You are on page 1of 9

Renewable Energy 33 (2008) 15231531

Optimization of a solar-assisted drying system for drying bananas


R. Smitabhindu
a,
, S. Janjai
b
, V. Chankong
c
a
Department of Industrial Engineering, Kasetsart University, Bangkok 10900, Thailand
b
Department of Physics, Faculty of Science, Silpakorn University, Nakhon Pathom 73000, Thailand
c
Department of Electrical Engineering and Computer Science, Case Western Reserve University, Cleveland, OH, USA
Received 26 March 2007; accepted 18 September 2007
Available online 19 November 2007
Abstract
This paper presents a mathematical model for optimal design of a solar-assisted drying system for drying bananas. The optimization
model consists of a simulation model of a solar-assisted drying system combined with an economic model. The simulation model is
composed of two systems of differential equations: one for the collector and other for the drying cabinet. These systems of the differential
equation were solved using the nite difference method. Values of the model parameters were determined experimentally. A computer
program in FORTRAN was developed to simulate the performance of the drying system. The model was validated by comparing the
simulation results with the experimental results and they were in good agreement. This simulation model was used for the optimization of
the solar-assisted drying system. An economic model was formulated to calculate the annual drying cost. The optimization problem was
dened as the optimization of the geometry and operational parameters of the drying system so as to minimize the drying cost per unit of
dried product. Currently used collector area and the air recycle factor were considered as the parameters for basic mode of operation of
the drying system. The adaptive pattern search technique was adopted to nd the optimum values of the solar collector area and the
recycle factor. The optimum values of the collector area and the recycle factor were found to be 26 m
2
and 90%, respectively. The
computer program developed in this study can be used to optimize similar drying systems.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Solar collector; Solar drying; Banana; Simulation; Optimization
1. Introduction
Bananas are a potential cash crop in Thailand and dried
banana is a popular snack food in this country. Huge
amount of bananas are consumed in every part of
Thailand. There is a bright future for dried bananas for
both internal consumption and export. The international
market can also be promoted with dried bananas if
the quality of the dried bananas can be maintained at
international standard. However, most of bananas in
Thailand are dried with the natural sun drying and dried
bananas are usually contaminated by insects.
The problem of drying bananas can be solved by using
solar drying technology. This is to due the fact that dried
bananas production areas are usually situated in tropical
regions with receive abundant of solar radiation. Recently,
a solar-assisted fruit drying system have been developed in
a collaborative research between Innotech Company,
Germany and the Royal Chitralada Project of the King
of Thailand. This drying system is now used to produce
dried bananas for the demonstration purpose at the Royal
Chitralada Projects in Bangkok, Thailand. As the system
has been designed for drying general fruits and the design
parameters are not optimal for drying bananas, it is
necessary to optimize these design parameters.
Solar drying systems must be properly designed in order
to meet particular drying requirements of specic products
and to give satisfactory performance with respect to
requirements. Designers should investigate the basic
parameters such as collector area, drying air recirculation
factor and product quality. However, full scale experiments
for different products, drying seasons and system cong-
urations are sometime costly and may not be feasible.
A simulation model is a valuable tool for predicting the
dryer performance and it is also least costly and less time
ARTICLE IN PRESS
www.elsevier.com/locate/renene
0960-1481/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2007.09.021

Corresponding author. Tel./fax: +66 2 2819255.


E-mail address: sermjanjai@yahoo.co.th (R. Smitabhindu).
ARTICLE IN PRESS
Nomenclature
A
c
collector area (m
2
)
A
max
maximum limit of collector area (m
2
)
C
b
specic heat of the absorber (J/kg-K)
C
c
specic heat of the cover (J/kg-K)
C
f
specic heat of the air (J/kg-K)
C
p
specic heat of the product (J/kg-K)
C
T
total cost (USD)
C
v
specic heat of vapor (J/kg-K)
C
w
specic heat of water (J/kg-K)
C
annual
annual cost (USD/yr)
C
cabinet
cost of the dryer cabinet (USD)
C
capital
total capital cost (USD)
C
collector
collector cost (USD)
C
electric
cost of the electricity consumption (USD/kWh)
C
gas
cost of gas consumption per year (USD)
C
labor
costs of construction and installation (USD)
C
labor,op
labor cost for operating the dryer (USD)
C
maint
maintenance cost (USD)
C
maint,i
maintenance cost in the ith year (USD)
C
op
operating cost of the dryer (USD)
C
op,i
operating cost in the ith year (USD)
C
unit,col
unit cost of the solar collector (USD/m
2
)
C
unit,elect
unit cost of the electricity (USD/kWh)
C
unit,gas
unit cost of the gas (USD/kg)
D
c
distance between the cover and the absorber of
the collector (m)
D
t
height of the gap between the trays (m)
D
h
hydraulic diameter (m)
G specic mass ow rate of the air in the collector
(kg/s-m
2
)
h
1
enthalpy of the air from the collector (J/kg)
h
2
enthalpy of the recycle air (J/kg)
h
3
enthalpy of the air to the drying chamber (J/kg)
h
fg
latent heat of vaporization (J/kg)
h
w,ca
convective heat loss coefcient of the cover
caused by wind (W/m
2
-K)
h
c,bf
convective heat transfer coefcient between the
absorber and the air in the collector (W/m
2
-K)
h
c,cf
convective heat transfer coefcient between the
cover and the air in the collector (W/m
2
-K)
h
c,fc
convective heat transfer coefcient between the
air in the collector and the cover (W/m
2
-K)
h
c,pf
convective heat transfer coefcient between the
product and the drying air (W/m
2
-K)
h
c,fb
convective heat transfer coefcient between the
air in the collector and the absorber (W/m
2
-K)
h
r,bc
radiative heat transfer coefcient between the
cover and the absorber (W/m
2
-K)
h
r,cs
radiative heat transfer coefcient between the
cover and the sky (W/m
2
-K)
H humidity ratio of the drying air (dimensionless)
H
1
humidity ratio of the air from the collector
(dimensionless)
H
2
humidity ratio of the recycle air (dimensionless)
H
3
humidity ratio of the air to the drying chamber
(dimensionless)
i
f
ination rate (decimal)
i
in
interest rate (decimal)
I
t
solar radiation incident on the collector (W/m
2
)
k thermal conductivity of the air (W/m1C)
_ m
1
mass ow rate of the air from the collector
(kg/s)
_ m
2
mass ow rate of the recycle air (kg/s)
_ m
3
mass ow rate of the air to the drying chamber
(kg/s)
M moisture of the product at time t
(kg
water
/kg
solid
)
M
e
equilibrium moisture content (kg
water
/kg
solid
)
M
0
initial moisture content (kg
water
/kg
solid
)
M
dry product
amount of dried products obtained from
the system per year (kg)
M
elect
amount of the electricity consumption per year
(kWh/yr)
M
gas
amount of gas consumption per year (kg)
N life span of the dryer (years)
Nu Nusselt number (dimensionless)
rh relative humidity of the air (%)
R
c
recycle factor (%)
Re Reynolds number (dimensionless)
R
max
maximum limit of the recycle factor (%)
t drying time (s)
T temperature of the drying air (K)
T
a
temperature of ambient air (K)
T
b
temperature of the absorber (K)
T
c
temperature of the cover (K)
T
f
temperature of the air in the collector (K)
T
f
temperature of the drying air (K)
T
p
temperature of the product (K)
T
s
sky temperature (K)
U
b
heat loss coefcient through the back side of
the collector to ambient air (W/m
2
-K)
V speed of the air in the collector (m/s)
W
c
width of the collector (m)
W
t
width of tray (m)
x distance (m)
a
c
absortance of the cover (dimensionless)
r
f
density of the air (kg/m
3
)
r
b
density of the absorber (kg/m
3
)
r
c
density of the cover (kg/m
3
)
r
s,p
density of product (kg/m
3
)
e
b
emittance of the absorber (dimensionless)
e
c
emittance of the cover (dimensionless)
s StefanBoltzmann constant (W/m
2
-K
4
)
d
b
thickness of the absorber (m)
d
c
thickness of the cover (m)
n viscosity of the air (m
2
/s)
(ta) transmittanceabsorptance product of the sys-
tem consisting of the cover and the absorber
(dimensionless)
R. Smitabhindu et al. / Renewable Energy 33 (2008) 15231531 1524
consuming. Again, the simulation model is essential to
optimize the dryer component and drying process, and
optimization techniques can be used for optimal design of
solar drying systems [1,2].
Several studies have been reported on the optimization
of natural convection and forced convection solar drying
systems [37]. Very few studies have been carried out for
simulation and optimization of solar-assisted drying system
for tropical fruits. The purposes of this study are to
formulate a mathematical model of a solar-assisted drying
system and to use this model to optimize the design
parameters of the system for drying bananas.
2. Description of the system
The solar-assisted forced convection drying system has
been installed at the Royal Chitralada Projects, Dusit
Palace in Bangkok, Thailand to demonstrate the potenti-
ality of solar-assisted convection dryer for production of
quality dried fruits. This drying system has a capacity of
drying 250 kg of bananas and it is being used for drying
bananas since the year 2000.
This drying system consists of two main parts namely:
(1) the solar collector, and (2) the drying cabinet (Fig. 1).
The solar collector is placed on the rooftop of the drying
building and the drying cabinet is inside this building. The
solar collector consists of polyurethane back insulator and
cover glass. There is an air gap between the cover glass and
the insulator through which ambient air is sucked from both
ends of the roofs through the collectors. The air is sucked at
the midpoint of the collector and supplied into the drying
cabinet with an auxiliary heat source using an LPG gas
burner. Each part of the collector was designed with a
modular concept. The parts of the collector such as
insulation and cover glass are in modular form so that
these can be easily transported and connected to each other.
The drying cabinet is a tray type and accommodates 15
trays in stacks with a total drying area of 8 m
2
and the
dimension of a tray is 1 m2 m1.5 m. This drying
cabinet has been specially designed in such a way that hot
air is guided to ow parallel through the products placed in
the trays in the stacks. This design has the advantage that
air temperatures in the cabinet are uniform. Ambient air
preheated by solar collector is sucked by an electrical
blower and additional heat if needed, is supplied by an
LPG gas burner. Then heated air is supplied to the cabinet.
The picture of the solar collector and the drying cabinet are
depicted in Fig. 2.
The ambient air from both ends of the collector is sucked
by the blower to move through the air gap of the collector.
The collector receives the energy from solar radiation and
heats air inside the collector. The polyurethane back
insulator is used to reduce heat loss. The hot air from the
solar collector is moved through the air duct to the LPG gas
burner placed on the top of the cabinet. This air is heated up
again by the gas burner if needed and forced to the drying
cabinet for drying the product. Part of this drying air is
recycled and circulated in the cabinet to save energy and the
other part of this air is released to ambient surroundings.
ARTICLE IN PRESS
LPG gas burner
Door
Air flap
Tray
m
3
, H
3
.
Exhausted Air
Blower
H
x
T
f
Insulator
Cover glass Solar collector
x
Bananas
60C
dx
dH
H
x
+ x
dx
dT
T
f
+ x
m
2
, H
2
.
m
1
, H
1
.
Drying cabinet
Fig. 1. Schematic diagram of the solar-assisted drying system.
R. Smitabhindu et al. / Renewable Energy 33 (2008) 15231531 1525
3. Modeling of the solar-assisted drying system
3.1. Solar collector
Considering an element dx of collector at a distance x
from the inlet as shown in Fig. 3. Energy balances on the
collector components give the following equations:
3.1.1. Cover of the collector
The heat balance in the element (x, x+dx) with the width
of the cover of W
c
in time Dt yields the following equation:
r
c
d
c
DxW
c
C
c
T
c

dT
c
dt
Dt
_ _
r
c
d
c
DxW
c
C
c
T
c
h
c;fc
DxW
c
T
f
T
c
Dt h
r;bc
DxW
c
T
b
T
c
Dt
h
w;ca
DxW
c
T
a
T
c
Dt h
r;cs
DxW
c
T
s
T
c
Dt
DxW
c
a
c
I
t
Dt. 1
This equation can be expressed as
r
c
d
c
C
c
dT
c
dt
h
r;bc
T
b
T
c
h
c;fc
T
f
T
c

h
w;ca
T
a
T
c
h
r;cs
T
s
T
c
a
c
I
t
.
2
3.1.2. Air in the collector
The heat balance in the air inside the collector in the
element (x, x+dx) yields
D
c
GW
c
C
f
T
f

dT
f
dx
Dx
_ _
Dt D
c
GW
c
C
f
C
f
HT
f
Dt
DxWh
c;fc
T
c
T
f
Dt DxW
c
h
c;bf
T
b
T
f
Dt.
3
This equation can be rewritten as
D
c
GC
f
dT
f
dx
h
c;cf
T
c
T
f
h
c;bf
T
b
T
f
. (4)
3.1.3. Absorber of the collector
The heat balance in the element (x, x+dx) of the
absorber and of width W in time Dt also yields the
equation:
r
b
DxW
c
d
b
C
b
T
b

dT
b
dt
Dt
_ _
r
b
DxW
c
d
b
C
b
T
b
h
c;fb
DxW
c
T
f
T
b
Dt h
r;bc
DxW
c
T
c
T
b
Dt
u
b
DxW
c
T
a
T
b
Dt DxW
c
taI
t
Dt. 5
ARTICLE IN PRESS
Fig. 2. Picture of the drying cabinet inside the building (a) and the solar collector on the roof of the same building (b).
Convection
Radiation
Air inlet
Absorber
Back insulator
x dx
Glass cover
h
r, cs
h
r, bc
h
w, ca
h
c, cf
h
c, bf
T
a
T
c
T
f
T
b
Conduction
Fig. 3. Schematic diagram showing heat transfers in the solar collector.
R. Smitabhindu et al. / Renewable Energy 33 (2008) 15231531 1526
This equation can be rearranged as
r
b
d
b
c
b
dT
b
dt
h
c;bf
T
f
T
b
h
r;bc
T
c
T
b

U
b
T
a
T
b
taI
t
. 6
Radiation heat transfer coefcient, h
r,cs
between the
cover and the sky is given as [8]:
h
r;cs

c
sT
2
c
T
2
s
T
c
T
s
, (7)
and T
s
0:552T
1:5
a
.
The radiative heat transfer coefcient, h
r,bc
between the
cover and the absorber is [8]
h
r;bc

sT
2
b
T
2
c
T
b
T
c

c
1
. (8)
Convective heat transfer coefcient, h
w,ca
at the upper
surface of the cover from ambient air is computed from the
following equation [8,9]:
h
w;ca
5:7 3:8V. (9)
The convective heat transfer in the inner surface of
the cover from the air inside the collector is computed
from the following correlation suggested by Kays and
Crawford [10]:
Nu 0:0158Re
0:8
. (10)
The Reynolds number is given by
Re
D
h
Vr
f
n
. (11)
The hydraulic diameter, D
h
is computed from the
relationship
D
h

4W
c
D
c
2W
c
D
c

. (12)
The convective heat transfer coefcient, h
c
,
bf
is
computed from Nusselt number Nu obtained from
Eq. (10), hydraulic diameter (D
h
) from Eq. (12) and the
heat transfer coefcient, h
c,bf
from the following equation:
h
c;bf

Nuk
D
h
. (13)
The value of h
c,cf
was assumed to be equal to h
c,bf
because of the similarity of the materials and geometry of
media: air-cover and absorber-air.
3.2. Method of solution for the collector model
The collector model (Eqs. (2), (4) and (6)) is a system of
differential equations which are difcult to solve using
analytical method. This system of equations was solved
implicitly using the nite difference method. The system of
equations in the nite difference form is as follows:
Equation for the cover
r
c
d
c
C
c
T
c;tDt
T
c;t
Dt
h
r;bc
T
b;tDt
T
c;tDt
h
c;fc
T
f;tDt
T
c;tDt

h
w;ca
T
a;tDt
T
c;tDt

h
r;cs
T
s;tDt
T
c;tDt
a
c
I
t;tDt
. 14
Equation for the air
DGC
f
0:5
T
f;xDx;tDt
T
f;xDx;tDt
2Dx

T
f;xDx;t
T
f;xDx;t
2Dx
_ _
h
c;bf
T
b;x;tDt
T
f;x;tDt
h
c;cf
T
c;x;tDt
T
f;x;tDt
.
15
Equation for the absorber
r
b
d
b
C
b
T
b;tDt
T
b;t
Dt
h
c;bf
T
f;tDt
T
b;tDt
h
r;bc
T
c;tDt
T
b;tDt
U
b
T
a;tDt
T
b;tDt

htaiI
t;tDt
. 16
The length of the collector was divided into a number
of sections so that the properties of the material are
nearly constant within each section. The time interval
should be small enough for the air conditions to be
constant at the inlet to and exit from the section. But for
the economy of computing time, a compromise between
the acceptability of the results and the intervals must be
used.
The system of equations consisting of Eqs. (14)(16) is
expressed in the following form for the interval Dt for the
entire length of collector unit:
a
11
a
12
a
13
a
21
a
22
a
23
a
31
a
32
a
33
_

_
_

_
T
c
T
f
T
b
_

_
_

_
b
1
b
2
b
3
_

_
_

_. (17)
This system of equations is a set of implicit equations for
the time interval Dt for the entire length of the drying unit
and was solved using the GaussJordan elimination
method.
3.3. Formulation of the model for the drying cabinet
The bananas were placed in a single layer in a series of
trays, which are parallel (Fig. 1) and the bananas in the
trays almost equally exposed to a uniform temperature and
ARTICLE IN PRESS
R. Smitabhindu et al. / Renewable Energy 33 (2008) 15231531 1527
only thin layer equation is sufcient to describe the drying
behavior inside the dryer.
3.3.1. Energy balance of the drying air
The change in enthalpy of the drying air is equal to the
convective heat transfer to the product:
D
t
GW
t
C
f
C
v
H T
f

dT
f
dx
Dx
_ _
Dt
D
t
GW
t
C
f
C
v
HT
f
Dt
DxW
t
h
c;pf
T
p
T
f
Dt. 18
This equation can be rewritten as
D
t
GC
f
C
v
H
dT
f
dx
h
c;pf
T
p
T
f
. (19)
3.3.2. Energy balance of the product inside the drying
cabinet
The change in enthalpy of the product is equal to the
convective heat transfer to the product minus heat supplied
to evaporate the moisture to the air. This energy balance
yields:
DxW
t
r
s;p
C
p
C
w
M T
p

dT
p
dt
Dt
_ _
DxW
t
r
s;p
C
p
C
w
MT
p
h
fg
D
t
G
dH
dx
DxDt C
v
T
f
T
p

D
t
G
dH
dx
DxDt DxW
t
h
c;pf
T
f
T
p
Dt. 20
This equation can be rearranged as
r
s;p
C
p
C
w
M
dT
p
dt
h
fg
C
w
T
p
T
f
D
t
G
dH
dx
h
c;pf
T
f
T
p
. 21
3.3.3. Moisture balance of the air
Moisture balance states that the moisture gain by the air
is equal to the moisture loss by the product:
D
t
GW
t
Dt H
dH
dx
Dx
_ _
D
t
GW
t
DtH
r
s;p
DxW
t
M r
s;p
DxW
t
M
dM
dt
Dt
_ _
. 22
Rearrangement of this equation yields
D
t
G
dH
dx
r
s;p
dM
dt
. (23)
3.3.4. Moisture balance of the product to be dried
The moisture content of the product was expressed by an
appropriate thin layer equation and the following empirical
equation was tted to experimental data [11]:
M M
e
M
0
M
e
Ae
Bt
, (24)
where the constant A and B were expressed as functions of
relative humidity (rh) in percent and temperature (T) in 1C
in the following form:
A a
0
a
1
rh a
2
Tt a
3
rh
2
a
4
T
2
, (25)
B b
0
b
1
rh b
2
T b
3
rh
2
b
4
T
2
, (26)
where a and b are empirical constants [11]
3.3.5. Recirculation
A part of the exhaust air from the dryer was recycled and
this amount was xed by design. The original design is for
a recirculation factor of 95%. The humidity and the
enthalpy of the mixed air are computed from the following
equations:
_ m
3
_ m
1
_ m
2
, (27)
_ m
3
H
3
_ m
1
H
1
_ m
2
H
2
, (28)
_ m
3
h
3
_ m
1
h
1
_ m
2
h
2
. (29)
3.4. Method of solution for the drying cabinet model
The banana bed was divided into a number of sections
(x j Dx, j 1, 2, 3, y) along the length of the drying
trays. The drying time was also divided into a number of
intervals (t i Dt, i 1, 2, 3, y). On the basis of the air
temperature, relative humidity and airow at the entry of
the drying cabinet, the drying constant (B) and equilibrium
moisture content (M
e
) of the banana were computed.
Using these B and M
e
values, the changes of moisture
content of bananas, DM, for a time interval, Dt, were
calculated using Eq. (24). Using the recent value of air
temperature and drying rate, the product temperature of
rst section of the drying trays was computed using
Eq. (21). On the basis of the recent value of product
temperature, the air temperature inside the rst section was
estimated using Eq. (19). The change in air humidity was
computed using Eq. (23). This process was repeated
section-by-section until the end of the section was reached.
This process was then repeated for each time increment.
When air relative humidity exceeds 99%, the condensation
routine deposits back the moisture from the over saturated
air. Air and banana temperature were adjusted for this
condensation [12]. The numerical solution was pro-
grammed in FORTRAN language.
4. Model validation
To build up condence in the models, drying experi-
ments were carried out using the drying system and
the experimental results were compared with simulated
ARTICLE IN PRESS
R. Smitabhindu et al. / Renewable Energy 33 (2008) 15231531 1528
results. The measured collector outlet air temperatures
were compared with the simulated results. Fig. 4 shows
that the patterns of the variations of the simulated
temperatures agree well with those of the measured
temperatures.
The measured moisture contents of bananas were also
compared with the simulated moisture contents. Fig. 5
shows that the simulated moisture contents agree well with
the measured results. This simulation model will be used
for the optimization of the dryer components and drying
process.
5. Economic model of the solar-assisted drying system
The various cost of drying system was estimated based
on the real cost of the construction and operation of this
drying system because the system has been used since the
year 2000. The capital cost of the system consists of
the cost of the collector, drying cabinet and labor cost for
the construction and installation. The capital cost of this
drying system is listed as follows:
Solar collector
Drying cabinet
Labor (construction and installation)
2000 USD
2833 USD
500 USD
(1 USD 40 Bahts)
The collector material costs are those for glass and
polyurethane back insulator. The drying cabinet material
costs are those for the trays, LPG burner, air duct, blower,
system control, insulation materials and structural materials.
The cost of the collector (C
collector
) can be expressed as
C
collector
C
unit;col
A
c
, (30)
where C
unit, col
is the unit cost of the collector and A
c
is the
collector area.
The total capital cost for the solar-assisted drying system
(C
T
) is given by next equation
C
T
C
collector
C
f
, (31)
where C
f
is the material cost of the drying cabinet and the
labor cost for construction and installation.
The annual cost calculation method proposed by
Audsley and Wheeler [13] yields:
C
annual
C
T

N
i1
C
maint;i
C
op;i
o
i
_ _
o 1
oo
N
1
_ _
,
(32)
where C
annual
is the annual cost of the system. C
maint,i
and
C
op,i
are the maintenance cost and the operating cost at the
year i, respectively. o is expressed as
o 100 i
in
=100 i
f
, (33)
ARTICLE IN PRESS
0
20
40
60
80
100
8:00 11:00 14:00 17:00 8:00 11:00 14:00 17:00 8:00 11:00 14:00 17:00
Time, hr
T
e
m
p
e
r
a
t
u
r
e
,

C
Measurement
Simulation at R = 95 %
9 October, 2003 10 October, 2003 11 October, 2003
Fig. 4. Predicted and observed variation in the collector temperature.
0.0
1.0
2.0
3.0
4.0
8:00 11:00 14:00 17:00 8:00 11:00 14:00 17:00 8:00 11:00 14:00 17:00
Time, hr
M
o
i
s
t
u
r
e

c
o
n
t
e
n
t
,

d
b
Measurement
Simulation at R = 95 %
9 October, 2003
10 October, 2003 11 October, 2003
Fig. 5. Simulated and observed moisture content of bananas for the drying test.
R. Smitabhindu et al. / Renewable Energy 33 (2008) 15231531 1529
where i
in
and i
f
are the interest rate and the ination rate,
respectively.
The maintenance cost of the rst year was assumed to be
1% of the capital cost.
The operating cost consists of the gas consumption cost
(C
gas
), electricity consumption cost (C
electric
) and the labor
cost for operating the system (C
labor,op
). This cost can be
written as follows:
C
op
C
gas
C
electric
C
labor;op
. (34)
The cost of the gas consumption is calculated as follows:
C
gas
C
unit;gas
M
gas
, (35)
where C
unit,gas
is the unit cost of the gas and M
gas
is the
mass of the gas used per year.
The cost of the electricity consumption can be calculated
as
C
electric
C
unit;elect
M
elect
, (36)
where C
electric
is the unit cost of the electricity and M
electric
is the amount of the electricity used per year.
The annual cost per unit of dried product is called the
drying cost (Z). It can be written as
Z
C
annual
M
dry product
, (37)
where M
dry product
is the dried product obtained from this
drying system per year.
Now the optimization of the system component and
recycle factor can be stated mathematically as
Minimize Z
C
T

N
i1
C
maint;i
C
op;i
o
i
_ _
o1
oo
N
1
_ _
M
dry product
,
(38)
subject to the constraints
0pA
c
pA
max
,
0pR
c
pR
max
,
where A
max
is the maximum collector area. As the collector
is on the roof of the building, A
max
is the available area of
the roof (A
max
50 m
2
), R
max
is the maximum recycle
factor (R
max
100%).
The adaptive pattern search technique was used for this
constrained optimization problem [14]. This was found to
be very effective as reported by Bala and Woods [6] and
Hussain et al. [7]. The technique consists essentially of an
exploratory search and a pattern search under constrained
conditions. The optimization program was written in
FORTRAN.
6. Results and discussions
As the material properties are usually xed and the
environmental parameters cannot be controlled, the para-
meters which we can optimize are the geometrical
parameters and the operating parameters. To identify the
parameters, which are necessary to optimize, we simulate
the performance of the optimization model based on the
physical simulation model and economic model. The
parameters used in this simulation were taken from the
drying system installed at the Royal Chitralada Projects.
Solar radiation data were taken from Solar Energy
Research Laboratory at Silpakorn University, Nakhon
Pathom located at about 60 km from the Royal Chitralada
Projects since this Laboratory maintains a data base of
solar radiation for a long period. The relative humidity and
ambient air temperature were also taken from the same
Laboratory. These data are in the form of typical
meteorological year (TMY). The optimum values obtained
from the optimization model and actual values used in the
real drying system at the Royal Chitralada Project are
shown in Table 1. It is found from Table 1 that the
collector area of the real drying system is 18 m
2
less than
the optimum (26 m
2
). The optimum recycle factor is 90%
which is slightly lower than the actual recycle factor in use
(95%). The optimum drying cost (0.225 USDper kg) is half
of the actual drying cost (0.45 USDper kg).
7. Conclusions
The solar-assisted drying system in the Royal Chitralada
Projects has been simulated to provide design data and it
has also been optimized for the solar collector area and the
air recycle factor for drying bananas using the simulation
model combined with an economic model.
The simulation model was validated by comparing the
simulation results with the experimental results. The
patterns of the variations of the simulated temperatures
agree well with those of the measured temperatures. Also
the simulated moisture contents agree well with the
measured results. This simulation program was used as a
tool for the optimization process. The adaptive pattern
search technique was use to optimize the collector area
and the air recycle factor. The optimum values of the
collector area of 26 m
2
and the optimum recycle factor of
90% were obtained from the optimization process. These
optimum values resulted in the minimum drying cost with
0.225 USDper kg.
As this drying system is for the demonstration of drying
technology for small-scale food industry, the results
obtained from this work can be used as a guideline
for the construction of other units of the solar drying
system. The simulation and the optimization programs
ARTICLE IN PRESS
Table 1
The optimum and the actual values of the solar collector area (A
c
), the
recycle factor (R
c
) and the drying cost
Optimum value Actual value
Area of solar collector (m
2
) 26 18
Air recycle factor (%) 90 95
Drying cost (USD/kg) 0.225 0.45
R. Smitabhindu et al. / Renewable Energy 33 (2008) 15231531 1530
developed in this work can be applied to optimize similar
drying systems.
Acknowledgement
The rst author was awarded a scholarship from the Her
Royal Highness Princess Maha Chakri Sirindhorn. We
would like to acknowledge for this support. The authors
would also like to thank Dr. A. Esper and Mr. Thada
Keawprasert for the valuable advice and supports.
References
[1] Bala BK. Solar drying systems: simulation and optimization.
Udaipur, India: Agrotech Publishing Academy; 1998.
[2] Steinfeld A, Segal I. Simulation model of solar thin layer drying
process. Dry Technol 1986;4(4):53554.
[3] Radajewski W, Jolly P, Abawi GY. Optimization of solar grain
drying in a continuous ow dryer. J Agric Eng Res 1987;38:12744.
[4] Radajewski W, Jolly P, Wollin AS. Optimization of solar drying
system. In: Proceedings of the 11th international congress on
agricultural engineering, Dublin, 1989. p. 22438.
[5] Radajewski W, Graydon D, McGaha EJ. Optimization of solar hay
drying in a step-ow dryer. Trans ASAE 1990;33:142331.
[6] Bala BK, Woods JL. Optimization of the natural-convection solar
drying system. Energy 1995;20(4):28594.
[7] Hussain MA, Woods JL, Bala BK. Optimization of solar tunnel drier
for drying of chilli without color loss. Renew Energy 2005;30:72942.
[8] Dufe JA, Beckman WA. Solar engineering of thermal process. New
York: Wiley; 1991.
[9] McAdams WH. Heat transmission. 3rd ed. New York: McGraw-Hill;
1954.
[10] Kays WM, Crawford ME. Convective heat and mass transfer.
New York: McGraw-Hill; 1980.
[11] Smitabhindu R. Thin layer model of banana. Paper presented at the
ADC conference, Asian Institute of Technology, Thailand, 2004.
[12] Bala BK. Deep bed drying of malt. PhD thesis, University of
Newcastle upon Tyne, UK, 1983.
[13] Audsley E, Wheeler J. The annual cost of machinery calculated actual
cash ows. J Agric Eng Res 1978;23:189201.
[14] Taubert WH. A search decision rule for the aggregate scheduling
program. Manage Sci 1968;14B:34359.
ARTICLE IN PRESS
R. Smitabhindu et al. / Renewable Energy 33 (2008) 15231531 1531