normally, in the drying process heat energy is supplied from
fossil fuels whereas fan energy is supplied from electricity. These
energies are different grades therefore in order to sum up these
two kinds of energy, a weighted sum is used to calculate the total
equivalent thermal energy as follows:
Et = (FP · w1 + HP)A · t
where A is inlet area of bin, t is drying process duration in sec-
ond and w1 is thermal–electrical energy weight. Two factors affect
the weight of thermal to electrical energy. First, the conversion effi-
ciency of thermal to electrical energy and second, the transmission
and distribution network efficiency. The electricity supply industry
in Iran has been based mostly upon thermal power plants. There
are three categories of thermal power plants in Iran; steam turbine,
gas turbine and diesel generator. Based on Iran’s statistic, the con-
version efficiency of thermal to electrical energy is 37% on average
in 2006 and with the present technology the transmission and dis-
tribution network efficiency is about 80% (Karbassi et al., 2007).
Therefore the thermal–electrical energy weight can be formulated
as:
1 ity are also presented in Table 1. As it is shown, the mean relative
w1 =
th–elec · trans–dist
where th–elec is conversion efficiency of thermal to electrical energy
and trans–dist is transmission and distribution network efficiency.
Thus for Iran’s condition the average thermal–electrical energy
weight is about 3.38.
If the heat energy is supplied from electricity then the total
equivalent thermal energy is calculated as follows:
Et = (FP + HP) · w1 · A · t
Now the total equivalent thermal energy consumption for dry-
ing can be calculated using Eqs. (26) or (28) depending on heat
energy sources. For determining the specific energy consumption
it is essential to know the average moisture content of the batch
under each drying condition. Using the simulation model it is pos-
sible to predict the grain moisture content at different depth and
drying time. The average grain moisture content for a batch at the
end of drying process can be calculated as follows:
m
i=1
Mi
Mab =
m
where Mi indicates the ith predicted moisture content at ith layer
and m is the total number of layers.
For determining the specific energy consumption the model was
run under combinations of different drying conditions of air tem-
peratures (35, 40, 45, 50 ◦ C) and mass flow rates (0.01, 0.03, 0.06,
0.1, 0.16, 0.22 kg m−2 s−1 ). For all runs the initial grain tempera-
ture and moisture content were considered to be 27 ◦ C and 24.75%
(d.b.), respectively. The ambient air temperature and humidity were
assumed to be fixed at 30 ◦ C and 0.01 kg kg−1 , respectively. In order
to keep the breakage losses of rice to the minimum level in most
high temperature deep-bed paddy dryer the grain depth ranged
from 25 to 30 cm and here the depth of grain in dryer was con-
sidered to be 25 cm (Khoshtaghaza et al., 2007). The grain bulk
density was assumed to be 560 kg m−3 . Under each drying condition
the corresponding drying duration for decreasing the average grain
moisture content of a batch to the safe storage moisture content
of 12% (d.b.), were determined using the simulation model. Next,
for each run the heating and motive energy involved in drying was
83
calculated per square meter of dryer bed. Finally, specific energy
consumption was determined for each drying condition.
(25)
3. Results and discussions
3.1. Comparison of grain moisture content, air temperature and
humidity
The predicted results for the variations of grain moisture con-
tent, air temperature, and absolute air humidity under different
drying conditions were obtained and compared with the exper-
imental results. In Figs. 4–6, the results for run 3 were shown.
(26)
Similar results were found in other runs as summarized in Table 1.
From Fig. 4, it can be seen that at both depth levels, the model can
give a very good prediction of grain moisture content when the
run duration was less than 60 min. However, the model somehow
underestimates the grain moisture content in the final stage of dry-
ing. In all runs, the average mean relative deviation values for the
prediction of grain moisture content vary between 6.77 and 10.01%
(Table 1).
The comparison of air temperature at different depths starting
from 30th min is shown in Fig. 5. It can be seen that at all depths, the
model underestimates the measured air temperatures. The mean
relative deviation values for air temperature at different runs vary
between 7.92 and 10.19% (Table 1).
The experimental and simulated results of absolute air humid-
(27) deviation values for absolute humidity of air vary between 8.04 and
10.15% for different runs. In contrast with the grain moisture con-
tent and air temperature, the absolute air humidity is overestimated
by the current simulation model (Fig. 6). This is expected because
the actual grain moisture content is higher than the predicted one.
This means that in the experimental tests, less moisture would be
removed from the grain in comparison with the computer simula-
tion. This leads to the actual air humidity around the grain being
(28) lower than theoretical prediction.
In many of the drying simulation studies, the difference between
the measured and predicted parameters was reported to vary
between 10 and 15%, and this is usually regarded as satisfactory
results (Sitompul et al., 2001; Madhiyanon et al., 2001; Dimitriadis
and Akritidis, 2004; Kalbasi, 2003; Tang et al., 2004). Overall,
it can be found that the accuracy of the present model predic-
tion is reasonable in comparison with the experimental results
and it has achieved similar or better level of accuracy than those
reported in the other studies (Pabis et al., 1998; Sitompul et al.,
2001; Madhiyanon et al., 2002; Dimitriadis and Akritidis, 2004;
(29)
Fig. 4. Experimental and predicted grain moisture content with drying time at dif-
ferent depths for run 3.
84 D. Zare, G. Chen / Computers and Electronics in Agriculture 68 (2009) 78–87

Fig. 5. Experimental and predicted air temperature with drying time at different depths for run 3.

Fig. 6. Experimental and predicted air humidity with drying time at different depths for run 3.

Kalbasi, 2003; Tang et al., 2004). It is identified that the main error
sources between the simulated and experimental data are due to:
(1) the simplification assumption made in derivation of mathe-
matical model, (2) the lack of accuracy of thin layer grain drying
equation, (3) the inadequacy of precise equation for estimating vol-
umetric heat transfer of paddy in a packed bed, (4) the insufficient
precision of moisture equilibrium isothermal equation at relative
humidity above 90% and (5) the error in measurement of input
parameters and actual performance of grain dryer. It is believed that
further investigations of the model sensitivity and the removal of
the uncertainty of some of the model inputs would further improve
the accuracy of the present model.
3.2. Prediction of grain temperature during drying
In the present research, the actual grain temperature has not
been measured. However, it is possible to use the computer simu-
lation to show the grain temperature profile at each depth during

Table 1
Results of validation tests for the simulation parameters.

Run Drying air Drying air mass Initial moisture % MRDa (grain % MRDa (air % MRDa (air
temperature flow rate content moisture content) temperature) humidity)
(◦ C) (kg m−2 s) (dry basis%)
Depth Depth Depth Depth Depth Depth Depth Depth
8–9 cm 17–18 cm 8–9 cm 17–18 cm 24–25 cm 8–9 cm 17–18 cm 24–25 cm

1 50 0.22 24.69 8.67 8.85 7.92 8.90 9.05 9.98 8.48 9.41
2 45 0.22 25.87 8.80 8.18 8.75 10.19 9.17 10.15 9.40 9.37
3 50 0.16 24.75 8.81 8.48 8.40 9.31 9.79 9.83 8.74 9.25
4 45 0.16 25.12 7.92 7.50 9.40 9.81 10.02 8.88 8.79 9.46
5 50 0.1 25.57 8.81 7.79 9.44 9.00 10.16 8.14 8.71 8.42
6 45 0.1 25.36 10.01 6.77 9.74 9.80 9.18 9.15 8.04 7.95
a
Mean relative deviation.
 D. Zare, G. Chen / Computers and Electronics in Agriculture 68 (2009) 78–87 85

mization were obtained without recalibration on the experimental

Fig. 7. Predicted grain and air temperature with drying time at different depths for
run 3.
the drying process. This information is particularly useful in deter-
mining the grain quality during drying process. It has been found
that for all present runs, the simulated grain and air temperature
almost coincides (Fig. 7). This may be related to the fact that the air
velocity in this experiment was relatively high, which has a direct
effect on the convective heat transfer coefficient. The reduction of
temperature in the first 20 min in Fig. 7 is related to the moisture
condensation on the cool surface of grain and the evaporation of
unbounded moisture from the grain.
3.3. Determining the optimal drying condition
In this study, the optimum drying condition was determined
based on the specific energy consumption using the simulation
model. It must be mentioned that the simulation results for opti-
data. Two strategies of heat supplying energy were considered here.
In the first strategy heat was supplied from fossil fuel (Table 2) and
in the second one heat was supplied from electricity (Table 3). For
all runs the average moisture content of the batch were decreased
from 25.47% (d.b.) to 12% (d.b). It can be seen that by increasing
both mass flow rates and drying air temperature the drying dura-
tion were decreased. This is due to the fact that with the increase of
mass flow rates the convective heat and mass transfer coefficient
increases which results in faster evaporation of the moisture. By
increasing the drying air temperature, the mass diffusion coefficient
increases and the rate of water evaporation increases as well.
For each treatment, the energy involved for heating and blow-
ing of the air was calculated and shown in Tables 2 and 3. The
energy required for heating of the air increases as the drying air
mass flow rate and temperature increase. The fan consumption
energy is calculated by multiplying fan power by drying dura-
tion. As it shown in Tables 2 and 3, at each level of mass flow
rate the fan consumption energy increases when the air tempera-
ture decreases because at lower drying temperature the fan should
be used for longer time to obtain the same moisture removal.
The fan consumption energy is more significant for higher mass
flow rate (0.22, 0.16 kg m−2 s−1 ) because higher mass rate produces
higher static pressure. In addition, for lower mass flow rate (0.01,
0.03 kg m−2 s−1 ) the fan consumption energy is relatively high this
is due to the prolonged drying time and low fan efficiency at these
mass flow rates. Based on simulated data in Tables 2 and 3 the min-
imum fan consumption energy is belonged to drying air mass flow
rate of 0.06 kg m−2 s−1 and temperature of 50 ◦ C.
The total energy consumption and the specific energy con-
sumption are shown in Tables 2 and 3. The lower specific energy
consumption means the higher energy efficiency. In the first strat-
egy of drying when heat is supplied from fossil fuel it is found that
the treatment with mass flow rates of 0.03 kg m−2 s−1 and drying
air temperature of 35 ◦ C is the optimum drying condition. On the
other hand when heat is supplied from electricity there is no opti-

Table 2
The specific energy consumption for different simulated treatments when heat energy is supplied from fossil fuel (shaded area shows minimum specific energy consumption).

Temperature Mass Initial Final Drying Fan Heat Total Mass of Specific
(◦ C) flow rate average average duration energya energya energya evaporated energy
(kg m−2 s−1 ) moisture moisture (min) (MJ m−2 ) (MJ m−2 ) (MJ m−2 ) water consumption
content content (kg m−2 ) (MJ kg−1 )
of a batch of a batch
(%d.b.) (%d.b.)

35 0.01 24.75 12 1400 14.37 4.23 18.59 14.30 1.30

40 0.01 24.75 12 1128 11.58 6.82 18.39 14.30 1.29
45 0.01 24.75 12 956 9.81 8.67 18.48 14.30 1.29
50 0.01 24.75 12 838 8.60 10.13 18.73 14.30 1.31
35 0.03 24.75 12 911 9.89 8.25 18.14 14.30 1.27
40 0.03 24.75 12 675 7.32 12.24 19.56 14.30 1.37
45 0.03 24.75 12 545 5.91 14.82 20.74 14.30 1.45
50 0.03 24.75 12 455 4.94 16.50 21.44 14.30 1.50
35 0.06 24.75 12 748 9.65 13.55 23.21 14.30 1.62
40 0.06 24.75 12 538 6.94 19.50 26.45 14.30 1.85
45 0.06 24.75 12 418 5.39 22.74 28.13 14.30 1.97
50 0.06 24.75 12 339 4.38 24.59 28.97 14.30 2.03
35 0.1 24.75 12 678 11.71 20.48 32.19 14.30 2.25
40 0.1 24.75 12 480 8.29 29.00 37.30 14.30 2.61
45 0.1 24.75 12 365 6.31 33.09 39.39 14.30 2.75
50 0.1 24.75 12 290 5.01 35.06 40.07 14.30 2.80
35 0.16 24.75 12 635 17.42 30.69 48.11 14.30 3.36
40 0.16 24.75 12 442 12.13 42.73 54.86 14.30 3.84
45 0.16 24.75 12 330 9.06 47.86 56.92 14.30 3.98
50 0.16 24.75 12 260 7.13 50.29 57.43 14.30 4.02
35 0.22 24.75 12 615 38.16 40.86 79.03 14.30 5.53
40 0.22 24.75 12 427 26.50 56.76 83.26 14.30 5.82
45 0.22 24.75 12 315 19.55 62.82 82.37 14.30 5.76
50 0.22 24.75 12 250 15.51 66.50 82.01 14.30 5.73
a
All energies are equivalent of thermal energy.
86 D. Zare, G. Chen / Computers and Electronics in Agriculture 68 (2009) 78–87

Table 3
The specific energy consumption for different simulated treatments when heat energy is supplied from electricity (shaded area shows minimum specific energy consumption).

Temperature Mass flow Initial average Final average Drying Fan energya Heat Total Mass of Specific
(◦ C) rate moisture moisture duration (MJ m−2 ) energya energya evaporated energy
(kg m−2 s−1 ) content of a content of a (min) (MJ m−2 ) (MJ m−2 ) water consumption
batch (%d.b.) batch (%d.b.) (kg m−2 ) (MJ kg−1 )

35 0.01 24.75 12 1400 14.37 14.29 28.66 14.30 2.00

40 0.01 24.75 12 1128 11.58 23.04 34.61 14.30 2.42
45 0.01 24.75 12 956 9.81 29.29 39.10 14.30 2.73
50 0.01 24.75 12 838 8.60 34.24 42.84 14.30 3.00
35 0.03 24.75 12 911 9.89 27.90 37.79 14.30 2.64
40 0.03 24.75 12 675 7.32 41.35 48.68 14.30 3.40
45 0.03 24.75 12 545 5.91 50.10 56.01 14.30 3.92
50 0.03 24.75 12 455 4.94 55.78 60.72 14.30 4.25
35 0.06 24.75 12 748 9.65 45.82 55.47 14.30 3.88
40 0.06 24.75 12 538 6.94 65.92 72.87 14.30 5.10
45 0.06 24.75 12 418 5.39 76.85 82.24 14.30 5.75
50 0.06 24.75 12 339 4.38 83.12 87.49 14.30 6.12
35 0.1 24.75 12 678 11.71 69.21 80.93 14.30 5.66
40 0.1 24.75 12 480 8.29 98.03 106.32 14.30 7.43
45 0.1 24.75 12 365 6.31 111.84 118.14 14.30 8.26
50 0.1 24.75 12 290 5.01 118.51 123.52 14.30 8.64
35 0.16 24.75 12 635 17.42 103.72 121.14 14.30 8.47
40 0.16 24.75 12 442 12.13 144.42 156.55 14.30 10.95
45 0.16 24.75 12 330 9.06 161.78 170.84 14.30 11.95
50 0.16 24.75 12 260 7.13 170.00 177.13 14.30 12.39
35 0.22 24.75 12 615 38.16 138.12 176.29 14.30 12.33
40 0.22 24.75 12 427 26.50 191.84 218.34 14.30 15.27
45 0.22 24.75 12 315 19.55 212.34 231.89 14.30 16.22
50 0.22 24.75 12 250 15.51 224.75 240.27 14.30 16.80
a
All energies are equivalent of thermal energy.

mum point, as decreasing drying air temperature and mass flow rate
increase the drying efficiency. As it is clear from Tables 2 and 3 when
heat energy is supplied from electricity, the value of specific energy
consumption is significantly higher than those supplied from fossil
fuel. Therefore it is not recommended to use electricity for heating
of the air in drying process.
4. Conclusion
It has been found that the discrepancies between the model pre-
diction and the experiments are acceptable and are well within the
reported level of accuracy in other studies. The simulation model
can be used as a tool to optimize drying process of paddy in a
deep-bed batch dryer from view points of energy consumption, effi-
ciency and moisture removal. Based on minimum specific energy
consumption the treatment with mass flow rate of 0.03 kg m−2 s−1
and air temperature of 35 ◦ C was found to be the optimal dry-
ing condition when heat was supplied from fossil fuel. For further
study the quality of paddy, the fixed and variable cost of dryer,
the drying duration or throughput capacity of dryer, the labour
cost and weather condition can be considered in the optimization
process.
References
Abe, T., Afzal, T.M., 1997. Thin-layer infrared radiation drying of rough rice. Journal
of Agricultural Engineering Research 67, 289–297.
Agrawal, Y.C., Singh, R.D., 1977. Thin layer drying studies for short grain rice. ASAE,
St. Joseph (MI), Paper No.: 77-3531.
ASAE, 2000. Moisture measurement unground grain and seeds. ASAE, Standards. St.
Joseph (MI). S352.2 DEC92 47th.
ASAE, 2004. Resistance to airflow of grains, other agricultural products, and perfo-
rated metal sheets. ASAE Standards. St. Joseph (MI). D272.3 DEC01.
ASHRAE, 1985. Fundamentals Handbook. Am. Soc. Heating, Refrigeration, and Air-
Conditioning Engineering, New York.
Barker, J., 1965. Heat transfer in packed beds. Ind. Eng. Chem. 57, 43–51.
Brooker, D.B., Bakker-Arkema, F.W., Hall, C.W., 1974. Drying Cereals Grains. AVI, West-
port, CT.
Brooker, D.B., Bakker-Arkema, F.W., Hall, C.W., 1992a. Drying and Storage of Grains
and Oilseeds. AVI Publishing Co., New York.
Brooker, D.B., Bakker-Arkema, F.W., Hall, C.W., 1992b. Drying and Storage of Grains
and Oilseeds. Chapter 11, Rice Drying. AVI Publishing Co., New York, pp. 309–
334.
Brooker, D.B., Bakker-Arkema, F.W., Hall, C.W., 1992c. Drying and Storage of Grains
and Oilseeds. Chapter 3, Drying-Air Properties. AVI Publishing Co., New York, pp.
27–65.
Chen, Y.L., 1998. A new thin-layer equation for intermittent drying of rough rice.
Journal of Agricultural Machinery 6 (2), 81–89.
Chen, Y.L., Shei, H.J., 1996. Development of thin-layer equation for rough rice drying
at beginning period. Memorial College of Agriculture 36 (4), 321–328 (National
Taiwan University, Taipei, Taiwan, ROC).
Chua, K.J., Chou, S.K., Hawlader, N.A., Mujumdar, J.C.Ho., 2002. Modelling the
moisture and temperature distribution within an agricultural product under-
going time-varying drying schemes. Biosystems Engineering 81 (1), 99–
111.
Dimitriadis, N.A., Akritidis, C.B., 2004. A model to simulate chopped alfalfa drying in
a fixed deep bed. Drying Technology 22 (3), 479–490.
Fontana, C., 1983. Concurrent flow versus conventional drying of rice. Ph.D. Thesis.
Michigan State University, East Lansing (MI).
Islam, M., Jindal, V.K., 1981. Simulation of paddy drying under tropical conditions.
AMA 12 (3), 37–41.
Jindal, V.K., Siebenmorgen, T.J., 1994. Simulation of low-temperature paddy and re-
wetting in shallow beds. Transaction of ASAE 37 (5), 224–228.
Kalbasi, M., 2003. Heat and moisture transfer model for onion drying. Drying Tech-
nology 21 (8), 1575–1584.
Karbassi, A.R., Abduli, M.A., Mahin Abdollahzadeh, E., 2007. Sustainability of energy
production and use in Iran. Energy Policy 35, 5171–5180.
Khoshtaghaza, M.H., Sadeghi, M., Amirichayjan, R., 2007. Study of rough rice drying
process in fixed and fluidized bed conditions. Iranian Journal of Science and
Technology of Agriculture and Natural Resources 14 (2), 1–11.
Lopez, A., Pique, M.T., Romero, A., 1998. Simulation on deep bed drying of hazelnuts.
Drying Technology 16, 651–665.
Madhiyanon, T., Soponronnarit, S., Tia, W., 2001. A two-region mathematical model
for batch drying of grains in a two-dimensional spouted bed. Drying Technology
19 (6), 1045–1064.
Madhiyanon, T., Soponronnarit, S., Tia, W., 2002. A mathematical model for con-
tinuous drying of grains in spouted bed dryer. Drying Technology 20 (3), 587–
614.
Mandas, N., Habte, M., 2002. Numerical simulation static-bed drying of barley.
Biosystems Engineering 82 (3), 313–319.
Noomhorn, A., Verma, L.R., 1986. A deep-bed rice drying simulation using two gen-
eralized single-layer models. Transaction of ASAE 29, 1456–1461.
Pabis, S., Jayas, D.S., Cenkowski, S., 1998. Grain Drying Theory and Practice. John
Wiley & Sons, New York.
Parry, J.L., 1985. A review of mathematical modelling and computer simulation of
heat and mass transfer in agricultural grain drying. Journal of Agricultural Engi-
neering Research 32, 1–29.
Sharp, J.R., 1982. A review of low temperature drying simulation models. Journal of
Agricultural Engineering Research 27, 169–190.
 D. Zare, G. Chen / Computers and Electronics in Agriculture 68 (2009) 78–87 87

Shei, H.J., Chen, Y.L., 2002. Computer simulation on intermittent drying of rough rice.
Drying Technology 20 (3), 615–636.
Sitompul, J.P., Istadi, I., Widiasa, I.N., 2001. Modelling and simulation of deep-bed
grain drying. Drying Technology 19, 269–280.
Tang, Z., Cenkowski, S., Muir, W.E., 2004. Modelling the superheated-steam drying
of a fixed bed of brewers’ spent grain. Biosystems Engineering 87 (1), 67–77.
Turner, I., Mujumdar, A.S., 1997. Mathematical Modelling and Numerical Techniques
in Drying Technology. Marcel Dekker, New York.
Wang, C.Y., Singh, R.P., 1978. A single-layer drying equation for rough rice. ASAE, St.
Joseph (MI), Paper No. 78-3001.
Zare, D., 2006. Computer simulation of rough rice drying in a deep-bed batch dryer.
Ph.D. Dissertation. Tarbiat Modares University.

View publication stats