You are on page 1of 9

Received: 15 July 2017 | Revised: 1 March 2018 | Accepted: 18 March 2018

DOI: 10.1111/jfpp.13649

ORIGINAL ARTICLE

An integrated heat-transfer-fluid-dynamics-mass-transfer model


for evaluating solar-dryer designs

Sappinandana Akamphon1 | Sittha Sukkasi2 | Korkiat Sedchaicharn2

1
Faculty of Engineering, Thammasat
University, Pathumthani, Thailand
Abstract
2
Design and Industry Solutions Lab, MTEC, This work aims to facilitate the design-improvement process of solar dryers by a computational
National Science and Technology model that accounts for the interrelationships between the conditions of the drying product (tem-
Development Agency, Pathumthani, perature and moisture content), the airflow inside the dryer (temperature, humidity, and velocity),
Thailand
and the surrounding conditions (temperature, humidity, and solar radiation). The model considers
Correspondence not only how the dryer and surrounding affect the drying of the product but also how the condi-
Sittha Sukkasi, Design and Industry tions of the product itself affect the conditions of the airflow inside the dryer and, by extension, its
Solutions Lab, MTEC, National Science and performance as well. To account for such intricate interrelationships, the model leverages and inte-
Technology Development Agency,
grates three physical domains: fluid dynamics, heat transfer, and mass transfer. The model
Pathumthani, Thailand.
Emails: sitthas@mtec.or.th; automatically simulates the natural convection inside the dryer and eliminates the need for obtain-
sittha@alum.mit.edu ing the experimental values of the temperature, humidity, and velocity of the internal airflow for
calculating the moisture level of the product. It can be used to quantify the effects of mixed-mode
solar dryer designs on their drying performance. Simulated results were validated against experi-
mental data and found to be reasonably accurate. Moreover, the model offers insights into how
the complex airflow inside the dryer can affect the drying of the product.

Practical applications
Experiments that are required to understand, compare, and improve the drying performance of dif-
ferent solar dryer designs can be impractical because of long operation time of solar drying and
variability of weather conditions, making a typical design process time-consuming. This work aims
to facilitate the design-improvement process. The model can quantify the effects of mixed-mode
solar dryer designs on their drying performance. The model can also offer useful insights for solar-
dryer designers. For instance, results from a case study show that the velocity and humidity of the
internal airflow could have stronger influence than that of the temperature on the drying, suggest-
ing that a solar dryer design that aims only to maximize airflow temperature may not lead to
optimal drying, and that airflow temperature should not be used as the sole benchmark for solar
dryer design.

NOMENCLATURE u air velocity vectors (m/s)


k thermal conductivity (W/m K)
Symbols Q heat transfer from sources other than viscous heating (W/m3)
Qrad radiative heat transfer (W/m2)
A surface area
a solar absorptivity
Aeff effective surface area normal to solar radiation
I incident solar radiation (W/m2)
q density (kg/m3)
r Stefan-Boltzmann constant (5.6703 3 1028 [W/m2 K4])
t time (s)
E thermal emissivity
C specific heat capacity (J/kg K)
h coefficient of heat convection (W/m2 K)
T temperature (K)
Nu Nusselt number

J Food Process Preserv. 2018;e13649. wileyonlinelibrary.com/journal/jfpp V


C 2018 Wiley Periodicals, Inc. | 1 of 9
https://doi.org/10.1111/jfpp.13649
2 of 9 | AKAMPHON ET AL.

performance. Validity of the simulated results is compared against


L characteristic length (m)
Lheat latent heat of vaporization of water (40,660 [J/mol]) experimental and simplified equation of thin-layer drying results.

Re Reynolds number
Pr Prandtl number 2 | LITERATURE REVIEW
2
g gravitational acceleration (9.8 m/s )
P pressure (N/m2) Solar drying has been extensively studied both experimentally and
Rair specific gas constant for air (287 J/kg K) computationally. Several studies explored the use of mathematical
c concentration of moisture (mol/m3) models to study the performance and optimization of solar dryer
2 designs. Generally, the existing work could be divided into three groups
D diffusivity of moisture (m /s)
2 based on the physical domains in which their mathematical models
D0 effective diffusivity at an infinitely high temperature (m /s)
Ea activation energy (J/mol) involved. The models in the first group focused on the drying charac-
R ideal gas constant [8.3144621 J/mol K] teristics of the products in terms of the diffusion of moisture, with the
m dynamic viscosity (Pa s) analysis boundaries set around only the products. Based on experimen-
n surface unit vector tal results, regressions were used to determine the dependence of the
N moisture flux (mol/m2 s) drying rate on various parameters, such as the inlet air temperature,
solar radiation intensity, and product size. Examples of the models in
Subscriptions
this group included various efforts to determine drying parameters in
x x direction thin-layered agricultural and marine products (Akpinar, Bicer, & Yildiz,
y y direction 2003; Basunia & Abe, 2001; To
grul & Pehlivan, 2002; Ertekin & Yaldiz,
a ambient 2004; Yaldiz, Ertekin, & Uzun, 2001; Yald n, 2001; Midilli &
yz & Erteky
s solar dryer cover surface Kucuk, 2003). These models did not take into account the designs and
i air inside the dryer configurations of the dryers.
p product The second group of the existing models involved both the mois-
0 initial ture diffusion of the products and the heat transfer inside the dryers.
These models assume uniform airflows and use conservation of energy
1 | INTRODUCTION to calculate the drying rate and inlet and outlet temperatures. For
example, Jain modeled the performance of multi-tray crop drying using
Solar drying is the process of using solar radiation to generate heat to an inclined multi-pass solar air heater with a built-in thermal storage.
facilitate moisture evaporation from the products that are the subject The model was used to evaluate the thermal performance of a flat plate
of the drying. The method and equipment have been continually solar air heater for crop drying. It was also able to predict the moisture
improved over the years. As it does not rely on fossil fuel or electricity, content, crop temperature, and crop-drying rate (Jain, 2005; Jain &
solar drying is one of the cleanest preservation processes. Jain, 2004). A model for deep bed drying of rice was developed by con-
There are many types of solar dryers (Fudholi, Sopian, Ruslan, sidering different air-flow configurations over absorber plates (Dubey
Alghoul, & Sulaiman, 2010). Direct solar dryers rely on direct solar & Pryor, 1996). It could calculate the heat removal factor, overall loss
exposure of target products to generate heat and evaporate the mois- coefficient, and top loss coefficient. A solar dryer was theoretically
ture. Indirect solar dryers use solar radiation to heat incoming airflow, modeled and used to compare the benefits of various types of solar
which is then used to dry the products. Mixed-mode dryers combine collectors on the drying rate (Hachemi, Abed, & Asnoun, 1998). It
the direct exposure and heated airflow to dry the products more effec- was found that offset-plate-fins collector provided the best perform-
tively, allowing for reduced operating time. ance. Ratti and Mujumdar (Ratti & Mujumdar, 1997) modeled the
Design optimization and performance improvement of solar dryers effects of shrinkage and variable inlet temperature during the drying
can be difficult due to the variability of environmental conditions and process in a cabinet solar dryer. The change of moisture content
long operating duration. Unless two or more models are tested simulta- along the length of a continuous solar tunnel dryer was considered
neously side-by-side, performance comparison of different designs is in the work by Condorı and Saravia (2003), investigating the condi-
hard to quantify. tions for efficiency improvement. They found a linear relationship
This work endeavored to facilitate solar dryer design by construct- between the incident solar radiation and dryer output temperature.
ing a mathematical model to evaluate drying performance of a product Janjai et al. studied experimental and modeling performance of solar
in a solar dryer. By integrating three physical domains—fluid dynamics, drying of rosella flower and chili using a rooftop dryer (Janjai,
heat transfer, and mass transfer—the model can comprehensively simu- Srisittipokakun, & Bala, 2008). In these models, the assumption of
late the solar drying process. The model explicitly represents natural uniform airflows, however, can oversimplify the heat transfer and
convection inside the dryer and thereby it is able to distinguish drying drying processes inside the dryer.
performances at different locations inside the dryer. This allows for a The third group of models applied computational fluid dynamics
more in-depth design optimization to further improve drying (CFD) to drying ovens, not specifically solar dryers. Mathioulakis et al.
AKAMPHON ET AL. | 3 of 9

used CFD to simulate air movement inside a batch-type tray dryer. It


was found that the degree of dryness of fruits or vegetables depended
on their positions inside the dryer due to the lack of spatial homogene-
ity of air velocities (Mathioulakis, Karathanos, & Belessiotis, 1998). A
study of airflow using CFD in an industrial sausage dryer was conducted
by Mirade and Daudin (Mirade & Daudin, 2000). The developed CFD
model was able to predict airflow patterns based on the filling level,
although there was still a wide discrepancy between the measured and
simulated air velocities. Margaris and Ghiaus (Margaris & Ghiaus, 2006)
used CFD to optimize the tray arrangement and inlet configuration
within a tray-drying chamber. Conjugate models of convective drying of FIGURE 1 Simplified geometric model of solar dryer with x, y, and
a porous flat plate were explored and investigated (Defraeye, Blocken, z directions
& Carmeliet, 2012; Lamnatou, Papanicolaou, Belessiotis, & Kyriakis,
2010), although these studies did not include an external heat source
the surface (A), and the effective surface area normal to the incident
such as solar radiation and assumed incompressible flow.
solar radiation (Aeff).
Conjugate CFD modeling is very powerful, yet it has not been
Aeff
applied to help determine the thermal and drying performance of solar Qrad 5aI 2rEðTsurface
4
2Tsurrounding
4
Þ (2)
A
dryers. This work applied the framework of a conjugate CFD model that
incorporated heat transfer, mass transfer (diffusion-convection equa- The dryer’s surfaces are exposed to the convective heat transfers from

tion), and computational fluid dynamics, but allowed for external heat the ambient surrounding and internal airflow, and as well as the radia-

source and compressible flow modeling to simulate the performance of tive heat from the sun, some of which is radiated back. It is assumed
mixed-mode solar dryers, which has not been previously attempted in that there is no heat flux across the outer surfaces of the dryer. The
solar dryer applications. The resulting model can be used to quantify, continuity equation for heat flux at the inner surface – internal airflow
compare, and optimize the performance of different solar dryers. of the dryer is

Aeff
2hsi ðTs 2Ti Þ1as I 2rEs ðTs4 2Ti4 Þ52n  ks rTs (3)
3 | METHODOLOGY A
It is assumed that the all surfaces except the bottom of the dryer are
3.1 | Formulation of an integrated heat-transfer-fluid- transparent, and thus do not absorb the solar radiation, rendering the
dynamics-mass-transfer (IHTFDMT) model of mixed- second term negligible. As the product is assumed to be a thin flat
mode solar dryers plate, the coefficients of heat convection (h) at the inner surface of the
dryer and product surface are dependent on the air velocity:
The mathematical model developed to study the thermal performance
and drying capability of mixed-mode solar dryers is a combination of Nu  k
h5
heat transfer equations, nonsteady Navier-Stokes theorem, and L
Nu50:664  Re1=2  Pr1=3 (4)
diffusion-convection equation. To limit the complexity, the problem
quL lC
was modeled as two-dimensional in which all properties along the z- Re5 ; Pr5
l k
direction in Figure 1 were assumed uniform, which is normally the case
The surfaces subsequently transfer the heat to the internal airflow
in solar dryers.
by convection, and the air then transfers the heat to the product. It is
3.1.1 | Heat transfer model assumed that the air does not absorb heat directly from solar radiation.
The general heat transfer equation for solar dryers is Therefore, the continuity equation for heat flux at the internal air
flow–product interface is
oT
qC 1qCu  rT5r  ðkrTÞ1Q (1)
ot Aeff
Np  nLheat 2hip ðTi 2Tp Þ1ap I 2rEp ðTp4 2Ti4 Þ52n  ðkp rTp Þ (5)
where the four terms represent the heat capacity of the medium, con- A

vective heat transfer, conductive heat transfer, and heat transfer from where Np is the moisture flux at the surface of the product. A deriva-
external sources other than viscous heating, respectively. (The para- tion for the flux is detailed in Section 3.1.3. A boundary condition is
metric symbols in all equations are listed in the Nomenclature section imposed at the inlet that
at the beginning.) In this case, the primary external heat source is solar Ti 5Ta (6)
radiation. The net radiative heat transfer on a surface due to solar radi-
ation is the difference between the absorbed solar radiation and the
radiative heat transfer from the surface to the surrounding. It depends 3.1.2 | Fluid dynamics model
on the incident solar radiation (I), solar absorptivity (a), thermal emissiv- As most solar dryer designs tend to be uniform in the direction of the
ity (E) of the surface, surface and ambient temperatures (T), the area of width of the solar dryer (the z-direction in Figure 1) with no
4 of 9 | AKAMPHON ET AL.

dependence of any property on z, the fluid dynamics model of the negligible effect on the drying process. The rate of moisture convection
internal airflow can be simplified into two dimensions. is so small relative to the volumetric flow rate of the air that the subse-
Continuity equation: quent change in internal air humidity should be minimum.

oq
1r  ðquÞ50 (7) 3.1.4 | Integration of models
ot
The heat transfer, fluid dynamics, and mass transfer models above
Momentum equation in the direction along the length of the solar dryer
can be integrated and solved concurrently. The three models are
(the x-direction):
interrelated. In the heat transfer model, the temperatures of the
  !
oux oux oux op o2 ux o2 ux surfaces of the dryer are affected by the properties of the materials,
q 1ux 1uy 52 1l 1 2 (8)
ot ox oy ox ox2 oy solar radiation, velocity of the internal airflow, and temperatures of

Momentum equation in the direction along the height of the solar dryer the surrounding and internal airflow. The temperature of the internal

(the y-direction): airflow is affected by the properties of air, its velocity, and tempera-
! tures of the dryer surfaces and products. The temperatures of the
 
ouy ouy ouy op o2 uy o2 uy products depend on the properties of the materials, solar radiation,
q 1ux 1uy 52 1l 1 2 1gðq2q0 Þ (9)
ot ox oy oy ox2 oy
and velocity and temperature of the internal airflow. In the fluid
The final term in Equation 9 represents the buoyancy force resulted dynamics model, the velocity of the air is affected primarily by its
from the change in temperature. The air density inside the dryer temperature and pressure. In the mass transfer model, the moisture
decreases with increasing temperature, creating buoyancy force that concentration of the internal airflow depends mainly on the air veloc-
pushes the air up toward the outlet. The dependence of air density on ity, while the moisture concentrations of the products are affected by
temperature can be described using the ideal gas expression: the effective diffusivity, which is driven by the temperatures and
properties of the materials. With this integrated IHTFDMT model, the
P
q5 (10) effects of the dryer designs on the drying performance can be com-
Rair T
prehensively simulated.
It is assumed that there is no slip along the internal walls of the solar
dryer.
3.2 | Model validation
3.1.3 | Mass transfer model
To validate the IHTFDMT model, three solar-drying experiments were
The drying process taking place in the dryer is assumed to be limited to conducted. Their initial conditions are input into the model and the
the product and the internal airflow. The moisture is transported by experimental and computational results are compared. Additionally,
convection in the internal airflow and by diffusion in the product. There the simplified equation of thin-layer drying, which is commonly used
is no creation of additional moisture in the dryer; the moisture in the to describe drying processes, was also used to provide a baseline
dryer is either in the internal airflow or product. comparison for the goodness of fit.
The mass transfer of moisture in the dryer is determined by the
convection-diffusion equation: 3.2.1 | Solar dryer for experiments

oc The solar dryer that was used in the experiments (Figure 2) was a
1r  ð2DrcÞ1u  rc50 (11)
ot passive, mixed-mode, thin-tunnel type, according to the classification
Continuity of moisture flux at the internal air flow–product interface by Leon (Leon, Kumar, & Bhattacharya, 2002). The dryer consisted
can be expressed as of two parts. The lower part served as an air inlet and solar-
radiation collector. The upper part served as a drying space, second-
Np  n5Ni  n (12)
ary solar-radiation collector, and air outlet. Ambient air entered
where the fluxes in the product and the internal air flow are through the air inlet and absorbed heat as it flowed upward through
Np 52Dp rc the lower part of the dryer to the upper part. The lower part of the
(13)
Ni 5uc dryer was designed for maximum heating of the passing air—foam-
insulator on the bottom and air-insulator on the top minimized heat
It is assumed that the flux in the product is diffusive, while the flux in
loss from the black solar-collector plate to the ambience; and the
the internal air flow is convective. The product’s effective diffusivity
height of the air-flow channel was kept at a few centimeters to
follows Arrhenius relationship as
make sure that the layer of air was thin and can absorb heat quickly.
 
E0 The upper part of the dryer contained multiple meshed metal
Dp 5D0 exp 2 (14)
RT
shelves that could hold the products to be dried while minimizing
Values of D0 and Ea are obtained from literature. The u term is obtained unnecessary blockage of the airflow. The layout of the shelves
by solving the heat transfer and fluid dynamics models simultaneously. ensured that all of the products are directly exposed to sunlight
Continuity of moisture concentration is assumed at the dryer’s inlet without being shaded. The meshed shelves also allowed incident
and outlet. Humidity of ambient air and internal were assumed to have solar radiation that did not fall onto the products to reach and get
AKAMPHON ET AL. | 5 of 9

absorbed by the secondary black solar-collector plate under the samples were measured at the beginning, 4 hrs into the drying, and the
shelves. Thus, the products could receive heat from multiple sources: end.
the direct sunlight, and the internal airflow that had received heat Drying of apples was carried out in Experiment 3. The apples
from the lower part of the dryer and the secondary solar-collector were cut into thin slices of approximately 5-mm thickness and
plate in the upper part. Like the lower part, the upper part of the spread out on four meshed metal shelves. Each shelf held 30 pieces
dryer was also insulated with foam- and air-insulator on the bottom of samples. One shelf of samples was placed in an open space out-
and top, respectively, to minimize heat loss to the environment. The side the dryer. The other three shelves, which constituted three sam-
internal airflow absorbed moisture from the materials to be dried ples, were arranged in a 1-column-by-3-rows configuration in the
and left the dryer through the outlet at the top. drying space.
The calculation of remaining moisture from the measured weights
was done using the following equation

W2Wd
3.2.2 | Experimental setup MR5 ; (15)
W0 2Wd
Experiment 1 involved drying of common snakehead fish (Channa
where MR is the remaining moisture ratio, W is the product weight at
striata), which were filleted and salted for 12 hrs before the drying.
any time, Wd is the dry weight, and W0 is the starting weight. In this
There were four fish samples. The weights of the samples were meas-
work, the dry weight of a product is its weight after 2 days in the solar
ured at the beginning of the experiments and hourly thereafter. The
dryer.
samples were arranged inside the drying space in a 2-columns-by-2-
rows configuration. The samples were dried for 6 hrs. The outside
3.2.3 | Inputs for IHTFDMT model
weather conditions (air temperature, relative humidity, and solar radia-
tion) were recorded throughout the experiment. The model is implemented using conjugated heat transfer, laminar fluid
In Experiment 2, bananas, peeled and halved lengthwise, were flow, and diffusion of single species models in COMSOL Multiphysics
used as the products to be dried. There were four samples in total. The 4.4. The physical domains included in the model were heat transfer
samples were arranged inside the drying space in a 2-columns-by-2- with surface-to-surface radiation, laminar flow, and transport of diluted
rows configuration. The drying lasted for 7 hrs. The weights of the species.
The values of absorptivity and emissivity of the solar collector
plate are 0.9 and 0.8, respectively. These values are used in Equation 2.
The values of absorptivity a, emissivity E, activation energy Ea , and
effective diffusivity D0 of fish, bananas, and apple were obtained from
the literature as followed:
The absorptivities and emissivities are used in Equation 5,
while the activation energies and effective diffusivities are used in
Equation 14.
The initial moisture concentrations c0 of fish, banana, and apple
(43,000, 55,000, and 30,000 mol/m3) used in the model were experi-
mentally determined from initial and dry weights of the samples.
The humidity of ambient and internal air was assumed to have
negligible effect on the drying process.
The recorded hourly solar radiation and ambient temperature were
used in the heat transfer model.
The IHTFDMT model was used to calculate the moisture concen-
tration of the products during the drying process. The moisture con-
centration can be translated into remaining moisture, which is the
parameter for comparison across the three results (experiment, model,
and simplified) by
c
MR5
c0

where c is the product moisture concentration during the drying


process.

3.2.4 | Simplified equation of thin-layer drying

FIGURE 2 (a) A photo and (b) schematic of the passive, mixed- Simplified equation of thin-layer drying is a mathematical model that has
mode, thin-tunnel solar dryer used for model validation been commonly used to describe drying processes, in the following form
6 of 9 | AKAMPHON ET AL.

Material a E Ea D0

Fish 0.5 (Kituu 0.8 (Kituu 28,359 (Park, 1998) 6.24 3 1026 (Park, 1998)
et al., 2010) et al., 2010)

Banana 0.26 (Njie 0.6 (Njie 9,124 (Demirel 1.05 3 1028 (Demirel & Turhan, 2003)
& Rumsey, 1998) & Rumsey, & Turhan, 2003)
1998)

Apple 0.25 (Arinze, 0.6 (Arinze, 24,512 (Wang 9.68 3 1026 (Wang et al., 2007)
1978) 1978) et al., 2007)

M2Me relatively fast even when some of the moisture has already been evapo-
MR5 5exp ð2ktÞ (16)
M0 2Me rated, as evident in the first 4 hrs of the experiment. As shown in Figure
where MR is the remaining moisture ratio, M is the moisture content at 3, the simulated result from the IHTFDMT model was very close to the
time t, M0 is the starting moisture content, Me is the equilibrium moisture regression result of the simplified equation. The error of regression of
content, and k is the drying constant. The drying constant k is calculated the model was 3.17%, while that of the simplified equation was 2.30%.
by means of curve fitting: finding the slope of a semi-log plot between
MR and t obtained from an actual drying process. The drying constant k is
4.2 | Bananas
used for comparing performances of different solar dryers or different
products in the same conditions (El-Beltagy, Gamea, & Essa, 2007; Sacilik The incident solar radiation of Experiment 2 ranged from 450 to 702 W/
& Elicin, 2006). In this work, the simplified equation for thin-layer drying is m2. The weights of the banana samples at the beginning of Experiment 2
used to provide a baseline goodness of fit for comparison with that of the were 58, 62, 64, 64, 66, and 70 g. Figure 4 shows the calculated hourly
IHTFDMT model. The value k for each product is determined using least- remaining moisture percentages. The simulated data shows good fit with
square regression of the experimental remaining moisture ratio results. the experimental ones. Note that the experimental results have slower
drying rate than the simulated results after the first 4 hrs. The late, slower
4 | RESULT COMPARISON AND drying stage agrees with the findings in (Koua, Fassinou, Gbaha, & Toure,
DISCUSSION 2009), which attributed the effect to the product being in a hygroscopic
domain where water exists only in tied form by sorption, which is more
In this section, results obtained from the IHTFDMT model simulation difficult to be released, rather than in vapor form, which accounts for the
are validated against those obtained experimentally. On each figure, faster drying earlier. The error of regression of the model was 6.06%,
the goodness of fit is indicated using root mean square percent devia- slightly higher than that of the simplified equation (4.20%).
tion (e), whose formula is
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uX  
u Xi 2Yi 2 4.3 | Apples
u
t Xi
e5 ; (17) The incident solar radiation of Experiment 3 ranged from 502 to 823
N
W/m2. The combined weight of the apple slices on the shelf that was
where Xi, Yi, and N are the simulated result values, experimental result
placed outside the dryer was 520 g at the beginning. The weights of
values, and number of experimental results, respectively. The standard
the samples on the three shelves inside the drying space were 417,
errors of regression of the IHTFDMT model and the simplified equation
418, and 422 g.
are denoted by em and en, respectively.

4.1 | Common snakehead fish (Channa striata)


The incident solar radiation of Experiment 1 ranged from 420 to 752
W/m2. The weights of the samples at the beginning of the experiment
were 193, 226, 252, 284, and 292 g. The average remaining moisture
percentages, calculated from the hourly measured weights, are plotted
in Figure 3.
The experimental data show that the fish samples dried at a slightly
faster rate in the first 4 hrs than in the simulation, after which the model
becomes slightly faster. This is because the simulated results do not take
into account shrinking during the drying. Shrinking of the product keeps FIGURE 3 Remaining moistures in the fish samples from the
the moisture more concentrated, allowing for the drying to remain experiment, IHTFDMT simulation, and simplified equation
AKAMPHON ET AL. | 7 of 9

FIGURE 4 Remaining moistures in the banana samples from the


experiment, IHTFDMT simulation, and simplified equation

Figure 5 shows that the remaining moisture percentages in the


experimental and simulation results match reasonably well, despite
the fast-drying rates in the first 4 hrs and much slower drying rates in
the last 3 hrs. In this case, the model outperformed the simplified
model with regard to the goodness of fit (2.53% vs. 4.93% errors of
regression).

4.4 | Comparison to simplified equation of drying


The simplified equation relies on linear regression to derive drying con-
stants by means of least square error, which makes it accurate by
nature; yet, the proposed physics-based IHTFDMT model was still able FIGURE 6 Remaining moistures in the fish samples at different
to give solutions that were comparably, and sometimes even more rack positions (a) from the experiment and (b) IHTFDMT simulation
accurate, compared to those from the simplified equation. This proves
From Figure 6, the IHTFDMT model was able to predict the rela-
the predictive capabilities of the model, which can be used to facilitate
tive drying performances with respect to the different rack position in
future solar dryer design, estimate drying time of product, and model
the dryer. That is, the products on the top rack still had more moisture
thermal performances of available solar dryers.
left at the end than those on the middle and bottom racks.
Additionally, the model can be used to provide further insight into
4.5 | Design optimization with IHTFDMT model
the causes of the slower drying in the top rack. In this case, three possi-
The benefit of integrating a fluid dynamics model into the IHTFDMT ble causes from the internal air flow—velocity, temperature, and humid-
model is the insight into how the complex airflow inside the dryer can ity—were investigated using simulated results.
affect the thermal and drying performances of the product. For dryers Figure 7a shows that the air passing through the top rack had
that do not have lengthwise uniform airflow, air temperature and veloc- the highest temperature, since it had the longest time to accumulate
ity can vary greatly at different locations inside the dryer. For example, heat from the solar collector. However, according to Figure 6, the
the model can simulate the remaining moistures of the products placed fish slice in the top rack was the slowest to dry. Figure 7b,c illustrate
at different rack positions inside the dryer. that the air humidity around the top rack was the highest and the air
velocity there was not the fastest, due to the narrow air outlet at the
exit. Those 2 factors could account for the slower drying rate of the
top rack. Figure 7 illustrates that air temperature is not the only driv-
ing factor in drying rate. In fact, it may have less of an effect com-
pared to velocity and humidity. This suggests that a solar dryer
design that aims only to maximize internal airflow temperature may
not lead to optimal drying, and that the airflow temperature should
not be used as the sole benchmark for solar dryer design.

5 | CONCLUSIONS

FIGURE 5 Remaining moistures in the apple samples from the This article proposed a mathematical model for determining the per-
experiment, IHTFDMT simulation, and simplified equation formance of solar dryers, by integrating heat transfer, computational
8 of 9 | AKAMPHON ET AL.

tied form by sorption, which is more difficult to be released, rather


than in vapor form. Incorporating these two variations into the
IHTFDMT model should improve the accuracy of the model even
more. The IHTFDMT model also offers insights into how the complex
airflow inside the dryer can affect the thermal and drying performances
of the product. In some cases, the velocity and humidity of the internal
airflow could have a stronger influence than that of the temperature
on the drying of the product. Based on these findings, mixed mode
solar dryers should be designed with relatively narrower air path to
allow for higher internal airflow velocity, which leads to better ventila-
tion, reduced air humidity, and consequently better drying
performance.

ORC ID
Sittha Sukkasi http://orcid.org/0000-0002-6499-8901

RE FE RE NC ES
Akpinar, E. K., Bicer, Y., & Yildiz, C. (2003). Thin layer drying of red pep-
per. Journal of Food Engineering, 59(1), 99–104.
Arinze, E. A. (1978). Solar energy absorption properties of some agricul-
tural products. Masters thesis by Arinze, Edwin A. at the University
of Saskatchewan in the year 1978.
Basunia, M. A., & Abe, T. (2001). Thin-layer solar drying characteristics
of rough rice under natural convection. Journal of Food Engineering,
47(4), 295–301.
Condorı, M., & Saravia, L. (2003). Analytical model for the performance
of the tunnel-type greenhouse drier. Renewable Energy, 28(3),
467–485.
Defraeye, T., Blocken, B., & Carmeliet, J. (2012). Analysis of convective
heat and mass transfer coefficients for convective drying of a porous
flat plate by conjugate modelling. International Journal of Heat and
Mass Transfer, 55(1), 112–124.
Demirel, D., & Turhan, M. (2003). Air-drying behavior of Dwarf Cavendish
and Gros Michel banana slices. Journal of Food Engineering, 59(1), 1–11.
FIGURE 7 Internal air (a) temperatures (b) velocities and (c)
humidity at the positions 2 cm in front of the top, middle, and Dubey, O., & Pryor, T. (1996). A user oriented simulation model for deep
bottom racks bed solar drying of rough rice. Renewable Energy, 9(1–4), 695–699.
El-Beltagy, A., Gamea, G., & Essa, A. A. (2007). Solar drying characteris-
tics of strawberry. Journal of Food Engineering, 78(2), 456–464.
Ertekin, C., & Yaldiz, O. (2004). Drying of eggplant and selection of a
fluid dynamics, and mass transfer domains. The integrated IHTFDMT
suitable thin layer drying model. Journal of Food Engineering, 63(3),
model could determine the remaining moisture content of the drying
349–359.
products, taking into account the effects of the solar radiation, sur-
Fudholi, A., Sopian, K., Ruslan, M. H., Alghoul, M. A., & Sulaiman, M. Y.
rounding conditions, and dryer’s design and configuration, while also (2010). Review of solar dryers for agricultural and marine products.
able to determine the values of intermediate parameters, such as the Renewable and Sustainable Energy Reviews, 14(1), 1–30.
internal airflow velocity and temperatures of the products, dryer surfa- Hachemi, A., Abed, B., & Asnoun, A. (1998). Theoretical and experimental
ces, and internal airflow. The experiments with a mixed-mode solar study of solar dryer. Renewable Energy, 13(4), 439–451.
dryer using bananas, fish, and apples were conducted to validate the Jain, D. (2005). Modeling the system performance of multi-tray crop dry-
model. The experimental results agree well with those simulated by the ing using an inclined multi-pass solar air heater with in-built thermal
storage. Journal of Food Engineering, 71(1), 44–54.
model. Furthermore, when compared with the linear-regression-based
Jain, D., & Jain, R. K. (2004). Performance evaluation of an inclined
simplified equation of drying, the IHTFDMT model demonstrates a
multi-pass solar air heater with in-built thermal storage on deep-bed
comparable level of accuracy. The small deviations between the experi- drying application. Journal of Food Engineering, 65(4), 497–509.
mental and simulated results could be contributed to the fact that the
Janjai, S., Srisittipokakun, N., & Bala, B. K. (2008). Experimental and mod-
model does not account for (a) product shrinkage during the drying and elling performances of a roof-integrated solar drying system for dry-
(b) the product being in hygroscopic domain where water exists only in ing herbs and spices. Energy, 33(1), 91–103.
AKAMPHON ET AL. | 9 of 9

Kituu, G., Shitanda, D., Kanali, C., Mailutha, J., Njoroge, C., Wainaina, J., Njie, D. N., & Rumsey, T. R. (1998). Solar absorptivity of peeled cassava
& Silayo, V. (2010). Thin layer drying model for simulating the drying root, yam tuber and unripe plantain fruit. Journal of Food Process
of Tilapia fish (Oreochromis niloticus) in a solar tunnel dryer. Journal of Engineering, 21(4), 317–326.
Food Engineering, 98(3), 325–331. Park, K. (1998). Diffusional model with and without shrinkage during
Koua, K. B., Fassinou, W. F., Gbaha, P., & Toure, S. (2009). Mathematical salted fish muscle drying. Drying Technology, 16(3–5), 889–905.
modelling of the thin layer solar drying of banana, mango and cas- Ratti, C., & Mujumdar, A. (1997). Solar drying of foods: Modeling and
sava. Energy, 34(10), 1594–1602. numerical simulation. Solar Energy, 60(3–4), 151–157.
Lamnatou, C., Papanicolaou, E., Belessiotis, V., & Kyriakis, N. (2010). Sacilik, K., & Elicin, A. K. (2006). The thin layer drying characteristics of
Finite-volume modelling of heat and mass transfer during convective organic apple slices. Journal of Food Engineering, 73(3), 281–289.
drying of porous bodies – Non-conjugate and conjugate formulations
grul, _IT., & Pehlivan, D. (2002). Mathematical modelling of solar drying
To
involving the aerodynamic effects. Renewable Energy, 35(7), 1391–
of apricots in thin layers. Journal of Food Engineering, 55(3), 209–216.
1402.
Wang, Z., Sun, J., Liao, X., Chen, F., Zhao, G., Wu, J., & Hu, X. (2007).
Leon, M. A., Kumar, S., & Bhattacharya, S. (2002). A comprehensive pro-
Mathematical modeling on hot air drying of thin layer apple pomace.
cedure for performance evaluation of solar food dryers. Renewable
Food Research International, 40(1), 39–46.
and Sustainable Energy Reviews, 6(4), 367–393.
Yaldiz, O., Ertekin, C., & Uzun, H. I. (2001). Mathematical modeling of
Margaris, D. P., & Ghiaus, A.-G. (2006). Dried product quality improve-
thin layer solar drying of sultana grapes. Energy, 26(5), 457–465.
ment by air flow manipulation in tray dryers. Journal of Food Engineer-
ing, 75(4), 542–550. z, O., & Erteky
Yaldy n, C. (2001). Thin layer solar drying of some vegeta-
bles. Drying Technology, 19(3–4), 583–597.
Mathioulakis, E., Karathanos, V. T., & Belessiotis, V. G. (1998). Simulation
of air movement in a dryer by computational fluid dynamics: Applica-
tion for the drying of fruits. Journal of Food Engineering, 36(2),
183–200. How to cite this article: Akamphon S, Sukkasi S, Sedchaicharn K.
Midilli, A., & Kucuk, H. (2003). Mathematical modeling of thin layer dry- An integrated heat-transfer-fluid-dynamics-mass-transfer model
ing of pistachio by using solar energy. Energy Conversion and Manage- for evaluating solar-dryer designs. J Food Process Preserv. 2018;
ment, 44(7), 1111–1122.
e13649. https://doi.org/10.1111/jfpp.13649
Mirade, P., & Daudin, J. (2000). A numerical study of the airflow patterns
in a sausage dryer. Drying Technology, 18(1–2), 81–97.

You might also like