Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Journal of Food Engineering 312 (2022) 110739

Contents lists available at ScienceDirect

Journal of Food Engineering


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

Numerical and experimental study of airflow resistance across an array of


sliced food items during drying
Petros Demissie Tegenaw a, *, Pieter Verboven b, Maarten Vanierschot a
a
Department of Mechanical Engineering, Celestijnenlaan 300, Leuven, B-3001, Belgium
b
Division of Mechatronics, Biostatistics and Sensors (MeBioS), Willem de Croylaan 42, Leuven, B-3001, Belgium

A R T I C L E I N F O A B S T R A C T

Keywords: Characterization of the airflow across food items involves identification of their viscous and inertial resistances.
Airflow resistance This paper presents an extensive numerical and experimental study to determine the resistances of food items
Sweet potato placed in an array. The experimental study used an array of sliced sweet potato with different initial slice di­
Porous media formulation
ameters. The airflow resistance of the array was determined from a series of wind tunnel measurements. The
CFD
Dehydration
sliced arrays were dehydrated and digital images were taken to determine the intermediate array-porosity. The
numerical study was based on CFD modeling of the airflow across the disc arrays. In the simulations, the size of
each individual disc in the array was gradually decreased to reproduce the array-porosity change similar to the
experimental dehydration. The airflow resistance of the array was found to be dominated by inertial effects. The
influence of geometrical parameters on the inertial resistance coefficient was also studied. Dimensional analysis
showed that the inertial resistance was exponentially decaying with array-porosity. To describe this relationship,
a model equation was proposed that fitted very well with both the experimental and numerical results. Finally,
the findings of the current study were used in a porous media based modeling and simulation of the momentum
transport in the dehydration of an array of sliced food items in a solar dryer.

1. Introduction 2011). The cross-flow configuration is observed in most solar food


drying units developed recently (Getahun et al., 2021). Thus, airflow
Food items are dried to prevent perishability and to avail them for resistance determination across food items placed in an array is vital for
the processing industry in a better condition. Convective air-drying is engineering and optimization of existing or newly designed food pro­
the most frequently used dehydration operation in the food industry cessing units. The airflow resistance and factors affecting it are reported
(Veleşcu et al., 2013; Doymaz, 2004). In convective sliced food dehy­ for flows across various fruits and vegetables, such as root and bulb
dration, the drying airflow pattern near the slices can be identified as (Neale and Messer, 1976), potatoes (Irvine et al., 1993), apples and
co-flow and cross-flow, as shown in Fig. 1, or a combination of the two. chicory (Verboven et al., 2004), pistachio nuts (Kashaninejad and Tabil,
In cross-flow configuration, the drying airflow experiences a higher 2009), and green figs (Amanlou and Zomorodian, 2011). It is also
resistance from the sliced food items compared to the one in a co-flow delineated for different food items for example onion (Rapusas et al.,
configuration. However, the cross-flow configuration is found to have 1995) and garlic (Madamba et al., 1994). These food items are raw bulks
a more favourable impact on the airflow distribution and heating uni­ with spherical or nearly spherical shapes, or completely random in the
formity (Udomkun et al., 2015). Besides dehydration, the cross-flow case of onion and garlic. Nevertheless, no such data have been portrayed
configuration is encountered in various other food processing applica­ in the literature or compiled in the ASABE standard D272.3 (Resistance
tions such as refrigeration, storage, and transport. The engineering to Airflow of, 2016) for a sliced array of circular food items as shown in
design, operation, optimization, and control of these processes are Fig. 1b, although this flow configuration is commonly encountered in,
dependent on the magnitude and distribution of the airflow resistance. amongst others, solar dehydration (Belessiotis and Delyannis, 2011;
Moreover, in food dehydration applications, the final quality of the dried Tegenaw et al., 2019a, 2019b).
food product is also affected by the magnitude and distribution of the The airflow resistance correlations are often presented in terms of a
airflow resistance (Verboven et al., 2004; Amanlou and Zomorodian, total pressure drop, ΔP as given by the Darcy-Forchheimer, Equation (1)

* Corresponding author.
E-mail address: petros.tegenaw@kuleuven.be (P.D. Tegenaw).

https://doi.org/10.1016/j.jfoodeng.2021.110739
Received 15 March 2021; Received in revised form 4 July 2021; Accepted 5 July 2021
Available online 8 July 2021
0260-8774/© 2021 Elsevier Ltd. All rights reserved.
P.D. Tegenaw et al. Journal of Food Engineering 312 (2022) 110739

Fig. 1. Common airflow pattern configurations in sliced food dehydration.

(Verboven et al., 2004, 2006; Amanlou and Zomorodian, 2011; Ye et al., combined infrared and hot air drying was investigated using an optical
2016; Guo et al., 2013; Malavasi et al., 2012), imaging method (Onwude et al., 2018, 2019).
η Once the relationship between airflow resistance and array-porosity
ΔP = ξ1 ρa V 2 + ξ2 V, (1) changes is determined, it can be used in porous media based Compu­
d
tational fluid dynamics (CFD) modeling and simulations of macro-scale
where ρa = 1.225 kg/m3 is the air density, η = 1.802 × 10− 5 Pa s the air transport phenomena. In the porous media approach, the food bulk is
dynamic viscosity at a temperature and pressure of 20 ◦ C and 1 atm, regarded as a porous zone through which simultaneous heat and mass
respectively, (Cengel, 2014), d (m) the disc diameter, and V (ms) is the transfer takes place (Darabi et al., 2015; Amanlou and Zomorodian,
airflow velocity. The total resistance is treated as the sum of inertial and 2010; Delele, 2009; Getahun et al., 2017a, 2017b; Delele et al., 2012).
viscous resistance terms, with their non dimensional coefficients, ξ1 and This modeling approach uses a space averaging technique to model the
ξ2, respectively. These coefficients were found to be dependent on phenomena with reasonable computational effort (Verboven et al.,
physical parameters such as the diameter, bulk porosity, stacking 2006). It was successfully used in the CFD simulation of a tray drier of
pattern, fluid property, product shape, roughness, and confinement ratio kenaf core (Misha et al., 2013), corn (Sanghi et al., 2018), and deep-bed
(Verboven et al., 2004, 2006). The exact effect of those physical pa­ paddy (Ranjbaran et al., 2014). Moreover, it is incorporated in most
rameters on the coefficients is given by Ergun’s equation for a bulk of commercial codes and the inputs needed are the airflow resistance co­
spherical or nearly spherical food items (Ergun and Orning, 1949). For efficients and porosity. Thus, determining these coefficients for an array
such shaped food items, Ergun’s equation and its subsequent modifica­ of food slices will play a vital role in the porous media based CFD
tions have shown that the coefficients are a function of the bulk porosity modeling of such food items. This involves obtaining the resistance co­
(Ergun and Orning, 1949; Delele, 2009; van der Sman, 2002; Verboven efficients as presented in Equation (1).
et al., 2004). However, for sliced food items, to the authors’ knowledge, The main objectives of this research are, (1) to determine the airflow
there are no such relations expressing the coefficients in Equation (1), in resistance for flows across an array of sliced food items numerically and
terms of geometrical parameters like bulk porosity. experimentally, (2) to study the effect of shrinkage, due to dehydration,
The relative dominance of the two terms (viscous and inertial) in on the airflow resistance, (3) to develop a relationship between the
Equation (1) determines how momentum transport is modeled. In resistance coefficients and geometrical parameters like initial disc
literature, a porous medium Reynolds number (Rep), a ratio of inertial diameter and array-porosity, and (4) to show the application of the
term to viscous term, was reported to be a quantitative relation to judge model in a macro-scale CFD simulation of a solar dryer taken from
on the relative importance of the two terms (Verboven et al., 2006; literature.
Gruyters et al., 2018). At high flow speeds, Rep ≫ 1, the pressure drop is
dominated by inertial effects (first term of Equation (1)) while at very 2. Material and methods
low flow speeds Rep ≪ 1 as in the case of Stokes or creeping flow re­
gimes, the viscous effects become dominant (second term of Equation To determine the resistance coefficients as shown in Equation (1) for
(1)) (Verboven et al., 2006; van der Sman, 2002). sliced food items, both experimental and numerical studies were per­
During dehydration of sliced food items, the moisture content re­ formed. In the experimental study two initial slice diameters (33.3 and
duces significantly, this results in shrinkage of an individual slice due to 40 mm) were used and in the numerical study three (33.3, 40, and 50
the contraction of the visco-elastic matrix (Yadollahinia and Jahangiri, mm).
2009). Thus, the relative open space between the individual slices,
array-porosity, will change when sliced food items are dried by arran­
ging them in an array. This will in turn affect airflow resistance of the 2.1. Laboratory experiments on sweet potato slices
array of sliced food items. For a stack of green figs the change in bulk
porosity was found to affect airflow resistance notably (Amanlou and 2.1.1. Sample preparation
Zomorodian, 2011). Therefore, to determine the relationship between White fleshed sweet potato was purchased from local farmers around
the changes in the resistance with array-porosity, the associated Bahir Dar, Ethiopia. The samples were washed with pure water to
shrinkage needs to be identified. Shrinkage in fruits and vegetables has remove the soil, and the surface was dried afterwards. The samples were
been studied by using prediction models that can be categorized as stored in an incubator at a temperature set point of 13 ◦ C to prevent
empirical, theoretical, and physics-based models (Mahiuddin et al., germination (Devereau, 1994). The average initial moisture content of
2018). Recently, direct experimental measurements involving computer the sweet potato, obtained by drying 24 random samples of slices at a
vision have been employed to determine shrinkage during dehydration temperature of 105 ◦ C for 48 h, was 2.53 g of water per gram of dry
(Yadollahinia and Jahangiri, 2009; Singh and Talukdar, 2019; Sun et al., matter. Slices with 4 mm thickness were cut using a professional slice
2019). Specifically, the shrinkage of sliced sweet potatoes during cutter (EH 158-L, Graef, Germany) from the sweet potato. Then, two
experimental sets with different initial slice diameter were prepared:

2
P.D. Tegenaw et al. Journal of Food Engineering 312 (2022) 110739

Fig. 2. A 2D representation of the experimental array of sliced sweet potatoes with different initial slice diameters. a) Set-I: diameter 33 mm (6 × 6) and b) Set-II:
diameter 40 mm (5 × 5).

Fig. 3. Schematic representation of the experimental set up for air flow resis­
tance measurements (Side-view): ①- Centrifugal fan, ②- Airflow speed mea­
surement, ③- Wind tunnel, ④- Array of sliced sweet potato, ⑤- Measurement
channel, and ⑥data logger. A and B are pressure taps. All dimensions are
in meter.

set-I with 36 (6 × 6) discs of 33 mm diameter, and set-II with 25 (5 × 5)


discs of 40 mm diameter. The sliced discs were packed in between a wire
mesh (20 mm × 20 mm size) as shown in Fig. 2. This made sure that the
slices stay in place throughout the drying and subsequent airflow
resistance measurements. The airflow resistance of the open wire mesh
was measured and subtracted from the pressure drop measurements.
However, it was found that the resistance is much lower compared to the
one of the full pack as shown in Fig. 2. To be more specific, the pressure
drop measured at the highest airflow speed (around 2 m/s) was found to
be only 1.1 Pa.

2.1.2. Experimental setup


Fig. 3 shows a schematic representation of the test set up. The airflow
was created by a centrifugal fan ①driven by a frequency controller. In Fig. 4. A step by step representation of the procedure followed to determine
order to remove fluctuations and to create a uniform flow velocity pressure drops at each dehydration level.
profile with a thin boundary layer, a wind tunnel ③was employed. The
same wind tunnel was reported to have a very low turbulence intensity
B, see Fig. 3) were used to measure the change in pressure, ΔP between
of 0.088% at a mean velocity of 1.73 m/s with a uniformity of 99.8%
the upwind and downstream side. Pressure tapping point B was carefully
(Vanierschot et al., 2013). A 1.5 m long test channel with a squared cross
chosen at 0.25 m from the outlet of the test section, avoiding the wake of
section (0.2 m × 0.2 m) ⑤was used. The sliced sweet potatoes, as packed
the airflow in the downwind side. At each pressure tapping location, the
in Fig. 2, were placed at 0.25 m from the inlet of the test channel
pressure was tapped from four sides of the square cross section. The four
covering the entire cross section ④. Two pressure tapping points (A and
taps were interconnected ensuring the measurement of the average cross

3
P.D. Tegenaw et al. Journal of Food Engineering 312 (2022) 110739

Table 1
The diameters of the individual discs (di) constituting the array as used in the
CFD simulation sets (I, II, and III) given in millimeters. The respective array-
porosity (ϵ) in percentage is determined using Equation (2).
CFD Set array-porosity, %

I (6 by 6) II (5 by 5) III (4 by 4)

1 33.3 40 50 21.5
2 30.1 36.1 45.1 36.0
3 27.2 32.6 40.7 47.9
4 24.5 29.4 36.8 57.6
5 22.1 26.5 33.2 65.4

array-porosity, ϵ was then calculated as in Equation (2), where n was 36


and 25 for experiment set-I and set-II, respectively. The initial
array-porosity is around ϵ0 = 21%. This value is constant for any initial
slice diameter forming a squared array, as also reported by Beasley et al.
(Beasley and Clark, 1984).
∑n 2
At − i=1 π r i
ϵ= (2)
At
Fig. 5. A representative 3D computational grid used in the CFD modeling (ø =
2.1.5. Data analysis
50 mm) with closeup views showing the grid refinements made at
the interfaces. Airflow resistance data for sliced sweet potato at different stages of
drying were fitted to Equation (1) using a linear regression model. The
resulting values of the constants ξ1 and ξ2 at those stages were deter­
sectional value. The airflow velocity at the inlet of the channel was
mined and plotted against the corresponding array-porosity from
measured by a direction independent flow velocity probe ②(hot wire
Equation (2). Several statistical criteria such as the coefficient of
anemometer, Testo, Ternat, Belgium) with a velocity range from 0 to 10
determination r2, root mean square error (RMSE) and mean relative
m/s and an accuracy of ±(0.03 m/s + 5% measured value). The pressure
percentage deviation modulus (P%) were used to evaluate the goodness
drop was measured by a pressure transducer ⑥(Testo480, Testo, Ternat,
of fit.
Belgium) having a range from − 100 to 100 hPa with an accuracy ±
(0.3 Pa + 1% measured value).
2.2. CFD model formulation
2.1.3. Measurement procedure
The pressure drop measurements were taken for an airflow velocity 2.2.1. Computational domain and boundary conditions
range from 0 to 2 m/s. This range ensures coverage of airflow velocities A 3D symmetric computational domain with discs of 50 mm is shown
used in most convective food drying (Velić et al., 2004), and cooling in Fig. 5. It was constructed in one to one scale with a quarter of the
(Delele, 2009; Defraeye et al., 2013) applications. The airflow resistance measurement channel. The symmetry boundary condition was moti­
of the array of sliced sweet potato was characterised through a six step vated owing to the square cross section of the measurement channel and
procedure. First, the array was placed in the measurement channel. the discs arrangement inside. Moreover, the wind tunnel used in this
Second, the airflow speed was set to a small value around 0.25 m/s. study has shown 99.8% uniformity ensuring a mirrored flow distribution
Third, fifteen samples were recorded at a frequency of 1 Hz. Fourth, the along the symmetry planes. The airflow field was simulated at four
airflow speed was increased step by step up to 2 m/s, and at each step different inlet velocities (0.5, 1, 1.5 and 2 m/s), prescribed normal to the
pressure measurements were recorded. Fifth, the sample was taken out inlet face. A pressure outlet with zero static pressure was imposed at the
of the measurement channel and the digital images were taken as outlet. The measurement channel walls and the disc surfaces were
described in section 2.1.4. Sixth, the sample array of sliced sweet potato treated as no-slip walls.
was placed in an electric oven at 85 ◦ C and dried for about 10–20 min. In this part of the study, three initial disc diameters were investigated
Then, the entire procedure was repeated after each oven drying until the namely CFD set-I, set-II and set-III with initial disc diameters (d0) of
pressure drop measured at the highest airflow speed was below 10 Pa. 33.3, 40 and 50 mm, respectively. These initial disc diameters were then
Fig. 4 depicts a step by step procedure followed in the pressure drop successively reduced to mimic shrinkage due to dehydration. The indi­
experiments. The air temperature during the pressure drop measure­ vidual disc diameter (di) was reduced geometrically as
ments varies from run to run in a range between 18 ◦ C and 21 ◦ C.
di = d0 Ri+1 , (3)
2.1.4. Determination of the array-porosity
where the common reduction ratio (R), the amount by which the
The flow resistance measurements were performed at various
dimension of the discs reduces, was chosen to be 95%. The real sizes of
dehydration points of the packed array of sliced sweet potato. After each
the individual discs used in the simulations are reported in Table 1. The
airflow resistance measurement, the two experimental sets (set-I and set-
corresponding array-porosity ε, representing the relative open space,
II) were dried as described in section 2.1.3. As a consequence of this
was thus changed from 21.5% to 65.4% according to Equation (2). The
dehydration there was mechanical deformation in the form of shrinkage,
range of array-porosity presented here covers the entire shrinkage range
which increases the array-porosity of the array of slices. To quantify the
for potato slices reported by Alireza and Mehdi (Yadollahinia and
array-porosity, a digital image of the sample was taken by a 16 MP BSI
Jahangiri, 2009).
CMOS camera at a given dehydration stage. The image pre-processing
and segmentation was performed according to procedures outlined in
2.2.2. Computational grid
(Mery and Pedreschi, 2005; Pedreschi et al., 2006). Then, from the
A quad-dominant unstructured mesh was generated on the disc
pre-processed image, the total size of the image At, and radii of each
surfaces and on the interstitial spaces between them (column 1 in Fig. 6),
slice, ri was extracted in pixels using a MATLAB routine. The
while in columns 2 to 5 of the same figure, structured meshes were

4
P.D. Tegenaw et al. Journal of Food Engineering 312 (2022) 110739

Fig. 6. Surface grids used for the 3 CFD modeled sets with the successive reduction in the diameter of the discs in the array.

applied on the disc surfaces, keeping the interstitial space unstructured.


Table 2
These meshes were extruded further up to the inlet and outlet of the
The predicted area weighted average pressure (Pa) from the four turbulence
computational domain. The entire disc surfaces interacting with the
models and the respective percent deviation (shown in brackets) as compared to
fluid flow were refined ensuring a maximum computed y+ at the highest
the validation measurements at four inlet velocities (for a disc diameter of 40
flow rates to be 4.1. Thus, implementation of a low Reynolds number mm).
formulation (low-Re) was guaranteed for better resolving the boundary
m Experiment Simulated turbulence models
layer close to the interaction regions (Defraeye and Radu, 2017; Cac­ V0 ( )
s
cavale et al., 2016). RNG k-ε Realizable k- SST k-ω Standard k-
To avoid highly skewed mesh elements near the tangent points, small ε ε

fillets with a radius of 0.01 mm were introduced as shown in Fig. 5. 0.5 4.0 4.8 (18.2) 4.9 (20.8) 5.7 (42.4) 5.15 (28.1)
These simplifications in the disc geometry were applied only for simu­ 1 16.7 19.4 20.6 (23.3) 23.5 19.4 (16.4)
(16.4) (40.5)
lations involving the initial disc diameters (row 1 in Table 1 and column 1.5 40.3 45.2 47.6 (18.2) 53.2 (32) 44.6 (10.7)
1 in Fig. 6) as only in those cases the discs touched each other. A similar (12.2)
approach was used to perform a highly resolved CFD simulations of fluid 2 71.9 83 (15.5) 88.2 (22.7) 95 (32.1) 78.7 (9.5)
flow in the interstitial spaces of densely packed spheres representing
sand grains (Knight, 2018).
2.2.4. Numerical simulation details
The finite volume based numerical solutions of Equations (4) and (5)
2.2.3. Governing equations
were calculated using ANSYS Fluent 2020 R1. The equations were dis­
The governing equations for predicting the flow field within the
cretized using a second order upwind scheme. The turbulent nature of
computational domain were derived from mass and momentum con­
the simulated airflow regimes requires usage of appropriate turbulence
servation principles. The airflow was modeled using a steady 3D
models. In this study, pressure drop predictions of four commonly used
Reynolds-averaged Navier Stokes (RANS) approach for incompressible
turbulence models (standard k − ε, RNG k − ε, realizable k − ε, and k − ω
fluid, which reads as:
SST) were compared against a validation experiment. In all these tur­
Mass conservation:
bulence models, a low-Re formulation was implemented rather than

∇⋅V = 0 (4) wall functions to fully resolve the boundary layer. As outlined in Table 2,
the standard k − ε model shows the smallest deviation with the experi­
and momentum conservation: mental measurements. Therefore, it was used in all the simulations
→→ (μ ( → →T ) ) ∇Pm carried out. Nevertheless, significant differences can be observed,
(5)
eff
∇⋅(V V ) − ∇⋅ ∇V + ∇V
ρ
=− ,
ρ especially for the low flow rates. These differences are attributed to the
isotropic turbulent viscosity assumption of the turbulence models. As
( ) the pressure drop across the discs is mainly inertial related, the structure
kg
where the effective viscosity, μeff m.s is the sum of the dynamic and of the wake flows behind the discs plays a significant role. These wakes
turbulent viscosity, Pm = P + 2/3ρk (Pa) is the modified pressure, and V are highly anisotropic and hence the isotropic viscosity assumption in­
(ms) the airflow velocity. The density and viscosity were taken as those of troduces discrepancies, especially at low flow rates. To overcome this
air at 20 ◦ C from Cengel (2014). discrepancy and capture the wake in a better accuracy, the more

5
P.D. Tegenaw et al. Journal of Food Engineering 312 (2022) 110739

Table 3 Thus, the medium grid and its grid topology were implemented in all the
Grid independence analysis of the prediction of pressure drop using the GCI from simulations carried out.
m
simulations at V0 = 2 and at V0 = 0.5 , respectively.
s
3. Results and discussion
Number of Area weighted average pressure (Pa) Maximum GCI
elements (%)
V0 = V0 = V0 = V0 = 3.1. Experimental study of airflow resistance across the array of sliced
m m m m
0.5
s
1
s
1.5
s
2
s sweet potato
Coarse 5.11 19.78 43.82 77.20 –
(2054870) As dehydration in an electric oven proceeds, mechanical deforma­
Medium 5.15 20.09 44.63 78.74 4.39 tion in the form of shrinkage of the sweet potato slices was observed.
(3,570,992) This results in an increase of the array-porosity of the array, which in
Fine (8,032,967) 5.16 20.10 44.67 78.80 0.20
turn significantly affects the airflow resistance. This effect was presented
as a pressure drop versus airflow velocity in Fig. 7 for experimental set-I
computationally intensive large eddy simulations can be employed. The and set-II. The experimental data were fitted to the Darcy-Forchheimer
SIMPLEC algorithm was used for pressure-velocity coupling. The nu­ equation given by Equation (1) at each array-porosity level by a non-
merical solutions were considered converged when the unscaled resid­ linear regression analysis. The R2 for each test condition exceeded
ual dropped to a value 10− 6 for all equations. 0.99 as shown in Table 4. Thus, equation (1) was found to represent the
A grid independence analysis was carried out with three different experimental results with sufficient accuracy, over the entire range of
computational grids (coarse, medium and fine) for the flow through an airflow velocities studied.
array of discs with a diameter of 40 mm. The grids were constructed by The airflow resistance across the slices of sweet potato is increases
successive refining in all directions for reporting the discretization error with increasing airflow velocity. Moreover, if the airflow velocity is
in terms of the grid convergence index (GCI) (Roache, 1994). The fixed, it is found to be decreasing as array-porosity increases. This is
pressure drop predictions for a range of inlet velocities from 0.5 to 2 m/s because, at a fixed airflow velocity, the higher the amount of void
were used for the analysis. As summarized in Table 3, the maximum spaces, the lower the viscous and the inertial resistances by the slices,
discretization error was found to be about 4.39% between the course which results in lower flow energy losses.
and medium grid, and about 0.2% between the medium and fine grid. Table 4 summarizes the values of the dimensionless inertial and
viscous coefficients ξ1 and ξ2 respectively for the two experimental sets.

Fig. 7. The effect of array-porosity on the airflow resistance of the slices as determined from experimental measurements for a) experimental set I (ød0 = 33 mm),
and b) experimental set II (ød0 = 40 mm): Note: In the legends, the values indicated in % represent array-porosity (ϵ).

Table 4
Curve fitting summary for experimental sets I and II in terms of dimensionless inertial and viscous resistance coefficients, ξ1 and ξ2 respectively, according to Equation
(1).
Experiment set-I Experiment set-II

ϵ(%) 26.5 34.76 39.49 44.78 22.51 29.29 36.64 38.84 46.49 50.85
ξ1 10.4 4.8 4.0 3.2 14.9 7.85 5.48 4.56 3.18 2.26
ξ2 15.37 429 − 1114 − 959 2359 3654 − 418.3 − 1095 − 1365 − 886
R2 1 1 1 1 0,99 0,99 1 1 1 1
RMSE 0.45 0.28 0.06 0.06 1.24 0.61 0.21 0.13 0.04 0.04

6
P.D. Tegenaw et al. Journal of Food Engineering 312 (2022) 110739

Fig. 9. The effect of array-porosity on the pressure drop under different


simulated airflow velocities. Three sets of numerical studies with initial disc
diameter (d0) are shown: 33 (left), 40 (middle), and 50 mm (right).

3.2. Numerical study of airflow resistance across the array of sliced sweet
potato

3.2.1. Airflow velocity distribution with changing array-porosity


A representative simulation result showing contours of airflow ve­
locity distribution is shown in Fig. 8. The plots show the distribution of
airflow velocity approaching the discs, and the resulting wake at the
back of the disc array. The magnitude of the maximum airflow velocity
was observed in the interstitial spaces and it decreases when the array-
porosity increases from 21.5 to 65.4 percent.

3.2.2. The effect of array-porosity and initial disc diameter on airflow


resistance across an array of discs
The simulated pressure drops for the air flows across the array of
discs (with three initial diameters) are illustrated in Fig. 9, which shows
airflow resistance variations, in terms of pressure drop, at the five array-
porosity levels simulated for those disc arrays. The pressure drops re­
ported here are area weighted average values at a plane 0.15 m before
the disc arrays. For a constant airflow velocity, the pressure drop is
Fig. 8. Contours of airflow velocity magnitude for flows across the array of observed to decrease as the array-porosity increases. This is due to the
discs with varying array-porosity at an inlet velocity V0 = 0.5 m5 . Note: The plots lower drag forces and momentum losses associated with air flows arrays
shown are taken in a YZ plane passing through the center of the discs. of higher array-porosity. The pressure drop at a given airflow velocity
was decreased by around 96% when the array-porosity increases from
The two coefficients are observed to decrease as array-porosity increases 21.5% to 65.4% for the three CFD sets. The results show that the airflow
due to dehydration. This is expected as an increase in array-porosity resistance is highly dependent on the array-porosity and hence accurate
introduces more open space for the air to pass with less flow resis­ models describing this shrinkage are necessary if one wants to optimise
tance. This is also reported in other investigations of airflow resistances the drying process of food slices.
across various food items (Irvine et al., 1993; Amanlou and Zomorodian, Similar to the experimental study, fitting simulated pressure drops
2011; Kashaninejad and Tabil, 2009; Neale and Messer, 1976; Rapusas and associated airflow velocities into equation (1) show the dominance
et al., 1995), and across porous metals (Dukhan and Patel, 2008; Oun of the inertial resistance. Table 5 shows the variation of this resistance
and Kennedy, 2014). The porous media Reynolds number (Rep) was coefficient with the simulated array-porosity values and initial disc di­
defined by dividing the first term by the second term of Equation (1). ameters. As can be observed from this table, ξ1 highly depends on the
Using the coefficients shown in Table 4 at airflow speeds that range from interim array-porosity.
0.01 m/s to 2 m/s, the corresponding Rep was found to lay between 16
and 3 × 103. As Rep in all cases is much larger than one, the airflow 3.3. Inertial resistance coefficient (ξ1) as a function of array-porosity
resistance across the array of sliced sweet potato is entirely dominated
by the inertial resistance term. A closer look at the decrease in the In equation (1), the inertial and viscous resistance coefficients are
viscous resistance coefficients reveals that for a higher array-porosity, represented as ξ1 and ξ2 respectively. Both the numerical and experi­
the value of ξ2 becomes negative. This negative values are merely mental studies on the disc arrays revealed the dominance of the inertial
from numerical fitting of the measurements but physically this means at resistance. Hence, equation (6) can be used to describe the entire pres­
these points viscous resistance becomes less important as more space is sure drop (ΔP) and airflow velocity (V):
open. In this region, only the Forchheimer (inertial resistance) suffices to
describe the relation between the pressure drop and airflow velocity. ΔP = ξ1 ρa V 2 . (6)

The coefficient ξ1 is the highest at the start of the dehydration (low


array-porosity) and vanishes to rather small value at later stages in the
dehydration (higher array-porosity). Tables 4 and 5 have presented an

7
P.D. Tegenaw et al. Journal of Food Engineering 312 (2022) 110739

Table 5
Values of inertial resistance coefficient ξ1 evaluated by fitting the data from CFD study sets I, II, and III into Equation (1). Note: the numbers between square brackets
indicate the range of predicted ξ1 values with 95% confidence.
CFD study set array-porosity(ϵ), %

21.5 36.0 47.9 57.6 65.4

I (ø33mm) 18.5 [18.48–18.6] 5.1 [5.08–5.18] 2.3 [2.29–2.34] 1.2 [1.23–1.25] 0.75 [0.75–0.76]
II(ø40mm) 19.3 [19.26–19.4] 5.27 [5.21–5.33] 2.38 [2.36–2.39] 1.28 [1.27–1.28] 0.77 [0.77–0.78]
III(ø50mm) 20.1 [20.03–20.2] 5.48 [5.44–5.51] 2.42 [2.40–2.44] 1.33 [1.31–1.35] 0.80 [0.79–0.81]

Fig. 10. The relationship between ξi and array-porosity ε from the experi­
mental and numerical studies with the fitted model according to Equation (7).
The three sets of studies with initial disc diameters (d0) are shown: 33 (left), 40 Fig. 12. Comparison of the relationship 7 for discs with results from the
(middle), and 50 mm (right). literature: for perforated plate (Özahi, 2015) and Ergun’s equation (Ergun and
Orning, 1949).

drop relationships for air flows across disc arrays. In addition, due to a
better control over the geometry of constituting discs, it was possible to
delineate the influence of a changing array-porosity on airflow resis­
tance accurately in the case of the numerical study.
Further dimensional analysis was performed to describe the coeffi­
cient ξ1 in terms of geometrical parameters (array-porosity, initial
diameter, and/or diameter to thickness ratio) only. This is only possible
if the dimensionless Euler number (ξ1) is independent of the channel
Reynolds number (Re) (Huang et al., 2013; Özahi, 2015). It can be seen
from Fig. 11 that this validity holds for Re greater than 1 × 104. Similar,
critical Reynolds number value was also communicated for the
discharge coefficients in perforated orifices (Huang et al., 2013). Thus,
the expression shown by Equation (7) is suggested to calculate ξ1, in
terms of geometrical parameters (array-porosity (ϵ) and initial disk
diameter (d0)) for Re greater than 1 × 104.

ξ1 = (A + Bd0 )ϵ− C
(7)

The experimentally and numerically found results did fit very well to the
model Equation (7) with R2 = 0.999. The predicted model parameters A,
Fig. 11. The relationship between Euler number (ξ1) and Reynolds number B and C are 0.279, 1.685 (1/m), and 2.613, respectively.
(Re) at different array-porosity (ϵ) from the data at D0 = 50 mm. Note: The
Fig. 12 shows comparison of Equation (7) with the Ergun’s equation
dimension (L = 0.2 m) in the Re definition is the characteristics length of the
for a pack of spherical objects (Ergun and Orning, 1949), and resistance
square channel.
coefficient for a perforated plate (Özahi, 2015). As it can be observed,
the overall shape of the curve is the same. But, the Ergun’s equation,
exponentially decaying relationship between ξ1 and ϵ for each initial
developed for spherical objects, over and under predicts the resistance
disc diameter (d0). Moreover, Fig. 10 confirms this exponential rela­
coefficient for lower and larger porosity levels, respectively. This is
tionship for the three sets of numerical simulations and it also juxtaposes
because at lower porosity levels, the flow is so much confined and it will
results from the experimental investigation. The comparison between
lose larger amount of energy due to friction on the surfaces of a sphere
both studies shows that the CFD model could be used to find pressure
which is twice as large as that of a disc. Thus, the resistance coefficient as

8
P.D. Tegenaw et al. Journal of Food Engineering 312 (2022) 110739

Fig. 13. Effect of changing array-porosity due to potato slice dehydration on air velocity magnitude profiles near lower(A) and upper(B) rack shelves of the solar food
dryer in the study of Tegenaw et al. (Tegenaw et al., 2019b). Note: The various array-porosity levels are indicated in percentage, and the corresponding pressure drop
relationships are indicated between brackets.

determined by Ergun’s equation is larger. However, at larger porosity 3.5. Case study: CFD modeling of indirect solar food dryer
levels the flow is not confined as such and most of it will chose the path
of least resistance, whereby reducing losses due to friction on the sur­ In this section, a numerical study to predict the airflow distribution
faces. In the case of the perforated plate, lower resistance is expected due inside a solar food dryer based on the porous media approach is pre­
to the smaller solid surface area to block the fluid flow for the same sented to showcase the application of Equation (7). For the details of the
geometric size of the discs. The found differences revealed the limitation CFD modeling and the dimensions of the solar food dryer, the readers are
of the shape factor approach such as sphericity in adapting the Ergun’s directed to previous work (Tegenaw et al., 2019b). In this study, the
equation for calculations involving discs. porous media approach was employed to model slices of sweet potato
placed on the lower and upper rack shelves of an indirect solar dryer,
3.4. Relationship between airflow resistance coefficient with moisture whereby a constant resistance coefficient is prescribed using Ergun’s
content equation with a shape factor adopted for cylinders. The pressure drop
relationship from the equation as seen in Fig. 13 is composed of both the
The relationship described in Equation (2) was converted to Equa­ inertial and viscous term.
tion (8) by assuming that all the discs shrink equally and only in the Fig. 13 shows the magnitude and distribution of the airflow velocity
radial direction, starting from an initial disc radius (r0). in the vicinity of the rack shelves. By applying the findings of the current
√̅̅̅̅̅̅̅̅̅̅̅̅̅ study, it was possible to predict the changing distribution of airflow
r
=
1− ϵ
(8) velocity inside the drying chamber as dehydration proceeds (e.g.
ro 1 − ϵ0 changing array-porosity). It can be observed from Fig. 13, that the ve­
locity distribution significantly changes as the drying proceeds and
The expression given in Equation (8) can be used to relate the airflow hence taking this change not into account would result in large modeling
resistance with the average moisture content change. The shrinkage of a errors in the transient drying process (see the reference distribution in
potato slice versus its dimensionless moisture content (XX0 ) was studied to the figure). Although illustrated for the drying process in a solar dryer,
fit a linear relationship (rr0 = 0.593 + 0.381(XX0 )) (Singh and Talukdar, other drying processes involving the drying of an array of slices food
2019). Substituting this linear equation into Equation (8) results in items can benefit from the results of the current study.
Equation (9):
4. Conclusion
( )2
X
ϵ = (− 1 + ϵ0 ) 0.593 + 0.381 + 1. (9)
X0 In this study, the airflow across an array of sliced food items is
investigated using wind tunnel experiments and numerical simulations.
Equation (9) can be substituted into Equation (7) so as to use it as an The airflow resistances from both studies were presented as a pressure
input for developing appropriate CFD models of transport phenomena in drop and fitted accurately to the Forchheimer equation. The airflow
food dehydration. resistance was observed to be mainly inertial, and being affected by
geometrical parameters like the initial disc diameter and array-porosity.
A model equation was proposed to describe the relationship between the

9
P.D. Tegenaw et al. Journal of Food Engineering 312 (2022) 110739

resistance coefficients and the geometrical parameters. The proposed Getahun, S., Ambaw, A., Delele, M., Meyer, C.J., Opara, U.L., 2017b. Analysis of airflow
and heat transfer inside fruit packed refrigerated shipping container: Part i – model
model equation fits very well with the results from the experiments and
development and validation. J. Food Eng. 203, 58–68. https://doi.org/10.1016/j.
the numerical simulations. The airflow resistances identified in this jfoodeng.2017.02.010.
research were used in the formulation of an appropriate porous media Getahun, E., Delele, M.A., Gabbiye, N., Fanta, S.W., Demissie, P., Vanierschot, M., 2021.
based CFD modeling of the momentum transport phenomena in sliced Importance of integrated cfd and product quality modeling of solar dryers for fruits
and vegetables: a review. Sol. Energy 220, 88–110. https://doi.org/10.1016/j.
food items drying. Although, demonstrated in a solar dryer, the results in solener.2021.03.049.
this study can be incorporated into other macro-scale CFD modeling of Gruyters, W., Verboven, P., Diels, E., Rogge, S., Smeets, B., Ramon, H., Defraeye, T.,
heat and mass transfer phenomena involving sliced food items. Nicolaï, B., 2018. Modelling cooling of packaged fruit using 3d shape models. Food
Bioprocess Technol. 11 (11), 2008–2020. https://doi.org/10.1007/s11947-018-
2163-9.
Author’s contribution Guo, B., Hou, Q., Yu, A., Li, L., Guo, J., 2013. Numerical modelling of the gas flow
through perforated plates. Chem. Eng. Res. Des. 91 (3), 403–408. https://doi.org/
10.1016/j.cherd.2012.10.004.
Petros Demissie Tegenaw: Conceptualization, Methodology, Inves­ Huang, S., Ma, T., Wang, D., Lin, Z., 2013. Study on discharge coefficient of perforated
tigation, Software, Formal analysis, Writing – original draft, Visualiza­ orifices as a new kind of flowmeter. Exp. Therm. Fluid Sci. 46, 74–83. https://doi.
tion, Data curation. Pieter Verboven: Conceptualization, Supervision, org/10.1016/j.expthermflusci.2012.11.022.
Irvine, D.A., Jayas, D.S., Mazza, G., 1993. Resistance to airflow through clean and soiled
Methodology, Writing – review & editing, Resources. Maarten Vanier­ potatoes. Transactions of the ASAE 36 (5), 1405–1410. https://doi.org/10.13031/
schot: Conceptualization, Supervision, Methodology, Writing – review & 2013.28478.
editing, Project administration, Funding acquisition, Resources Kashaninejad, M., Tabil, L., 2009. Resistance of bulk pistachio nuts (ohadi variety) to
airflow. J. Food Eng. 90 (1), 104–109. https://doi.org/10.1016/j.
jfoodeng.2008.06.007.
Declaration of competing interest Knight, C., 2018. Fluid Flow and Drag in Polydisperse Granular Materials Subject to
Laminar Seepage Flow. Imperial College London. Ph.D. thesis.
Madamba, P.S., DriscolL, R.H., Buckle, K.A., 1994. Bulk density, porosity and resistence
The authors declare that they have no known competing financial
to airflow of garlic slices. Dry. Technol. 12 (4), 937–954. https://doi.org/10.1080/
interests or personal relationships that could have appeared to influence 07373939408960003.
the work reported in this paper. Mahiuddin, M., Khan, M.I.H., Kumar, C., Rahman, M.M., Karim, M.A., 2018. Shrinkage
of food materials during drying: current status and challenges. Compr. Rev. Food Sci.
Food Saf. 17 (5), 1113–1126. https://doi.org/10.1111/1541-4337.12375.
Acknowledgement Malavasi, S., Messa, G., Fratino, U., Pagano, A., 2012. On the pressure losses through
perforated plates. Flow Meas. Instrum. 28, 57–66. https://doi.org/10.1016/j.
The Interfaculty Council for Development Cooperation (IRO) of KU flowmeasinst.2012.07.006.
Mery, D., Pedreschi, F., 2005. Segmentation of colour food images using a robust
Leuvenis gratefully acknowledged for funding this research. algorithm. J. Food Eng. 66 (3), 353–360. https://doi.org/10.1016/j.
jfoodeng.2004.04.001.
References Misha, S., Mat, S., Ruslan, M.H., Sopian, K., Salleh, E., 2013. The cfd simulation of tray
dryer design for kenaf core drying. In: Advances in Manufacturing and Mechanical
Engineering, Vol. 393 of Applied Mechanics and Materials, Trans Tech Publications
Amanlou, Y., Zomorodian, A., 2010. Applying cfd for designing a new fruit cabinet dryer.
Ltd, pp. 717–722. https://doi.org/10.4028/www.scientific.net/AMM.393.717.
J. Food Eng. 101 (1), 8–15. https://doi.org/10.1016/j.jfoodeng.2010.06.001.
Neale, M., Messer, H., 1976. Resistance of root and bulb vegetables to airflow. J. Agric.
Amanlou, Y., Zomorodian, A., 2011. Evaluation of air flow resistance across a green fig
Eng. Res. 21 (3), 221–231. https://doi.org/10.1016/0021-8634(76)90079-2.
bed for selecting an appropriate pressure drop prediction equation. Food Bioprod.
Onwude, D.I., Hashim, N., Abdan, K., Janius, R., Chen, G., 2018. Modelling the mid-
Process. 89 (2), 157–162. https://doi.org/10.1016/j.fbp.2010.03.011.
infrared drying of sweet potato: kinetics, mass and heat transfer parameters, and
Beasley, D.E., Clark, J.A., 1984. Transient response of a packed bed for thermal energy
energy consumption. Heat Mass Tran. 54 (10), 2917–2933. https://doi.org/
storage. Int. J. Heat Mass Tran. 27 (9), 1659–1669. https://doi.org/10.1016/0017-
10.1007/s00231-018-2338-y.
9310(84)90278-3.
Onwude, D.I., Hashim, N., Abdan, K., Janius, R., Chen, G., 2019. The effectiveness of
Belessiotis, V., Delyannis, E., 2011. Solar drying. Sol. Energy 85 (8), 1665–1691. https://
combined infrared and hot-air drying strategies for sweet potato. J. Food Eng. 241,
doi.org/10.1016/j.solener.2009.10.001 progress in Solar Energy 1.
75–87. https://doi.org/10.1016/j.jfoodeng.2018.08.008.
Caccavale, P., De Bonis, M.V., Ruocco, G., 2016. Conjugate heat and mass transfer in
Oun, H., Kennedy, A., 2014. Experimental investigation of pressure-drop characteristics
drying: a modeling review. J. Food Eng. 176, 28–35. https://doi.org/10.1016/j.
across multi-layer porous metal structures. J. Porous Mater. 21 (6), 1133–1141.
jfoodeng.2015.08.031 virtualization of Processes in Food Engineering.
https://doi.org/10.1007/s10934-014-9863-y.
Cengel, Y., 2014. Heat and Mass Transfer: Fundamentals and Applications. McGraw-Hill
Özahi, E., 2015. An analysis on the pressure loss through perforated plates at moderate
Higher Education.
Reynolds numbers in turbulent flow regime. Flow Meas. Instrum. 43, 6–13. https://
Darabi, H., Zomorodian, A., Akbari, M., Lorestani, A., 2015. Design a cabinet dryer with
doi.org/10.1016/j.flowmeasinst.2015.03.002.
two geometric configurations using cfd. J. Food Sci. Technol. 52 (1), 359–366.
Pedreschi, F., León, J., Mery, D., Moyano, P., 2006. Development of a computer vision
Defraeye, T., Radu, A., 2017. Convective drying of fruit: a deeper look at the air-material
system to measure the color of potato chips. Food Res. Int. 39 (10), 1092–1098.
interface by conjugate modeling. Int. J. Heat Mass Tran. 108, 1610–1622. https://
https://doi.org/10.1016/j.foodres.2006.03.009 physical Properties VI.
doi.org/10.1016/j.ijheatmasstransfer.2017.01.002.
Ranjbaran, M., Emadi, B., Zare, D., 2014. Cfd simulation of deep-bed paddy drying
Defraeye, T., Verboven, P., Nicolai, B., 2013. Cfd modelling of flow and scalar exchange
process and performance. Dry. Technol. 32 (8), 919–934. https://doi.org/10.1080/
of spherical food products: turbulence and boundary-layer modelling. J. Food Eng.
07373937.2013.875561.
114 (4), 495–504. https://doi.org/10.1016/j.jfoodeng.2012.09.003.
Rapusas, R., Driscoll, R., Srzednicki, G., 1995. Bulk density and resistance to airflow of
Delele, M.A., 2009. Engineering Design of Spraying Systems for Horticulturalapplications
sliced onions. J. Food Eng. 26 (1), 67–80. https://doi.org/10.1016/0260-8774(94)
Using Computational Fluid Dynamics. Bio-ingenieurswetenschappen K.U.Leuven.
00043-9.
Ph.D. thesis. https://limo.libis.be/primo-explore/fulldisplay?docid=LIRIAS171
Resistance to Airflow of Grains, Seeds, Other Agricultural Products, and Perforated Metal
4483&context=L&vid=Lirias&search_scope=Lirias&tab=default_tab&lang=en_US.
Sheets, Standard, 2016. ASABE STANDARDS 2009, St. Joseph, MI.
Delele, M., Vorstermans, B., Creemers, P., Tsige, A., Tijskens, E., Schenk, A., Opara, U.,
Roache, P.J., 1994. Perspective: a method for uniform reporting of grid refinement
Nicolaï, B., Verboven, P., 2012. Cfd model development and validation of a
studies. J. Fluid Eng. 116 (3), 405–413. https://doi.org/10.1115/1.2910291.
thermonebulisation fungicide fogging system for postharvest storage of fruit. J. Food
Sanghi, A., Ambrose, R.P.K., Maier, D., 2018. Cfd simulation of corn drying in a natural
Eng. 108 (1), 59–68. https://doi.org/10.1016/j.jfoodeng.2011.07.030.
convection solar dryer. Dry. Technol. 36 (7), 859–870. https://doi.org/10.1080/
Devereau, A., 1994. Tropical Sweet Potato Storage: A Literature Review. University of
07373937.2017.1359622.
Greenwich, UK, Chatham, UK. Technical report. http://gala.gre.ac.uk/id/epri
Singh, P., Talukdar, P., 2019. Determination of shrinkage characteristics of cylindrical
nt/13793/.
potato during convective drying using novel image processing technique. Heat Mass
Doymaz, İbrahim, 2004. Convective air drying characteristics of thin layer carrots.
Tran. 1. https://10.1007/s00231-019-02771-2.
J. Food Eng. 61 (3), 359–364. https://doi.org/10.1016/S0260-8774(03)00142-0.
Sun, Q., Zhang, M., Mujumdar, A.S., 2019. Recent developments of artificial intelligence
Dukhan, N., Patel, P., 2008. Equivalent particle diameter and length scale for pressure
in drying of fresh food: a review. Crit. Rev. Food Sci. Nutr. 59 (14), 2258–2275.
drop in porous metals. Exp. Therm. Fluid Sci. 32 (5), 1059–1067. https://doi.org/
https://doi.org/10.1080/10408398.2018.1446900 pMID: 29493285.
10.1016/j.expthermflusci.2007.12.001.
Tegenaw, P.D., Gebrehiwot, M.G., Vanierschot, M., 2019b. On the comparison between
Ergun, S., Orning, A.A., 1949. Fluid flow through randomly packed columns and
computational fluid dynamics (cfd) and lumped capacitance modeling for the
fluidized beds. Ind. Eng. Chem. 41 (6), 1179–1184. https://doi.org/10.1021/
simulation of transient heat transfer in solar dryers. Sol. Energy 184, 417–425.
ie50474a011.
https://doi.org/10.1016/j.solener.2019.04.024.
Getahun, S., Ambaw, A., Delele, M., Meyer, C.J., Opara, U.L., 2017a. Analysis of airflow
Tegenaw, Petros, Hayelom, M., Kassaye, A., Hailesilassie, A., Gebrehiwot, M.,
and heat transfer inside fruit packed refrigerated shipping container: Part ii –
Vanierschot, M., 2019a. Design, development and cfd modeling of indirect solar food
evaluation of apple packaging design and vertical flow resistance. J. Food Eng. 203,
83–94. https://doi.org/10.1016/j.jfoodeng.2017.02.011.

10
P.D. Tegenaw et al. Journal of Food Engineering 312 (2022) 110739

dryer. Energy Procedia 158, 1128–1134. https://doi.org/10.1016/j. Velić, D., Planinić, M., Tomas, S., Bilić, M., 2004. Influence of airflow velocity on kinetics
egypro.2019.01.278 innovative Solutions for Energy Transitions. of convection apple drying. J. Food Eng. 64 (1), 97–102. https://doi.org/10.1016/j.
Udomkun, P., Argyropoulos, D., Nagle, M., Mahayothee, B., Janjai, S., Müller, J., 2015. jfoodeng.2003.09.016.
Single layer drying kinetics of papaya amidst vertical and horizontal airflow. LWT - Verboven, P., Hoang, M., Baelmans, M., Nicolai, B., 2004. Airflow through beds of apples
Food Sci. Technol. (Lebensmittel-Wissenschaft -Technol.) 64 (1), 67–73. https://doi. and chicory roots. Biosyst. Eng. 88 (1), 117–125. https://doi.org/10.1016/j.
org/10.1016/j.lwt.2015.05.022. biosystemseng.2004.02.002.
van der Sman, R., 2002. Prediction of airflow through a vented box by the Verboven, P., Flick, D., Nicolaï, B., Alvarez, G., 2006. Modelling transport phenomena in
Darcy–forchheimer equation. J. Food Eng. 55 (1), 49–57. https://doi.org/10.1016/ refrigerated food bulks, packages and stacks: basics and advances. Int. J. Refrig. 29
S0260-8774(01)00241-2. (6), 985–997. https://doi.org/10.1016/j.ijrefrig.2005.12.010 issue with Special
Vanierschot, M., Timmermans, J., Van den Bulck, E., 2013. Application of Particle Image Emphasis on Data and Models on Food Refrigeration.
Velocimetry (Piv) in the Study of Perforated Plate Wake Flow, Dirckx, Joris. Shaker Yadollahinia, A., Jahangiri, M., 2009. Shrinkage of potato slice during drying. J. Food
Publishing BV, Maastricht, pp. 423–431. Eng. 94 (1), 52–58. https://doi.org/10.1016/j.jfoodeng.2009.02.028.
Veleşcu, I., Ţenu, I., Cârlescu, P., Dobre, V., 2013. Convective air drying characteristics Ye, X.L., Su, Y.B., Guo, B.Y., Yu, A.B., 2016. Multi-scale simulation of the gas flow
for thin layer carrots. Bull. Univ. Agric. Sci. Vet. Med. Cluj-Napoca. Food Sci. through electrostatic precipitators. Appl. Math. Model. 40 (21), 9514–9526. https://
Technol. 70, 11. https://doi.org/10.15835/buasvmcn-fst:9619. doi.org/10.1016/j.apm.2016.06.023.

11

You might also like