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

net/publication/277005071

CFD Simulation of fixed bed dryer by using porous media concepts: Unpeeled
longan case

Article in International Journal of Agricultural and Biological Engineering · January 2013

DOI: 10.3965/j.ijabe.20130601.0010

CITATIONS READS

7 619

4 authors, including:

Sarawut Phupaichitkun
Silpakorn University
8 PUBLICATIONS 38 CITATIONS

SEE PROFILE

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

Sorption isotherm of longan fruit View project

All content following this page was uploaded by Sarawut Phupaichitkun on 21 May 2015.

The user has requested enhancement of the downloaded file.

 100 March, 2013 Int J Agric & Biol Eng Open Access at http://www.ijabe.org Vol. 6 No.1

CFD simulation of fixed bed dryer by using porous media

concepts: Unpeeled longan case

Wuttichai Prukwarun, Wasan Khumchoo, Waraporn Seancotr,

Sarawut Phupaichitkun*
(Department of Material Science and Engineering, Faculty of Engineering and Industrial Technology,
Silpakorn University, Nakorn Pathom, 73000, Thailand)

Abstract: Quality of dried product depends on the temperature and velocity at each position in the dryer. Simultaneous
microscopic and macroscopic simulation on Computational Fluid Dynamic (CFD) is a general problem of fixed bed dryer
consisting of water transportation in porous media and dynamic flow of hot air in the dryer. Simplifying the dryer by
assuming the packed bed as porous volume, viscous and inertial resistances (1/ and C2) are necessary for calculating the
pressure drop and velocity change in the bulk. Comparing the P/L of the standard packing with experimental results, the
porosity and resistance parameters can be estimated. Simulation of unmodified, adding false floor and invest mesh, and
insulating the dryer wall are used for validation with previous results. Adding the round holed sieve as false floor and invert
mesh can produce better profile but cannot obtain uniform distribution. Air velocity distribution shows similar but the
calculating temperature is higher than that from the experiment. By analysis of thermal efficiency of dryer without insulator,
the heat loss rates with flue gas and heat flux at wall are in the range 14%-17% and 5.5%-7.3%. Integrating with single fruit
or thin layer drying kinetic in the future, the CFD simulation can be used for optimal design of fixed bed dryer.
Keywords: deep-bed dryer, longan, velocity and temperature distribution, Computational Fluid Dynamic (CFD), ANSYS
DOI: 10.3965/j.ijabe.20130601.0010

Citation: Prukwarun W, Khumchoo W, Seancotr W, Phupaichitkun S. CFD simulation of fixed bed dryer by using porous
media concepts: Unpeeled longan case. Int J Agric & Biol Eng, 2013; 6(1): 100－110.

 energy[1,2]. Dryer design for declining the temperature

1 Introduction
and velocity gradient in the bulk assures the moisture
Computational Fluid Dynamic (CFD) is a powerful homogeneity of dried product or storage food. Heat and
tool for calculating velocity profile of fluid flow and mass transfer with air flow in the bulk is necessary to
visualizing the air flow trajectory in the dryer for the calculate the rate of decline in water content from the
purpose of obtaining the better quality and conserving the product. Water can be transported through processes
such as water sorption kinetic at surface, water
Received date: 2012-11-29 Accepted date: 2013-01-15 evaporation, diffusing rate of inner liquid water and water
Biographies: Wuttichai Prukwarun, B.Sc. on Petrochemical and vapor from inside through the structure barrier and
Polymeric Materials, Email: [email protected].
convective water transfer from surface to bulk of air flow.
Wasan Khumchoo, B.Sc. on Petrochemical and Polymeric
Materials. Email: [email protected]. Waraporn CFD simulation is associated with transport phenomena
Seancotr, B.Sc. on Petrochemical and Polymeric Materials. Email: calculation of complex geometry in both microscopic and
[email protected]. macroscopic level. However, calculating in micron
*Corresponding author: Sarawut Phupaichitkun, PhD,
scale of food requires further information about geometry,
Agricultural Science, Lecturer, Department of Material Science and
Engineering, Faculty of Engineering and Industrial Technology,
material, and its properties which are inaccessible by
Silpakorn University, 73000 Nakorn Pathom, Thailand; Fax: +66 experiment. Simulation of the air flowing through the
34 219361; Email: [email protected]. fixed bed of food by using packing geometry in
 March, 2013 CFD simulation of fixed bed dryer by using porous media concepts: Unpeeled longan case Vol. 6 No.1 101

microscopic scale requires powerful computer to reduce Drying experiments were carried out at a farmer
the long data processing time. A Simple geometry by factory in the village of San Pa Tong, Chiang Mai,
considering the bulk of particles as a porous zone volume Thailand during 2004-2007 under collaboration project
lessens the calculation node and convergence time. between Hohenhiem University (Germany), Mae Jo
Simulation of the velocity and temperature distribution of University, Chaing Mai University and Silpakorn
hot air in the dryer by assuming the food as porous zone University (Thailand) (see Table 1). About two tons of
[3,4]
has been studied by Molenda . longan cultivar ‘E-Dor’ (or ‘Daw’) were dried in a fixed
Unpeeled longan dryer has an advantage for CFD bed dryer, called ‘Taiwan’ type: Suncue SKS-480A,
simulation and validation. This is due to the round Taichung, Taiwan; mfd. 1996. The geometry of
shape of fruit which is related to the spherical particle unmodified dryer and conventional drying procedure was
[5]
packed in the column as explicated by Eugan . The described by Phaphuangwittayakul et al[6]. The air was
fruit contains large amount of water and is dried for 48 supplied by axial fan blower (0.745 kW) and heated by
hours which is long enough to investigate the temperature liquefied petroleum gas (LPG) burner (48 kW/h). The
and velocity contour. Water evaporation at the surface air flows through round holed (8 mm) metal sieve which
is not affected by the temperature distribution of air also uses for supporting the bulk of fruits presented in
because the drying rate is very slow. The effect of hot Figure 1. L. Henz[7] investigated the thermal efficiency
air velocity on drying rate is less than the temperature by measuring ambient temperature, temperature of the air
therefore the drying kinetic is controlled by the inner before and after burner, temperature inside the longan bed
diffusion of water in the fruit. Less changing of bulk and longan fruit and air flow outlet and varying the air
porosity regarding the fruit has no shrinkage. It is flow rate at inlet (after the burner) from 6.0 m/s to
necessary to improve the uniformity of the momentum, 3.45 m/s. Dryer was modified by inserting the wood
heat and mass transfer by redesign of CFD software, wall to increase the bed height from 0.475 m to 0.600 m.
validation of hot air flow through the bulk by assuming as The longan fruits were separated into three layers by
the porous volume. enclosing them in a plastic net and later layers were
1.1 Unpeeled longan dryer and its modification and shifted and rotated to 180 degree after every shifting
experiment procedure.

Table 1 Drying condition of all trial experiments

Bed height Inlet velocity Outlet velocity Average ambient Average temperature Pressure drop
Experiment code Dryer
/m /m·s-1 /m·s-1 temperature/℃ in bulk/℃ /Pa

Henz (2005)[7]
Exp05A 0.475 6.00 0.17¹ 30.0 61.0 15.05 Unmodified
Exp05B 0.475 6.00 0.17¹ 29.0 60.7 15.08 Unmodified
Exp05C 0.475 6.00 0.17¹ 28.8 61.4 18.25 Unmodified
Exp05D 0.475 3.45 0.10¹ 28.2 62.7 20.97 Unmodified
[8]
Azcárraga (2006)
Exp06A 0.60 5.15² 0.15 28.5 56.7 15.29 Shifting A
Exp06B 0.60 4.88² 0.14 29.8 57.0 16.03 Shifting B
Exp06C 0.60 3.93² 0.11 29.5 55.8 16.22 Shifting C
Exp06D 0.60 na na na na na Shifting D
[9]
Nagle et al. (2008)
Exp07A 0.60 2.17² 0.06 29.4 55.3 na Unmodified
Exp07B 0.60 3.06² 0.09 29.5 57.0 na Invertion
Exp07C 0.60 1.91² 0.05 29.2 60.4 na False
Exp07D 0.60 3.76² 0.11 28.3 61.9 na Insulation
Note: ¹estimated from inlet velocity; ²estimated from outlet velocity; na: not available data.
 102 March, 2013 Int J Agric & Biol Eng Open Access at http://www.ijabe.org Vol. 6 No.1

This method improved the uniformity of the drying

rate and gained better product homogeneity which
developed by local farmer in 2005. Four applied methods
of shifting procedures (Figure 2) were investigated in term
of product uniformity and thermal efficiency improvement
by Azcárraga[8]. In 2008, Nagle and Azcárraga[9]
modified the dryer with (a) perforated sheet metal inside
the plenum, inverted against the air entrance, (b) false
plywood floor inside the plenum, inclined toward the air
entrance, and (c) insulation of the plenum floor and dryer
walls with fiber glass wool for improving the thermal
efficient and air distribution. The dryer performance
and product quality was reported by Nagle et al[9-10]. Figure 1 Longan dryer (‘Taiwan’ type) illustrated by Azcárraga[8]

Figure 2 Longan drying experiment for investigating the effect of shifting method by Azcárraga [8] (left) and the efficiency of three
modified dryers compared with original dryer by Nagle et al.[10] (right)

1.2 Fluid flow in porous media P  ( 1  2)  (1 2)

Ergun A vs  B vs (1)
Fluid flowing through the bulk of spherical particles, L d 2 3
d 3
and the pressure drop across the fixed bed are related to P 
Forchheimer  vs   Cvs2 (2)
the particle size, particle shape, particle alignment and L K
void fraction (or bulk porosity). The correlation P   C2 2
ANSYS  v v (3)
between fluid velocity and the pressure drop per the L  2
length/height of bed (P/L) was described by Ergun[5] in These equations are comparable with the pressure
1952. Ergun equation is shown in Equation (1) resulting drop equation in the porous volume (porous zone) and
of the Blake–Kozeny equation and Burke–Plummer porous wall (porous jump) on ANSYS software presented
equation. The Blake–Kozeny–Carman constant (A) and in Equation (3). The viscous resistance (1/) and
Burke–Plummer constant (B) are 150 and 1.75, inertial resistance (C2) are two important coefficients
respectively. Forchheimer presents this equation as which are assessable by using Ergun equation with
Equation (2) for high flow rate in porous media which diameter of particle (d) and bulk porosity () which is
contains two specific coefficients for the particle bed depended on the particle packing in the bulk[11].
named permeability tensor (K) and the inertial drag or 1.3 Effect of particle packing on the bulk porosity
form tensor (C). Both parameters depend on the grain porosity in a
 March, 2013 CFD simulation of fixed bed dryer by using porous media concepts: Unpeeled longan case Vol. 6 No.1 103

specific volume inside the bin during drying or storage, in Figure 4a. Air flows from the bottom to the top of
which is influenced by the type of grain, non-uniformity box. Determine the pressure drop across the sieve by
of bulk density, coefficient of friction between the grain varying the temperature of air flow from 40 to 100℃ at
[12,13]
and wall, moisture content, and filling method . 0.1 m/s and varying the air velocity from 0.01 to 100 m/s
Figure 3 presents the model of three standard packing every ten time magnitude while the temperature was fixed
patterns which are primitive, body center cubic (BCC) at 80℃. Similarly, the air flow through the longan
and face center cubic (FCC) with their porosity value. packed sizes 75×75×75 mm3, the bed height is kept
Comparing the pressure drop value with these standard constant at 50.2 mm. The diameter of each longan fruit
packing forms, the equivalent porosity of longan bed is (dp) is 25 mm and packed in primitive, BBC and FCC
possible to estimate which is necessary for calculating patterns are shown in Figures 4b-4d. Correlation
two resistance coefficients of longan bulk using ANSYS. between air velocity and pressure drop per length was
described in Ergun equation. The 1/ and C2 coefficient
were determined from curve fitting as a second order
polynomial function on Microsoft Excel.

Figure 3 Illustration of bulk porosity of primitive, BCC and FCC

packing

To unravel these problems for CFD simulation in this

research, the number of calculating node were reduced by
considering the fixed longan bed as the porous volume.
The holed sieve is considered as the porous wall.
Important factors (viscous and inertia loss coefficient)
and equivalent porosity of longan bulk are estimated from
the pressure drop and air velocity of previous drying
results. The overall objective of this study is to conduct
simulation of the temperature and air velocity profiles in
the dryer by using ANSYS (FLUENT) software and to Figure 4 Geometry of simulation volume of (a) 8 mm holed sieve
validate the results with the previous experiment. Finally, and longan fruit packed as (b) primitive (c) BCC and (d) FCC

a simple dryer model will be developed for redesign the

dryer in the future. 2.2 Simulation of the dryer with porous zone and
porous jump
2 Materials and methods
Geometry of the unmodified longan dryer for CFD
2.1 Viscous and inertia loss coefficients of holed sieve simulation is shown in Figure 5. The horizontal cross
2
and packed longan section area is 2.5×2.5 m with 0.875 m height for all
Microscopic simulation of hot air flowing through the experiments by Henz[7] and 1.0 m. Volume of air is
sieve with 8 mm diameter round holed and flowing simplified into two volumes of fluid. The first volume
through three types of longan packing are running on is the part with 0.4 m height called the plenum, the empty
ANSYS (FLUENT) V14 (ANSYS, Inc. Pennsylvania, space for homogenizing the hot air temperature and
US). Simulation of hot air flow through the sieve (50× velocity before flowing through sieve. Sieve is
50 mm) placed at one-third of air volume height as shown considered as the porous jump. The second volume is
 104 March, 2013 Int J Agric & Biol Eng Open Access at http://www.ijabe.org Vol. 6 No.1

air volume with the bulk of longan fruit and it is assumed 7.0 and the temperature contour taken by thermal camera
as a porous zone. The porosity and two resistance which was reported by Nagle et al[10].
coefficients of the sieve and longan bed is estimated from
3 Results and discussion
the previous section. The size of hot air inlet is
0.40×0.40 m2 and was placed at the front of dryer in the 3.1 Resistance coefficients of round holed sieve
x-direction. The side wall of the dryer was made from The average Reynold number (Re) at inlet and
steel with wall thickness of 3 mm. outflow of unmodified dryer running at 6.0 m/s air
velocity is lower than 1.0×105 which is in laminar
region[15]. The hot air flows through the sieve at 0.1 m/s
and variation of the temperature from 40℃ up to 100℃
were simulated when the pressure drop across the sieve is
constant. The pressure drop per bed height (P/L) is a
function of velocity only. The momentum resistance
coefficients play an important role with the size of
particle and porosity of bulk, which can be described by
Ergun relation. Fitting the curve between P/L and
velocity using the second order polynomial with zero
Figure 5 Geometry of longan dryer after simplify the longan bulk intercept, the viscous and inertia loss coefficients (1/
to porous zone and C2) of the sieve are determined and presented in
Table 2.
The results of the pressure drop for the first two years 3.2 Resistance coefficient and equivalent porosity of
of the experiment were used to estimate the equivalent packed of longan
porosity of longan bulk by comparing the pressure drop The results represent that the simulation of drying air
value with the three packing structures. Hence, this flowing through 75×75×75 mm3 consisting longan packed
porosity will be used to simulate the hot air flow of three
with primitive, BCC and FCC patterns, show no
modified dryers which was investigated by Nagle et al[10].
differences between P/L simulated on ANSYS and
The ambient temperature is an average value at 25℃; the
calculated from Ergun equation. This confirms the
inlet air velocity is 6 m/s. Enthalpy of inlet flow and
Blake-Kozeny-Carman constant (A) and Burke-Plummer
flue gas, and heat loss at wall of dryer is estimated and
constant (B) of Ergun equation are 150 and 1.75,
used for thermal efficiency analysis.
respectively. The 1/ and C2 of packed longan bulk
A low velocity of drying air resulted in a Mach
estimated from curve fitting are presented in Table 2.
number less than 0.1. Therefore, fluid can be considered
The P/L simulation of whole dryer (longan bed and
as incompressible flow. Dried air contained small
amount of vapor water while the viscosity remains sieve) for all experiments is presented in Figure 6. The

constant therefore, fluid can be assumed as a Newtonian correlation of regression

fluid. With constant density and low velocity, longan dryer with real porosity of packed longan is a
simulation with laminar model still validate for power function (R2 = 0.8879). The equivalent porosity
simulation in this study. Simulations were conducted of longan bed is 0.365 which is necessary for estimating
until continuity residue reaches a stable value and lower the 1/ and C2 of longan bed by using the Ergun equation.
than 6.3×10-5 for default convergence criterion of This porosity value was explained as a random packing of
[14] spherical particle which the porosity value is in the range
ANSYS (Fluent) . The simulated results of air
velocity and temperature distribution were compared with of 0.35-0.41[16]. The experimental pressure drop (P, Pa)
velocity contour which was estimated by using MATLAB value was measured by Henz[7] and Azcárraga[8]. The
 March, 2013 CFD simulation of fixed bed dryer by using porous media concepts: Unpeeled longan case Vol. 6 No.1 105

value is in the range of 15.2-20.9 Pa, depending on the air small. Non-uniform and no directional trajectory of air
velocity. flow in empty dryer is shown in Figure 7A. Simulation
Table 2 Resistance coefficients (1/ and C2) of Ergun presents the low efficient in distributing the hot air of
Equation of each dryer part and the porosity of each part of metal sieve. The Re of fluid flow in the plenum under
dryer the sieve is in the range of 85.79- 4116.34. The area
Viscous Inertial resistance Theoretical Equivalent
Dryer parts
resistance (1/) (C2) porosity () porosity ()
weight average value of Re is 1880.77 with very high
Tray 5.1544×106 2511.67 - - deviation (SD=1113.55) which represents high non-
Pack bed uniformity of velocity.
7
- Primitive 1.1175×10 460.64 0.482 0.356 According to CFD simulation of the dryer with
7
- BCC 3.2418×10 1799.80 0.319 0.260
longan bulk, the pack of spherical fruit does not only
- FCC 1.7994×108 1535.12 0.268 0.167
have low flow resistance but also distributes the hot air to
- Longan 4.3853×106* 4003.29* - 0.365
Note: * calculated by Ergun equation with porosity of 0.365. get the uniform velocity, presents in Figure 7B. The
average Re values of fluid flow inner the packed bed after
the sieve and at the outlet are 293.49 and 339.08,
respectively. The standard deviation value decreases
from 66.15 at sieve to 2.77 at outlet position. Packing of
longan fruit improves the air velocity distribution in the
dryer.
Measured velocity distribution generated from 25
points on a cross section area is presented in Figure 7C.
The contour plot of velocity from CFD simulation
presents that the air velocity nearby the dryer wall is zero
regarding the viscous effect. Measuring the air velocity
Figure 6 Correlation between P/L and porosity of longan bulk
at this point is not possible therefore this effect is not
3.3 Air velocity contour shown on contour plot from experimental results. The
The path line by velocity magnitude (m/s) of air flow air velocity near the burner measured from the dryer is
through (a) the empty dryer and (b) longan bulk height of the negative value. This might come from the air
0.475 m with porosity of 0.365 are presented in Figures flowing reverse from above the dryer due to the air
7A and 7D. Empty dryer with low momentum resist, temperature difference between the sieve and ambient
the effect of sieve on trajectory of air circulation is very which are not able to be found in simulation results.
 106 March, 2013 Int J Agric & Biol Eng Open Access at http://www.ijabe.org Vol. 6 No.1

Figure 7 CFD post processing of empty dryer (A-C) and dryer with packed bed with porosity equal 0.365 (D-F) as the pathlines of
velocity magnitude (A and C), the velocity distribution at the top layer of dryer (B and E), comparing with the velocity distribution from
experiment (C and F) by Nagle et al.[9]

3.4 Air temperature contour drying air near the outlet is higher than that in the real
Temperature contour in horizontal cross section near dryer which might be caused by no evaporative cooling
the wall of dryer simulated by ANSYS (FLENT) and effect at the longan surface during drying. Concerning
measured with thermal camera, is shown in Figure 8. this, the heat consumption according to water evaporation
The high temperature in plenum region of unmodified can be calculated through each node by adding user
dryer and modified dryer (false floor) is also seen in the define function (UDF) of the single fruit drying model or
simulation results. The temperature simulation of thin layer model.

Figure 8 CFD post processing of temperature distribution in unmodified and false floor comparing with thermo picture investigated by
Nagle et al.[10]
 March, 2013 CFD simulation of fixed bed dryer by using porous media concepts: Unpeeled longan case Vol. 6 No.1 107

The thermal picture and the contour plot of simulated that of wall near the inlet. Therefore the farmer has to
air temperature of unmodified dryer, modified dryer with rotate the longan layers 180 degree every shifting time for
additional false floor are presented in Figure 9. The obtaining uniform drying rate. Additional sieve for
wall temperature opposite the inlet is always higher than improving the air distribution purpose gets the better

Figure 9 CFD post processing of temperature distribution in (A) unmodified, (B) false floor and (C) invert mesh, comparing with
experimental results (D-F) measured by Nagle et al.[9]
 108 March, 2013 Int J Agric & Biol Eng Open Access at http://www.ijabe.org Vol. 6 No.1

trajectory in the plenum, however this still inefficient in contour shows better distribution but still not uniform
term of distribution. Figures 9A to 9C presents the after modified by adding sieve in invert and false
temperature distribution of air in vertical cross section direction. Invert floor result presents the highest flow
area in the longan bulk. Temperature contour rendering resistance. With lower air velocity, the momentum
from 25 point of measured value is shown in Figures transfer is small and less convective heat transfer,
9D-9E. CFD simulation shows similar temperature therefore the temperature of this case is higher than
contour as experimental results. The temperature others.

Table 3 Thermal analysis of trial longan experiment

[7]
Information Henz (2005) Azcárraga (2006)[8] Nagle et al. (2008)[9]
Drying experiments Investigate efficiency of dryer Investigate shifting method Modified dryer

Modified drying method Exp05A Exp05B Exp05C Exp05D Exp06A Exp06B Exp06C Exp06D Exp07A Exp07B Exp07C Exp07D
Dryer Original Original Original Original Bed hight Bed hight Bed hight Bed hight Unmodified Invertion False Insulation
Bed depth (m) 0.475 0.475 0.475 0.475 0.6 0.6 0.6 0.6 0.6 0.6 0.6 0.6
Material information
Initial moisture content (%) 69.56 73.50* 70.41 70.75 73.80 72.60
Final moisture content (%) 13.34 19.43* 18.48* 16.83* 9.85* 23.07 19.23 21.04 16.30
Mass of fresh fruit (kg) 2187 2107 2325 2220 1289 1944 1927 1990 1960 1806 1947 1987
Mass of dried fruit (kg) 706 1191 851 690 424 632 614 585 898 822 841 876
Mass of evaporated water (kg) 1481 916 1474 1530 865 1312 1313 1405 1062 984 1106 1111
Drying conditions
Drying time (hr) 51.3 51.6 47.5 47.8 51.6 56.3 58.0 53.8 54.8* 57.0* 54.8* 53.3*
Air flow rate (m³/hr) 3600 3600 3600 2070 3091 2930 2360 na 1301 1834 1147 2258
Drying experiment data
Fuel consumption (kg of LPG) 179 174 na 171 186 194 203 181 194 171 196 187
Electrical consumption (MJ) 137 138 127 128 138 151 156 144 147 153 147 143
air velocity at inlet channel (m/s) 6.00 6.00 6.00 3.45 5.15* 4.88* 3.93* na 2.17* 3.06* 1.91* 3.76*
air velocity at outlet channel (m/s) 0.17* 0.17* 0.17* 0.10* 0.15 0.14 0.11 na 0.06 0.09 0.05 0.11
Enthalpy of inlet air (MJ/kg) 130.79 126.56 125.69 70.57 106.73 105.57 84.34 na 46.30 65.60 40.56 77.39
Enthalpy of outlet air (MJ/kg) 236.54 235.14 237.82 139.66 188.64 179.76 141.74 na 77.51 112.48 74.51 150.53
Energy consumption and loss
Heat from LPG combustion (MJ) 8917 8668 na 8529 9293 9695 10119 9016 9681 8539 9783 9333
Heat consumption from water
-3343 -2067 -3327 -3453 -1952 -2961 -2963 -3171 -2397 -2221 -2496 -2508
evaporation (MJ)
Heat loss with flue gas (MJ) -5420 -5600 na -3299 -4223 -4178 -3329 na -1711 -2674 -1861 -3900
Heat loss at wall of dryer (MJ) -498 -501 na -464 -501 -547 -564 -522 -533 -628 -571 0
Others heat loss (MJ) -344 500 na 1312 2617 2008 3263 na 5041 3016 4855 2925
Others heat loss (%) -4+ 6 na 15 28 21 32 na 52 35 50 31
Economical calculations
Specific Energy for water
6.11 9.61 na 5.66 10.90 7.50 7.83 na 9.25 8.83 8.98 8.53
evaporation (MJ/kg)

Specific Energy for product

12.82 7.39 na 12.55 22.24 15.58 16.73 na 10.94 10.57 11.81 10.82
weight (MJ/kg)

Energy cost for product weight

5.56 3.20 na 5.44 9.64 6.75 7.25 na 4.74 4.57 5.12 4.69
(THB/kg)

Remarks: na: no data available; * estimated from data, + unreal data.

Thermal parameters of the dryers were determined of and electrical consumption which is the results of the fuel
twelve experiments from 2005 to 2007 is presented in consumption and drying time. Observation of air
Table 3. The standard heating value of LPG in Thailand velocity and temperature from trial experiments, the
is 49.88 MJ/kg with the cost of 0.44 THB/MJ. Total average value of them can be used for estimating the
heat requirement was estimated from the combustion heat enthalpy of inlet flow and outlet flow. Nagle et al.[10].
 March, 2013 CFD simulation of fixed bed dryer by using porous media concepts: Unpeeled longan case Vol. 6 No.1 109

have concluded that concerning the energy consumption with CFD software. This can be estimated from the
and thermal efficiency, the modified dryer with pressure drop per length and air velocity by using Ergun
insulation is the best alternative.. Additional holed equation. From simulation results, packing of spherical
sieves (invested and false floor) increased the drying rate fruit improves the air distribution. Since the optimal air
[10]
but only in the bottom level . Consistent with the velocity covers the wall with insulator, more efficient air
simulation results, heat loss is zero based on no heat flux distributor does not improve the non-uniform temperature
at the wall. Total heat loss of insulated wall dryer is and velocity distribution and both shifting and rotating
minimized (31%) when compared with that of other the layers at 180℃ are necessary for homogenizing the
modified dryer (35%-52%). Simulation of modified dried product. Integrating the water evaporation by
dryer with sieve results in higher velocity and applying the drying kinetic of single fruit or thin layer is
temperature in the bottom layer. Especially higher air important for temperature distribution and moisture
velocity at the wall opposite the inlet channel causes distribution during drying.
higher drying rate by convection.
Without insulation material covering, the invert mesh Acknowledgments
trial represents the higher thermal efficiency in term of This work was financed by Department of Material
specific energy for water evaporation, specific energy for Science and Engineering and Faculty of Engineering and
product weight and energy cost for product weight. The Industrial Technology, Sipakorn Univesity, THAILAND.
fourth experiment in 2005 (EXP05D) gave the lowest Authors would like to specially thank Prof. Joachim
specific energy for water evaporation. In terms of Mueller and Dr.agr.sc. Marcus Nagle, Institute ATS 440e,
slow drying rate of longan fruit, lower air velocity is still Hohenhim University, Stuttgart, Germany, for supporting
high enough to remove the water at the surface. and allowing us to use the experimental data of the dryer
Analogy of this phenomenon was also found in the trail for validation.
experiment in 2007. From CFD simulation, the enthalpy
loss with flue gas (17%-40%) and convective heat Nomenclatures
transfer at wall (5.5%-7.3%) are important thermal A The Blake–Kozeny–Carman constant
energy loss of the dryer. Covering the wall with B Burke–Plummer constant
insulation can save energy about 5%. The air inlet ∆P/L Pressure drop per unit length (Pa/m)
velocity of 6 m/s is too high. Additional sieve as false μ Viscosity of fluid
floor and invert mesh is not a good air distributor. ρ Density of fluid (kg/m3)
Calculation of water evaporation inner the bulk improved vs Average velocity of fluid flow through a cross
the CFD simulation but required the single fruit or thin section area (m/s)
layer drying model as UDF. Furthermore, simulation of ε Porosity
the new modified dryer concerning optimal air velocity, d Diameter of bed particle (m)
heat loss reduction and modern distributor design can be K Permeability tensor
possible with CFD software. C Inertial drag or form tensor

4 Conclusions C2 Inertial Resistance

1/ Viscous Resistance
The unpeeled longan bulk drying is a good sample for
fixing bed dryer. Transport phenomena of hot air flow
[References]
through the packed bed and hold sieve can be simplified
[1] Norton T, Sun D W. Computational fluid dynamics
by assuming this volume as the porous media (porous
(CFD)-an effective and efficient design and analysis tool for
zone and porous jump). Viscous and inertial resistances the food industry: A review. Trends in Food Science &
(1/ and C2) are important coefficients for simulating Technology, 2006; 17(11): 600-620.
 110 March, 2013 Int J Agric & Biol Eng Open Access at http://www.ijabe.org Vol. 6 No.1

[2] Xia B, Sun D W. Applications of computational fluid Mahayothee B, Haewsungcharern M, Janjai S, et al. Effects
dynamics (CFD) in the food industry: A review. Computers of operating practices on performance of a fixed-bed
and Electronics in Agriculture, 2002; 34(1-3): 5-24. convection dryer and quality of dried longan. International
[3] Lawrence J, Maier D E. Three-dimensional airflow Journal of Food Science & Technology, 2008; 43(11):
distribution in a maize silo with peaked, levelled and cored 1979-87.
grain mass configurations. Biosystems Engineering, 2011; [10] Nagle M, 1Azcárraga J C G, Mahayothee B,
110(3): 321-329. Haewsungcharern M, Janjai S, Müller J. Improved quality
[4] Molenda M, Montross M D, McNeill S G, Horabik J. and energy performance of a fixed-bed longan dryer by
Airflow resistance of seeds at diferent bulk density using thermodynamic modifications. Journal of Food Engineering,
Ergun's equation. Transactions of the ASAE, 2005; 48(3): 2010; 99(3): 392-399.
1137-1145. [11] ANSYS. Ansys Fluent Theory Guide. Canonsburg, PA:
[5] McCabe W L, Smith J C, Harriott P. Unit Operations of ANSYS Inc., 2011; 786 p.
Chemical Engineering. 6th Ed. New York: McGraw-Hill [12] Molenda M, Montross M D, McNeill S G, Horabik J.
Higher Education; 2001. Airflow resistance of seeds at different bulk density using
[6] Phaphuangwittayakul W, Alikhani Z, limpiti S. A batch Ergun’s equation. Transactions of the ASAE, 2005; 48(3):
dryer for un-peeled longan drying. Agricultural 1137-45.
Mechanization in Asia, Africa and Latin America, 2004; [13] Thompson S A, McNeill S G, Ross IJ, Bridges T C.
35(1): 41-44. Packing factors of whole grains in storage structures.
[7] Henz L C. Thermal efficiency investigation of a longan Applied Engineering in Agriculture, 1997; 3(2): 215-221
dryer in Chiang Mai, Thailand. Project report, Stuttgart, [14] ANSYS. ANSYS FLUNT 13.0 user’s guide. Canonsburg, PA:
Germany: University of Hohenheim; 2005. ANSYS, Inc.; 2009.
[8] Azcárraga J C G. Performance investigation of a Taiwan [15] Perry R H. Perry’s chemical engineer's handbook. New
type flatbed dryer currently used for longan (Dimocarpus York, USA: McGraw-Hill Companies, Inc., 1999.
longan Lour.) in Chiang Mai, Thailand. Master Thesis, [16] Ouchiyama N, Tanaka T. Porosity Estimation for Random
Stuttgart, Germany: University of Hohenheim; 2006. Packings of Spherical Particles. Industrial and Engineering
[9] Nagle M, Azcárraga J C G, Phupaichitkun S, Chemistry Research Fundamentals, 1984; 23(4): 490-493.

View publication stats