Chou 1997

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

INTERNATIONAL JOURNAL OF ENERGY RESEARCH, VOL.

21, 395—410 (1997)

A METHODOLOGY FOR TUNNEL DRYER CHAMBER DESIGN


S. K. CHOU*, M. N. A. HAWLADER, K. J. CHUA AND C. C. TEO
Department of Mechanical and Production Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 0511

SUMMARY
We present a methodology for estimating the required dimensions of tunnel dryers used in the drying of agricultural
products. The methodology incorporates a material model that describes the drying behaviour of the product based on
the ‘receding front’ concept, and a formulation of the transfer processes in tunnel drying. The material model is first
validated using experimental data before it is integrated into the dryer model. The overall dryer model in finite-difference
form is used to simulate the performance of a tunnel dryer under varying conditions. The resulting simulation algorithm
becomes an engineering tool for designers of tunnel dryers, enabling the estimation of required dimensions and the drying
time to reach the desired moisture content under different drying conditions and dryer performance. ( 1997 by John
Wiley & Sons, Ltd. Int. J. Energy Res., vol. 21, 395—410 (1997).
(No. of Figures: 8 No. of Tables: 0 No. of Refs: 12)
KEY WORDS: tunnel dryer; dryer modelling and simulation

1. INTRODUCTION
Despite numerous studies having been done on the drying of agricultural products, reliable simulation
models to aid the design of tunnel dryers are few. Many theoretical agricultural drying models such as those
formulated by Alvarez and Legues (1986) and Wang et al. (1980) serve to provide a better understanding of
the drying kinetics of foodstuff, while tunnel dryer models such as those presented by Mabrouk and Belghith
(1994) and Bertin et al. (1980) focus on the drying process within the dryer. Traditionally, designers of dryers
rely mainly on past experience of drying similar products to set operating conditions for trials. Having
achieved the desired product specifications using the pilot plant, they use empirical scale-up rules with
appropriate safety margins to estimate the dimensions of full-scale dryers. This results in a workable design
that requires further testing and considerable time and money to optimize. Therefore, there is a need for
a dryer model that the designer can use to obtain the dimensions of a purpose-built dryer and that can further
be used as a tool to assess the performance of a dryer as the process parameters vary.
In this paper, we describe a tunnel dryer by studying the drying time required for the product to reach the
required moisture content given the drying potential of the air. Therefore, it is expected that the total drying
time required to reach the desired moisture content is a function of several drying parameters. These drying
parameters refer to the air inlet velocity, its temperature and its absolute moisture content. The objective of
the study is to derive a simplified but accurate methodology for determining the chamber dimensions of
a tunnel dryer. Furthermore, the methodology can serve as a useful tool for system optimization of the dryer
based on performance criteria such as product quality and savings in energy and operating costs. To
calibrate the model, we use experimental results obtained from potato disks drying.

2. FORMULATION OF DRYING MODEL


The generalized methodology of dryer modelling and design as described by Reay (1989) comprizes both an
equipment model and a material model. The material model describes the drying kinetics and equilibrium,
* Correspondence to: S. K. Chou

CCC 0363—907X/97/050395—16$17.50 Received 25 September 1995


( 1997 by John Wiley & Sons, Ltd. Revised 7 November 1995
396 S. K. CHOU et al.

i.e. factors that are specific to the product being dried. The equipment model describes the condition of the air
flow and the exchange of sensible and latent heat between the air and the product in the adiabatic dryer. The
two models are then integrated to form the overall drying model for the prediction of the required chamber
dimensions. However, in the above method, experimentation is still needed to establish the drying character-
istics of the material.

3. MATERIAL MODEL
The formulation of the material model is based on the concept of the ‘generalized drying curve’ (GDC)
proposed by Ratti and Crapiste (1992). In the GDC model, it has been demonstrated that the ‘generalized
drying parameter’ (GDP) is independent of the drying conditions, particle geometry and food product being
dried and is only a function of the moisture content of the product. This experimentally obtained parameter
provides the means to determine the moisture content of the product during the drying process. In this study,
the material model is simplified to suit our application.
According to the GDC model, the moisture loss by the product may be interpreted by the ‘receding
front’ model. Experimental observations by Ratti and Crapiste (1992) indicate that during the drying
process of apple, carrot and potato, an evaporation zone is conspicuous. This evaporation zone segregates
the product into two regions, namely a wet zone and a dry zone. It is further observed that the front recedes
from the free surface into the product as drying progresses. The ‘receding front’ phenomenon is depicted in
Figure 1.
Accepting the receding front phenomenon, the following profile within the product during the drying
process may be considered.
For the wet zone:
X*"X for 0)z)m (1)
0
For the dry zone:

A B A B
X*!X 1!z
4 " for m(z(1 (2)
X !X 1!m
m 4

Figure 1. Scheme of the receding front model

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
TUNNEL DRYER CHAMBER DESIGN 397

Re-writing equation (2), we obtain

C D
X !X
X*"X # m 4 (1!z) for m(z(1 (3)
4 1!m
For a product being dried, the water mass flux movement may be characterized by the generalized Fick’s
law. For a one-dimensional diffusion process within the internal structure of the product, the water mass flux
can be expressed as
o D LX
g "! 1 % (4)
8 ¸ Lz
0
From the receding front moisture profile given by equations (1), (2) and (3), the water mass flux can be
evaluated by applying equation (4). Hence, in the dry zone, the mass flux can be re-written as

C D
o D X !X
g " 1 % m 4 (5)
8 ¸ 1!m
0
Within the product, there exists a thin layer of saturation vapour at the receding front so that the moisture
content of the product at the receding front can be assumed to be equal to ½ , the saturation humidity of the
4!5
air at the product temperature. Thus, equation (5) can be re-written as
o D
g " 1 % [½ !X ] (6)
8 ¸ (1!m) 4!5 4
0
where the term in the square brackets represents the driving force for the moisture movement in the dry
domain within the product itself. At the same time, the mass flux movement in the drying air is written as
g "k (X !½ ) (7)
8 ' 4 !
To account for the moisture movement within the product due to changes in the product’s chemical
composition as drying proceeds, it is appropriate to define the mass transfer Biot number, Bi , as
.
k ¸
Bi " ' 0 (8)
. D o
% 1
In determining the Biot number, the diffusion coefficient D is first obtained from experimental results. In the
%
subsequent simulation, we calculate the Biot number from equation (8) and the diffusion coefficient is
calculated using results from potato disks drying.
As described above, the GDP, ', which marks the exact co-ordinate of the ‘receding front’, is defined as
'"1!m (9)
Using equations (6), (7), (8) and the GDP given by equation (9), the water mass flux can be expressed as

C D
½ !½
g "k 4!5 ! (10)
8 ' 1#(')Bi
.
The overall moisture content of the product can be estimated from the moisture profile defined by
equations (1), (2) and (3). Integration of both the wet and dry zones gives

Pm C D
m 1
P0
(X !X )
X" X dz# X# m 4 ](1!z) dz (11)
0 4 (1!m)
Solving equation (11), we obtain
(1!X/X )
1!m" 0 (12)
[1!(X #X )/2X ]
m 4 0
( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
398 S. K. CHOU et al.

Ratti and Crapiste (1992) applied an ‘order of magnitude’ analysis to the above equation and showed that
the term (1!m), which is the GDP, is dependent only on the moisture content. This implies that the GDP
does not depend on drying conditions, the type of product or the geometry of the sample. Equation (9) can
therefore be expressed as

A B
X
'+ 1! (13)
X
0
Hence, it is now possible to formulate a new relation between the GDP and the moisture content of the
product that allows one to obtain the change in moisture content in the product with respect to time (cf.
equation (15)).
Using equation (13), Ratti and Crapiste (1992) then proposed the following function for the GDP based on
experimental data using different agricultural food products:

A B
X C2
'"C ] (14)
1 X
0
where the regression coefficients C and C obtained for the expression are 5·32]10~3 and !1·079,
1 2
respectively. Equation (14) allows one to obtain the change in the moisture content from equation (10),
thus

C D
m LX ½ !½
g " 1 "k 4!5 ! (15)
8 A Lt ' 1#(')Bi
1 .
and

C D
LX A k ½ !½
" 1 ' 4!5 ! (16)
Lt m 1#Bi (')
1 .
The derivation of equation (15) is given in the Appendix.
To find an expression for the change in the product temperature ¹ , with respect to a time step *t, the heat
1
and mass balance equation between the drying air and the product is examined. The resulting energy transfer
equation describes what happens to the heat that is transferred from the drying air to the product. This heat
is used to raise the temperature of the product as well as to evaporate moisture from it. Therefore, the energy
balance equation is expressed as

h A (¹ !¹ )*t"m M[C #XC )*¹ ]#[*X(h )]N (17)


' 1 ! 1 1 11 18 1 &'
The first term on the right-hand side of equation (17) represents the change in sensible heat within the
product due to a change in the product temperature. The second term represents the heat of vaporization of
the moisture leaving the product. Re-writing equation (17), we obtain

hA LX L¹
' 1 (¹ !¹ )! (h )"(C #XC ) 1 (18)
m ! 1 Lt &' 11 18 Lt
1
where LX/Lt is expressed by equation (16). Substituting for LX/Lt and re-arranging the terms in the equation,
we obtain the following expression for the change in product temperature with time:

G C DH
L¹ hA k (½ !½ )h
1" ' 1 [¹ !¹ ]! ' 4!5 ! &' (19)
Lt m (C #XC ) ! 1 h (1#Bi ')
1 11 18 ' .
Equations (16) and (19) constitute the material model of the drying process. A comparison between results
obtained from the material model and experimental data on the drying of potato disks is presented in
a subsequent section.

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
TUNNEL DRYER CHAMBER DESIGN 399

4. EQUIPMENT MODEL
To complete the formulation of the dryer model, we seek to characterize the heat and mass balance of the air
and the product in the drying tunnel. The equipment model is formulated for a continuously operated,
parallel-flow adiabatic tunnel dryer. The schematic for the local heat and mass balance for a differential
element is shown in Figure 2. The underlying assumptions used in the development of the equipment model
are as follows: (i) uniform feed rate with uniform moisture content; (ii) uniform product size; (iii) uniform
distribution of the drying product in the tunnel chamber; (iv) adiabatic drying process with no heat loss to
the environment; and (v) wet-bulb temperature coincides with adiabatic saturation temperature.
For an ideal adiabatic drying process in the steady state, Pakowski and Mujumdar (1987) presented
equations (20) and (21) describing the mass and energy balance between the air particles and the product for
a finite control volume:
M (½ !½ )"M (X !X ) (20)
! 1 2 1 2 1
M (h !h )"M (h !h ) (21)
! !1 !2 1 12 11
By including the dimensionless bed porosity parameter e, defined as the fraction of void space in the chamber
to the total volume of the chamber, the mass transfer equation (20) can be written as
» eo (½ !½ )"» (1!e)o (X !X ) (22)
! ! 1 2 1 1 2 1
Next, we consider the drying air entering control volume x#1 after a finite time *t after having undergone
the heat and mass transfer process at control volume x. *t is therefore envisaged as the product residence
time in control volume x before moving to control volume x#1 at a product feed rate » . Thus, the enthalpy
1
balance between the product and the air, described by equation (21), over a moisture differential interval dl
can be re-written as

GC D H
h A *t
» eo (C #½ C )(¹ !¹ )"» (1!e)o C (X !X )# ' 1 (¹ !¹ ) (23)
! ! 1! ! 17 !1 !2 1 1 17 1 2 m ! 1
1
where (C #½ C )(¹ !¹ ) is the change in specific enthalpy of the drying air passing through the
1! ! 17 !1 !2
control volume; [C (X !X )](¹ !¹ ) is the change in specific enthalpy of the product due to a change
17 1 2 ! 1
in moisture content; and [h A *t/m ](¹ !¹ ) is the change in specific enthalpy of the product due to an
' 1 1 ! 1
increase in sensible heat transferred from the air to the product.
The above equipment model together with the material model constitutes a complete modelling tool for
a tunnel dryer. In the next section, we present some empirical equations used to calculate the properties of the
drying air.

Figure 2. Scheme of a differential moisture content element

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
400 S. K. CHOU et al.

5. DRYING AIR PROPERTIES


To enable the above dryer model to be used for simulation, we require information on the thermodynamic
and transport properties of moist air. The information is obtained from empirical relationships described in
the following paragraphs.
The latent heat of vaporization of water is given by Li et al. (1992) in the temperature range 273·25 K to
563·15 K as

A B
647·3!¹ 0>3298
h "2501·05]103 ! (24)
&' 647·3!273·15
The saturation pressure of vapour can be calculated from the well-known Antoine’s equation (Li et al.,
1992) given by

C D
3816·44
p "exp 23·197! (25)
4!5 ¹ !46·13
!
Hence, the saturation moisture content of moist air can be calculated using the following formula:
p
½ "0·622 4!5 (26)
4!5 P!P
4!5
For the estimation of the heat transfer coefficient on the air side h , Sokhansanj and Jayas (1987) proposed
'
the following correlations for air flow over flat food products depending on whether the air flow in the drying
chamber is laminar or turbulent.
For laminar flow (Re(105):
Nu"0·664Re0>5Pr0>33 (27)
For turbulent flow (Re'105):
Nu"2#0·6Re0>5Pr0>33 (28)
Using the Chilton—Colburn analogy, we calculate the mass transfer coefficient k from the heat transfer
'
coefficient h :
'

A B
1
k" h (29)
' C #½C '
1! 17
for an air—water vapour system.

6. OVERALL DRYER MODEL


The overall dryer model is obtained by combining equations (16), (19), (22) and (23). To provide a simulation
structure for the model, we re-write the equations in finite-difference form.
For product moisture content:

G H
A k (½M !½M )
X !X "*t 1 ' 4!5 ! (30)
j j`1 m (1#Bi ')
1 .
For air humidity:
» (1!e)o
½ !½ " 1 1 (X !X ) (31)
j`1 j » eo j j`1
! !
For product temperature:

G H
h A *t k (½M !½M )h
¹ j`1!¹ j " ' 1 (¹M !¹M )! ' 4!5 ! &' (32)
1 1 m (C #XM C ) ! 1 h (1#Bi ')
1 11 18 ' .
( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
TUNNEL DRYER CHAMBER DESIGN 401

For air temperature:

GC D H
» (1!e)o h A *t
¹ j`1!¹ j " 1 1 C (X !X )# ' 1 (¹M !¹M ) (33)
! ! » eo (C #½M C ) 18 j j`1 m ! 1
! ! 1! ! 17 1
Equations (30) to (33) constitute the overall dryer model for calculating the conditions of the product and the
drying air leaving the control volume.

7. SIMULATION METHODOLOGY
The overall dryer model described by equations (30) to (33) is a one-dimensional incremental model. The
drying process in the chamber is discretized into a finite number of quasi-static product loading intervals. For
each interval, the desired decrease in moisture content of the product is specified. The intervals, representing
successive moisture content decrease, are made small enough to enable a local decoupling of the equations
and to improve the stability and precision of the finite-difference method. In each of the moisture intervals,
a corresponding solid residence time *t is calculated. The summation of the *t over all the moisture content
difference intervals provides the time span required to dry the product in the chamber from X (initial
*
moisture content) to X (desired moisture content). Once the time span is known, the required length of the
$
drying chamber can be conveniently determined.
Thus, the desired final moisture content of the product is achieved through a succession of loading
intervals., starting from the initial moisture content of the product. The ‘marching’ progress of the product is
initiated along the length of the chamber starting from the inlet of the tunnel dryer. For each interval, the
heat and mass transfer equations are solved as described below.
(i) Given the initial air condition and the required moisture content difference ½ is calculated from
j`1
equation (31) with known values of e, » , » , o and o .
1 ! 1 !
(ii) Next, ½M , ½M , k and h for the first moisture content difference interval are calculated based on the
! 4!5 ' '
condition of the inlet air to the drying chamber.
(iii) Equation (30) is then solved for the duration of the interval *t. With *t known, and assuming that
¹M "¹ j and ¹M "¹ j for the moisture content difference interval, the solution to the heat transfer
1 1 ! !
equation (32) and energy balance equation (33) results in the product temperature ¹ j`1 and the air
1
temperature ¹ j`1.
!
(iv) ¹M and ¹M are then calculated by taking the arithmetic average of the inlet ( j) and outlet ( j#1) values for
1 !
the same moisture content difference interval. h , k and ½M for the same interval are re-calculated using
' ' 4!5
the new average product and air temperatures. Both steps (iii) and (iv) are then repeated to yield the
corrected value of *t for the interval. The process is continued until the value of *t converges.
(v) The entire algorithm from steps (i) to (iv) is repeated for the next moisture content difference interval. The
computation is halted when the desired moisture content of the product is obtained.
The sum of *t for all moisture content difference intervals is then calculated, thus providing the overall
drying time needed for the product to reach the desired final moisture content or the equilibrium moisture
content, whichever is the higher value. This information together with the product velocity leads us to the
length of the drying chamber.

8. RESULTS AND DISCUSSION

8.1. Validation of material model


Before employing the overall drying model, we validate the material model with actual experimental data.
Potato disks are used as the drying specimen for the validation study. The disks are dried in a heat-pump
dryer, which provides an active apparatus for achieving and controlling the desired drying conditions. The

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
402 S. K. CHOU et al.

weight of potato disks is constantly monitored through a pair of compression load cells. The potato disks are
dried until the equilibrium condition is obtained.
In the material model, a small time step is chosen to ensure the decoupling of the equations and to enhance
the stability and precision of the model. The difference between experimental and predicted results is shown
in Figures 3 to 5. The simulated results obtained from the material model show good agreement with
experimental data. The mean difference in moisture content variation at any time is about 6%. Comparison
between experimental and predicted values for varying time interval *t further demonstrates the robustness
of the material model, as it is found that the time step does not alter the mean difference in moisture content
by more than 0·5% when the time step interval is increased from 10 to 20 seconds.
In estimating the measurement uncertainties arising from the instrumentation, we consider the random
and fixed errors associated with the load cell sensors and data acquisition system. The readings from the load
cell sensors are recorded at 30 second intervals and undergo a continuous averaging process. This helps to
eliminate the resulting random errors, thus reducing the uncertainty associated with the averaged weight to
only the fixed component. The fixed error is found from experimental calibration data. The load cell sensors
are calibrated with the data acquisition system using standard dead-weights under static condition. The
maximum percentage difference between the calibrated value and the actual value represents the fixed error
associated with the sensor-data acquisition system. Thus, the overall uncertainty of the experimental data
shown in Figures 3 to 5 is found to be $4%.

8.2 Application of overall model


In the following discussion, we show results of the application of the overall dryer model to a drying example.
The overall dryer model is further used to investigate the effects of various drying parameters on the overall
process drying time.

Figure 3. Comparison between experimental and predicted drying kinetics (material thickness: 7 mm; air flow velocity: 2·5 m s~1)

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
TUNNEL DRYER CHAMBER DESIGN 403

Figure 4. Comparison between experimental and predicted drying kinetics (material thickness: 7 mm; air flow velocity: 2·5 m s~1)

Figure 5. Comparison between experimental and predicted drying kinetics (air temperature: 323·15 K; air relative humidity: 70% RH;
air flow velocity: 2·5 m s~1)

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
404 S. K. CHOU et al.

Suppose there is a need to dry potato disks from an initial moisture content of 3·35 kg of water per
kilogramme of dry mass to a desired moisture content of 0·50 kg of water per kilogramme of dry mass at
a product feed rate M of 200 kg h~1 (dry basis). The following operating conditions are selected.
1
(i) Inlet temperature of material, ¹ "28°C
1
(ii) Bed porosity (loading), e "0·9
(iii) Initial material moisture content X "3·00 kg water/kg dry mass
i
(iv) Desired moisture content X "1·00 kg water/kg dry mass
$
(v) Velocity of material, » "0·0007 m s~1"2·52 m h~1
1
(vi) Moisture content difference interval, dl "0·00001 kg water/kg dry mass
Figure 6 shows the effects of air flow rate on the overall drying time. It can be seen that the reduction in
cycle drying time as the air flow rate increases is not linear. When operating in the low flow rate region,
increasing the air flow causes a significant reduction in the cycle drying time. On the other hand, in the high
flow rate region (greater than 3·0 m s~1), the improvement in the drying time from using higher flow rates is
insignificant. Beyond a certain air flow rate, the drying time reduces asymptotically towards a minimum
value. To explain this, we consider the movement of the air particles across the surface of the product. As
moisture leaves the surface of the product, it is passed to the air particles immediately adjacent to it.
Therefore, as the flow rate of the air increases, the air surrounding the product is replaced with drier air at
a faster rate. Moisture from the food product is then able to evaporate to the drier air at a correspondingly
higher rate.
In the low flow rate region, the degree of air particle replacement across the product’s surface is low.
Therefore, an increase in the air flow rate improves the air particle replacement rate resulting in a substantial
reduction in drying time. In contrast, in the high flow rate region, the residence time of the air particles across
the product shortens considerably, reducing the thermal communication period between the air particles and
the product. Therefore, any reduction in drying time resulting from higher air flow rates cannot be readily
realized. Similar results were observed by Ratti and Mujumdar (1993) in their simulation model for fixed-bed
batch drying of agricultural products. In the experimental work carried out by Wang and Brennan (1992) for

Figure 6. Influence of air flow velocity on total drying time (simulated data from overall dryer model)

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
TUNNEL DRYER CHAMBER DESIGN 405

studying the drying behaviour of potato, they reported that the air flow rate in the range of 3·30 to 4·50 m s~1
had little effect on the drying behaviour of potato.
In view of the observations and explanation given above, there appears to be an optimum flow rate beyond
which improvement in the reduction of drying time is insignificant. This suggests that a moderate air flow
rate should be chosen to maximize the effect of flow rate on the moisture removal rate. This moderate flow
range optimizes the effect of higher moisture evaporation due to faster air particle replacement rate and
longer thermal communication time between the air particles and the product. It also has implications
relating to capital and energy cost savings.
In the simulation study it is found that, below a flow rate of 2·5 m s~1, the humidity of the drying air
approaches saturation point before the product reaches the desired final moisture content. This can occur in
practice and two possible solutions can be used. One is to change the drying conditions of the air to provide
a higher drying potential. The second is to employ a cascading process whereby the product now enters
another drying chamber in cascade where the drying air inlet conditions, » , ¹ and ½, for the second
! !
chamber are different from or the same as those of the first. The drying process for the second chamber
continues until the desired moisture content is obtained.
Inspection of Figure 7 shows that increasing the drying air temperature reduces the drying time. A higher
drying air temperature increases the temperature gradient between the air and the product. This improves
the driving force for sensible heat transfer between the drying air and the product. Consequently, the
moisture transfer from the product to the air improves due to the greater difference in partial pressure of
water vapour in the product and the drying air. Studies by Sherwood (ASHRAE, 1989) have shown that the
mass diffusivity of water vapour in air increases as the air temperature increases. Hence, higher air
temperatures favour moisture migration from the product surface to the air and within the drying air
particles.
The driving force for mass transfer increases with higher humidity gradient between the drying air and the
product. Drier air has a greater capacity to evaporate water from a free surface. For a fixed air flow rate and

Figure 7. Influence of inlet air temperature on total drying time (simulated data from overall dryer model)

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
406 S. K. CHOU et al.

Figure 8. Influence of inlet air humidity on total drying time (simulated data from overall dryer model)

temperature, the air with the lower humidity ratio is able to remove more moisture from the product. This
can be observed from Figure 8 where the lower the humidity of the inlet air, the faster is the time for the
product to approach desired moisture content.
For each simulated drying process, the drying time is obtained by summing up the *t for all moisture
difference intervals. The drying time represents the required time to remove the specified amount of moisture
from the product based on a given drying potential of the air. Suppose the inlet air conditions comprise
a velocity of 3·0 m s~1, an air humidity of 0·008 kg per kilogramme of dry air and a drying temperature of
72·5°C. From Figure 7, the total time for the product to reach the required moisture content is approximately
143·4 minutes, which is the residence time of the product in the drying chamber. Multiplying the product
velocity with this time, the required length of the chamber is found to be 6·02 m. The corresponding
transverse cross-sectional area, A, of the chamber can be obtained from the following equation:
M
A" 1 (34)
» (1!e)o
1 1
For the above example, the required cross-sectional area of the chamber is found to be 2·25 m2.
In practice, there is a trade-off between faster drying rate and desired product quality. In our present
model, the residence time of the product in the dryer determines the required length of the drying chamber.
An oversized chamber will not only affect the final product quality but also first and operating costs. An
undersized chamber would fail to optimize the potential of the drying air stream for any given set of initial
conditions.

9. CONCLUSION
A design methodology for estimating the dimensions of the tunnel drying chamber has been developed. The
methodology incorporates a dryer model which depicts the kinetics of drying and the transfer processes in

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
TUNNEL DRYER CHAMBER DESIGN 407

a tunnel dryer. Experimental data are used to validate the drying kinetics ‘material’ model. Predicted results
from the model agree well with experimental data.
The dimensions of the dryer are obtained from simulation using the overall dryer model in finite-difference
form. The simulation model is also used to analyse the influence of various drying parameters on the drying
time. It has been shown that a shorter drying time can be achieved with a higher air temperature and lower
air humidity. The simulated results on the effect of air flow rate on drying time show the existence of an
optimum flow rate which maximizes moisture removal rate for a given set of drying conditions.
Finally, it has been demonstrated that a reliable engineering tool is necessary in order that the practical
difficulties of designing an optimum dryer can be overcome. There is scope for further work in cascade
drying, enhancing the material and equipment modelling and optimization for energy efficiency and product
quality.

APPENDIX: DERIVATION OF EQUATION (15)


Wet zone:
X*"X for 0)z)m (A1)
0
Dry zone

A B A B
X*!X 1!z
4 " for m)z)1 (A2)
X !X 1!m
m 4
Re-writing equation (A2), we obtain

A B
X !X
X*"X # m 4 (1!z) (A3)
4 1!m
Differentiating equation (A3), we have

A B
LX* X !X
"! m 4 (A4)
Lz 1!m
Diffusion within the product is defined by

A B
o D LX*
g "! 1 % (A5)
8 ¸ Lz
0
Therefore, in the drying zone the mass flux is written as

A BA B
o D X !X
1 %
g " m 4 (A6)
¸8 1!m
0
Taking note that X "½ , equation (A6) can be re-written as
m 4!5

A BA B
o D ½ !X
g " 1 % 4!5 4 (A7)
8 ¸ 1!m
0
At the same time, the mass flux from the surface to the drying air is given by
g "k (X !½ ) (A8)
8 ' 4 !
Substituting for X in equation (A7), we obtain
4

C AB D
o D g
g " 1 % ½ ! 8 #½ (A9)
8 ¸ (1!m) 4!5 k !
0 '
( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
408 S. K. CHOU et al.

Re-arranging equation (A9) leads to

C D
g k ¸ (1!m)
8 ' 0 #1 "[½ !½ ] (A10)
k o D 4!5 !
' 1 %
From the definition of the Biot number, given as

A B
k ¸
Bi " ' 0 (A11)
. D o
% 1
and the definition of GDP, written as
'"1!m (A12)
equation (A10) is finally written as
g
8 [1#Bi (')]"[½ !½ ] (A13)
k . 4!5 !
'
The final form of equation (15) is then obtained as
[½ !½ ]k
g " 4!5 ! ' (A14)
8 [1#Bi(')]

ACKNOWLEDGEMENTS
The authors wish to thank the Canadian International Development Agency for the financial support of this
project under the ASEAN-Canada Energy Project.

NOMENCLATURE
A "cross-section area of dryer (m2)
A "material mass transfer area (m2)
1
Bi "mass transfer Biot number (dimensionless)
.
C "specific heat (kJ kg~1 K~1)
1
D "effective diffusion coefficient (m2 s~1)
%
dl "differential moisture content interval, kg water/kg dry mass
h "enthalpy of drying air (J kg~1)
!
h "heat transfer coefficient (W m~2K~1)
'
h "heat of evaporation (J kg~1K~1)
&'
h "enthalpy of product (J kg~1)
1
k "mass transfer coefficient (kg m~2s~1)
'
M "mass flow rate of the drying air (kg s~1)
!
M "product feed rate (kg s~1)
1
m "dry mass of product (kg)
1
m "wet mass of product (kg)
8
¸ "characteristic length of product (m)
0
Nu "Nusselt number ("h ¸/i) (dimensionless)
#
P "total pressure of drying air (Pa)
Pr "Prandtl number ("kC /i) (dimensionless)
1!
Re "Reynold number ("» ¸ /k) (dimensionless)
! 1!
¹ "temperature (K)
¹M "average temperature of time interval *t (K)

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
TUNNEL DRYER CHAMBER DESIGN 409

t "time (s)
» "velocity (m s~1)
x "co-ordinate axis (m)
X* "localized moisture content, kg water/kg dry mass
X "moisture content of product, kg water/kg dry mass
XM "average moisture content of product, kg water/kg dry mass
X "initial moisture content of product, kg water/kg dry mass
*
X "desired moisture content of product, kg water/kg dry mass
$
½ "humidity of air, kg water/kg dry air
½M "average humidity of time interval *t, kg water/kg dry air
z "normalized co-ordinate axis ("x/¸ )
0
Greek symbols

m "receding front position (dimensionless)


g "water mass flux (kg s~1 m~2)
8
' "generalized drying parameter (dimensionless)
e "bed porosity (dimensionless)
b "psychrometric ratio (dimensionless)
o "density (kg m~3)
i "thermal conductivity of air (J s~1 mK~1)
k "dynamic viscosity of air (kg s~1 m~1)

Subscripts

1 "inlet condition of differential moisture content interval


2 "inlet condition of differential moisture content interval
a "air
0 "initial
p "product
s "product surface
sat "saturation
v "vapour
w "water
m "receding front co-ordinate
j "beginning of moisture content difference interval
j#1 "end of moisture content difference interval

REFERENCES
Alvarez, P. I. and Legues, P. (1986). ‘A semi-theoretical model for the drying of Thompson seedless grapes’, Drying ¹echnology, 4(1),
1—17.
ASHRAE (1989). ASHRAE Handbook of Fundamentals 1989, American Society of Heating, Refrigerating and Air-Conditioning
Engineers, Atlanta, Georgia, U.S.A., pp. 5.2.
Bertin, R., Pierronne, F. and Combarnous, M. (1980). ‘Modelling and simulating a distributed parameter tunnel drier’, J. Food Sci. 45,
122—125.
Li, S. Q., Gong, Z. X. and Mujumdar, A. S. (1992). ‘Calculations of the state variables of moist air’, in Drying ’92, Part B, A. S. Mujumdar
(Ed.), Elsevier, pp. 1790—1797.
Mabrouk, S. B. and Belghith, A. (1994) ‘Simulation and design of a tunnel drier’, in Renewable Energy, Vol. 5, Part I, pp. 469—473.
Pakowski, Z. and Mujumdar, A. S. (1987). ‘Basic process calculations in drying’, in Handbook of Industrial Drying, A. S. Mujumdar (Ed.),
Marcel Dekker, New York, U.S.A., pp. 83—129.
Ratti, C. and Crapiste, G. H. (1992). ‘A generalized drying curve for shrinking food material’, in Drying ’92, Part A, A. S. Mujumdar (Ed.),
Elsevier, pp. 864—873.
Ratti, C. and Mujumdar, A. S. (1993). ‘Fixed-bed batch drying of shrinking particles with time varying drying air conditions’, Drying
¹echnology, 11(60), 1131—1135.

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)
410 S. K. CHOU et al.

Reay, D. (1989). ‘A scientific approach to the design of continuous flow dryers for particular solids’, Multiphase Sci. & ¹echnol.
4, 1—102.
Sokhansanj, S. and Jayas, D. S. (1987). ‘Drying of foodstuffs’, Handbook of Industrial Drying, A. S. Mujumdar (Ed.), Marcel Dekker, New
York, U.S.A., pp. 517—554.
Wang, C. Y., Singh, R. P. and Zuritz, C. (1980). ‘A numerical approach to simulate rice drying’, in Drying ’80, A. S. Mujumdar (Ed.),
pp. 227—232, Vol. 1.
Wang, N. and Brennan, J. G. (1992). ‘Effect of water binding on the drying behaviour of potato’, Drying ’92, Part B, A. S. Mujumdar
(Ed.), Elsevier, pp. 1350—1359.

( 1997 by John Wiley & Sons, Ltd. Int. J. Energy Res., 21, 395—410 (1997)

You might also like