air flow patterns

8/18/2019 Air Flow Patterns

1/9

I C h e i T l E 0263-8762/04/S30.00 0.00

2004 Institution of Chemical Engineers

Trans

IChemE, Part A, October

2004

Chemical Engineering Research and Design, 82 A10): 1344—1352

IR FLOW

P TTERNS

IN DEHUMIDIFIER

WOOD DRYING KILNS

Z .

F. SUN

1

, C. G.

C A R R I N G T O N 1

J . A.

A N D E R S O N 2

and Q. SUN

1

1Physics

Department, University of Otago, Dunedin, New Zealand

^Energy

Group Ltd, Dunedin, New Zealand

B

y simulating the airflow

patterns,

velocity and

pressure

distributions in an industrial

dehumidifier wood drying

k i l n ,

it is shown that typical dehumidifier system configur

ations create a risk of high levels of air recirculation at the dehumidifier, wi th adverse

implications for dryer capacity and efficiency. The simulation results also show that, for high

efficiency,

it is important to avoid air recirculation. An alternative air flow configuration,

which

could

achieve

this result using a single set of

fans,

is

presented

and its

performance

assessed.

Keywords: wood drying;

heat

transfer; mass transfer; mathematical modelling;

heat

pump;

dehumidifier.

INTRODUCTION

Air f low

design is potentially important in the design of

dehumidifier drying kilns which

operate

as closed, fu l ly -

recirculatory

systems.

In this

paper,

we show how the

performance

of a dehumidifier wood drying ki ln can be

impaired by a mismatch between the ki ln airflow system

and the dehumidifier. In turn, the poor

performance

of

the dehumidifier reduces the overall efficiency of the

ki ln ,

resulting in

increased

drying time and energy use.

In industrial wood drying kilns, the effect of non-uniform

airflows

is particularly diff icul t to resolve. Nijdam and

Keey 2002)

have

investigated airflow

patterns

in conven

tional

heat-and-vent

timber kilns to determine design modi

fications that promote more uniform flows. Their velocity

measurements

down the height of the timber stack in a

k i l n wi th outward-swing overhead baffles showed that the

uppermost packets

of the

stack were starved

of airflow.

Nijdam and Keey 2002) demonstrated that, using con

toured right-angled

bends

and inward-swinging

overhead

baffles in higher-velocity wood drying kilns, there was a

3-fold

reduction in the

range

of the vertical velocity

distribution.

In a typical industrial dehumidifier

k i l n ,

the dehumidifier

comprises separate

modules, wi th their own fans, indepen

dent of other fans which may be used to maintain air move

ment in the

k i l n .

In this system the k i l n airflow and the

dehumidifier airflow interact closely, although they are nor

mally not designed as an integrated system. The impact of

mismatches

is diff icul t to anticipate. It is

also

diff icul t to

Correspondence to: Dr Z.F. Sun, Physics Department, University of

Otago, PO Box 56, Dunedin, New Zealand.

E-mail:

[email protected]

measure the distributions of air velocity and

pressure,

due

to the complex interactions of the

subsystems, such

as the

k i l n fans, wood stack, air ducts and the dehumidifier fans

and heat

exchangers.

In this

paper

we

present

an

analysis

of airflow in a commercial dehumidifier wood drying k i l n

as a whole, using a three-dimensional CFD model. We

show

how airflow

design

has the potential to affect the

per

formance of dehumidifier wood drying kilns in unexpected

ways. A modified k i l n configuration, in which an air duct

connects

the exit of the dehumidifier fans wi th the upper

duct space of the

k i l n ,

has

been assessed

using the CFD

model. The

results show

that this ki ln configuration can

significantly

improve the dehumidifier wood drying ki ln

performance. In particular, for dehumidifier wood drying

kilns to achieve high efficiency, it is

advantageous

to use

a single set of

fans

to drive the air flow and to

ensure

all

the flow

passes

the dehumidifiers without recirculation.

SYSTEM DESCRIPTION

Figure 1

shows

the

flow

configuration and the coordinate

system based on a commercial wood drying

k i l n .

The origin

of the coordinate

system

is at the mid-point of the left-hand

w a l l Figure la) of the k i l n at floor level. The system con

sists of two dehumidifier modules installed side-by-side, a

wood stack and six k i l n air recirculation fans, all located

wi th i n an insulated k i l n

chamber.

The two main

elements

of the dehumidifier, which

influ

ence

the flow

patterns

in the

k i l n ,

are the

condenser

and

evaporator

and their air

fans,

shown in Figure 1 a).

Each

module has two

condenser

fans and three evaporator fans.

For simplicity, in this investigation only the characteristics

of the volume flow rate and pressure drops of the

condenser

1344


2/9

A IR

FLOW PATTERNS

1345

5.3 m

4.24 m

kiln fans.

r .

2.2 m-

stack

3.B7m

0.8 m

6.8 m

(a)

U

1.9m

A

air duct

4.3 m

condenser

coils

and fans

evaporator coils

and

fans

0.54 m 1 0 8 m 1.08 m

0.63

m

4.51 m

baffle-board

central 2.75 m 0.5 m

symmetry line

A - A

(b)

B

(c)

Figure

1.

Schematic

diagram of an industrial dehumidifier wood-drying

k i ln .

coils

and fans are considered, since the airflow of the evap

orator enters the

condenser

and the flow rate is typically

only 10% of the

condenser

air f low. Each dehumidifier

module also has an electric air heater to preheat the

k i ln

before drying commences. Here the flow

resistance

of the

heater,

which is located between the condenser fans and

condenser

coils, has

been

incorporated

wi th

the

condenser

coils.

As

shown in Figure 1(a), the cross section of the

k i ln

chamber along the x-direction has a trapezoidal shape,

wi th

the front higher than the rear by 1 m. The backward-

sloping roof would produce stronger vortices in the front

top corner and thus a larger

pressure

drop than the peak-

top and barrel-top kilns discussed by Nijdam and Keey

(2002).

In Figure 1(a), the dashed lines represent the

walls

of a simple air duct which connects the top ceiling

space wi th the exit of the dehumidifier in a modified con

figuration

of

the k i ln .

Stacks of wood are normally several packets deep in the

air

flow

direction and several packets high. As shown in

Figure 1(a),

three

in-line

packets,

being 2.1 m

deep

each

in the air

flow

direction, are considered. The three packets

are supported by bearers 100 mm thick. The gaps

separated

the packets in the airflow direction are 0.25 m wide and

the gaps

separated

the packets vertically are neglected.

Concrete

slabs,

150 mm thick, are placed on the top of

the timber stack to reduce warp during drying (Nijdam

and Keey, 2002). It is

assumed

that the k i l n is fully

loaded

wi th

60 layers of wood boards and thus the horizon

tal ceiling of the drying zone is just on the top of the

concrete slabs.

The wood stack is assumed to be a normally aligned

stack of wood

boards

(Sun and Carrington, 1999), wi th

the geometry shown in detail in Figure 1. It is

assumed

that the side baffles

which

prevent air bypassing the side

of

the timber stack, as shown in Figure 1(b), are fully effec

tive, and air bypass is neglected. The details of the horizon

tal

board layers are shown in Figure 1(c). The

space

between board layers is 20 mm, which is maintained by

longitudinal stickers (or fillets across the width of the

stack at intervals of 450-600 mm (Keey et ai, 2000).

Since the

aspect

ratio of the ducts formed by the board

layers and stickers is relatively large, more than 20, the

effects of the stickers inside the stack on flow patterns

have been neglected. Stickers at the side end of the stack,

however, prevent air flow from the stack to the side

bypass

space.

As demonstrated by previous authors,

the effect of small gaps between in-line boards on the press

ure drop of

airflow

across wood stacks (Langrish and Keey,

1996) and on external mass and

heat

transfer rates (Sun,

2001) can be neglected. Thus, small gaps between in-line

boards have been ignored in the simulation. The ratio of

the inlet plenum-space w id th to the sum of the heights of

the sticker spaces is equal to 0.67 (=0.8/1 .22 m), which

is smaller than the minimum value

(unity)

which was

suggested

by

Nijdam

and Keey (2002) to mitigate the

adverse

effect of the frictional

pressure

drop down the

height of the plenum chamber.

NUMERICAL METHOD

In

order to characterize the distributions of air velocity

and pressure and air recirculation occurring in the k i ln ,

the renormalization group (RNG)

k-e

turbulence model

has been used to solve the turbulent momentum transport

equations. The RNG

k-e

model employs a differential

form of the relation for the effective viscosity,

yielding

an accurate description of how the effective turbulent trans

port varies wi th the effective Reynolds number. This allows

accurate extension of the model to near-wall flows and low-

Reynolds-number or transitional flows (Fluent Incorpor

ated, 1997). The RNG k-e model can be expressed by

the fo l lowing momentum equations and the equations for

turbulent kinetic energy

k)

and its

rate

of dissipation (e)

(Fluent Incorporated, 1997):

9

, , 3 3

(P i) + X - pUiUj)

OX;

/

Peir

3 duj BP

9bc

(1)

d,

„ 9 , ,% 9 /

dk\

p k )

p u t ) = -

U M e f f

- J

+

n , S - p e

and

(2)

3

, N 3

^-(pe) + —(pw,e)

at ax,

3

/ 9e\ , e2

eMeffT- +CUTH,S -C

2

eP-T - R (3)

3.v,

dxi

In

the simulation, the standard RNG k-e model constants,

derived analytically by RNG theory,

have been used:

CM =

0.0845, C

l 6

= 1.42, C

2 e

= 1.68, a

0

= 1.0. The standard

Trans

I C h emE

Part A,

Chemical Engineering Research and Design, 2004,

82(A10): 1344-1352


3/9


4/9

A IR FLOW

PATTERNS

1347

300

250

200

0_

150-

£ 100

Q_

50

0-

Woods Air

Movement 2101

Fantech

0714/10

Extension of

fitted relation

2

4 6 8

Volume

flow

rate m3 s~ 1

igure

3

Fan curves of the Woods air movement k i l n fans and the Fantech

0714/10/25 condenser fans.

fo r

the

pressure

rise across each of the

k i l n

fans, and

= 203.029 + 28.7837v - 2.6677v2, 9.1 < v < 15.7


5/9

1348 SU N et al.

Figure 5 shows the velocity field in a vertical y-z plane

coinciding

wi th the

axes

of the condenser fans of the dehu-

midifiers. In this figure, the two horizontally oriented grid-

surfaces represent the condenser fans. It can be

seen

that air

flow eddies are produced at the top side-corners of the k i ln

and

in the downstream

spaces

between the condenser fans.

More

generally, the CFD results indicate that air flow

eddies are produced at all the top and bottom corners of

the

k i ln

and at the corners of the timber packets. The

eddies represent dissipative processes that reduce the effi

ciency

of

both

the

k i l n

fans and the

dehumidifier

fans.

The calculated mass flow

rates

of the two condenser fans

are the same, at 4.89 kg s _ .

Figures 6 and 7 show profiles of the

x-velocity

along

lines from

the stack inlet to stack outlet on two typical

hori

zontal

planes, at the top ( z = 3.71m) and bottom

(z = 0.11 m) of the stack.

I t is

seen from

Figure 6 that the

velocity profiles

on the

top plane z 3.71 m) are not uniform at the entrance

region

of the

first

packet. The x-velocity

spans from

0.5

to 1.5 m s~ approximately. This

variation

in the

velocity

appears to be caused by non-uniform flow in the ceiling

space

of the

k i l n illustrated

in Figure 4. The large range

in the x-velocity indicates that there are large recircula

t ion

eddies in the top entrance

region

of the

first

packet.

Th e flow patterns of the

airflow

on the top plane of the

second packet are similar to those of the first packet, but

the velocities are larger. At the entrance of the

third

packet, the x-velocity span is smaller than those at the

. s . m m

- - > i

i \

• A)

•

• 4 1

1

Figure 5 Vector of

velocity

magnitude on the vertical plane where

x = 8.725 m.

3.50e+00 -|

3.00e+00 -

2.50e+00 -

2.00e+00-

E,

1.50e+00-

?

o

o

1.00e+00-

>

5.00e-01 -

0.00e+00-

-5.00e-01 -

-1.00e+00-

0 1 2 3 4 5 6 7 8

x-coordinate (m)

Figure 6 .r-velocity profiles on the plane z = 3.71 m.

entrances of the first and second packets. However, the

large span in

x-velocity

in the

exit

region

of the

third

packet indicates the presence of recirculation eddies at

the exit of the stack. These appear to be caused by the

k i ln

fans installed in the ceiling space of the

k i l n

and by

the strong resistance of the airflow impelled by the conden

ser fans

illustrated

in Figure 5. Since the effects of the

longitudinal

stickers inside the stack have been neglected

in the simulation, the air velocity spans in the cross y-

direction

may be overestimated. However, this would not

produce serious errors in the average mass flow rates.

The

x-velocities averaged over the cross-sectional

area

of

the top airflow channel at the locations of 1 m from the

leading edge

of each of the packets are 0.94, 1.93, and

2.39 m s~ in the

first,

second and

third

packets,

respectively.

A s shown in Figure 7, unlike the air

velocity

profiles on

the top plane, which increase

from

the

first

packet to the

third packet, the air

velocity

on the bottom plane decreases.

The

average values of the x-velocity in the

bottom airflow

channel are 1.92, 1.59 and 1.37 ms~ in the

first,

second

and third packets, respectively.

2.50e+00

2.25e+00

2.00e+00

„

1.756+00

| 1.50e+00

£ • 1.258+00

o

g 1.00e+00

7.50e-01

5.00e-01

2.50e-01

0.00e+00

. y

= 0

4

- y

= 0.45

• y = 0.9

o

y=1.35

•

o y = 1.8

1

= y = 2.25

a y = 2.7

1 2

3

4 5 6

x-coordinate (m)

Figure 7

.v-velocity

profiles

on the plane z = 0.11 m.

Trans I C h emE

Part A,

Chemical Engineering Research and Design,

2004, 82(A10): 1344-1352


6/9

A IR FLOW

PATTERNS

1349

Th e velocity

profiles

for the

middle

plane of the stack

z

= 1.91 m), which are not shown here, are similar to

the

velocity

profiles at z = 3.71 m and z = 0.11 m shown

in Figures 6 and 7. However, the profiles of x-velocity on

the middle plane of the stack are more uniform than those

on the top plane (z = 3.71 m) and on the

bottom

plane

z

= 0.11 m). The average values of the x-velocity in the

middle

airflow channel are 1.96, 1.88 and 1.89 m s 1 in

the first, second and th i rd packets respectively.

Profiles of the x- and y-velocity along lines from the

stack inlet to stack outlet indicate that, in the

vertical

z-

direction, the velocity

profiles

are also not

uniform.

Figure 8 shows the vertical profiles of the x-velocities aver

aged over the cross-sectional areas of the corresponding

flow

channels at the locations of 1 m from the leading

edge of each of the three packets (solid lines, test 1). It is

seen that, in the first packet,

w i t h

increasing in the height

o f

the stack, the air velocity increases from 1.92 m

s>~

at

the bottom of the stack to 2.20 m s

_

at z = 0.53 m, and

then

decreases to 0.94 m s 1 at the top of the stack. The

vertical

profile of the air

velocity

in the second packet is

more

uniform than that in the first stack, and the velocity

in

the top region is much larger than that in the first

packet. The

vertical

profile of the air

velocity

in the

third

packet shows that the air velocity in this packet gradually

increases w i t h the height of the stack

from

1.37 m s~ at

the

bottom

of the packet to 2.39 njs. at the top of the

packet. The increase in the air velocity in the th i rd packet

may be due to the effect of the k i l n fans, since the upper

part of the th i rd packet is close to the k i l n fans. It appears

that the empty spaces between successive packets have the

effect of

redistributing

the airflow rate through the packets.

The

variations and fluctuations of the

vertical

profiles of the

ai r velocity in the first and second packets are consistent

wi th the measurements by Nijdam and Keey (2002) in a

traditional

peak-top

k i l n

and in a newer barrel-top

k i ln .

Nijdam

and Keey (2002)

found

that the upper part of the

stacks were normally starved of

airflow

at the air entry

end, due to a

recirculation

zone adjacent to the stack

wi th i n the inlet plenum chamber.

A i r

flow recirculation

between the outlet of the dehumi

difier

and its inlet is also illustrated in Figure 9, which

shows pathlines of

massless

particles discharged

from

the

condenser fans. Here, the three vertically oriented gr id-

surfaces represent the

k i l n

fans, the two horizontally

1 2 3

Height

from

base of timber stack (z-direction) (m)

igur 8 Ai r velocity profile along height of the stack.

igur 9 Illustrative pathlines of particles leaving condenser fans. Par

ticles are coded by the grey colour.

oriented grid-surfaces denote the condenser fans, and the

grid-box represents

the

dehumidifier

walls.

This

figure

shows that a significant fraction of air leaving the dehumi

difier

mixes w i t h the

airflow

out of the stack and re-enters

the

dehumidifier,

because the k i l n fans and the condenser

fans are not configured in an integrated way. Indeed, part

o f the airflow recycles several times from the outlet of

the dehumidifier to its inlet.

Overall calculated results for the k i l n

w i t h

the present

configuration (Figure 1) are listed in Table 1 along w i t h

results for a modified configuration discussed below. In

Table

1, ws is the x-velocity averaged over the whole

cross-sectional area of the stack at the locations inside

each of the three packets, 1 m from the inlet of each

packet. It is seen that, for this

k i ln

(test 1 in Table 1), the

total mass

flow rate (19.55 kg s~ ) delivered by the conden

ser fans is larger than that (15.70 kg s_ ) delivered by the

k i l n

fans. The difference (3.85 kg s~ ) between the

mass

flow

rates delivered by the condenser fans and k i l n fans

represents the minimum amount of

airflow

recirculation

from

the

exit

of the dehumidifier to its

inlet.

This would

happen i f all the

airflow

leaving the stack were to enter

the dehumidifier. For the present

k i l n ,

the minimum

amount

of the airflow

recirculation

is 19.7% of the total

ai r mass

flow rate delivered by the condenser fans. How

ever, the

simulation

indicates that the real

recirculation

mass flow rate is l ike ly to be much larger than this

m i n i

mum

value, since much of the

airflow

delivered by the

k i l n fans comes from the stack directly.

Figures 10 and

11

show the pathlines

o f massless

particles

which are discharged from the upper and

lower

rear-end

surfaces of the stack respectively. In Figures 10 and 11,

the vertical rectangular gray surfaces represent the rear-

end

surfaces of the upper 21 board layers and lower 39

board layers of the stack, respectively. It can be

seen from

Trans I C h emE

Part A, Chemical Engineering Research and Design 2004, 82 A10): 1344-1352


7/9

1350 SUN

et

al.

Table

I Calculated results for different k i l n configurations.

Number of Number dehumidifier

AP

wk fan

Test

k i l n

fans modules Duct

(Pa)

k g s

')

k g s

')

(kgs

- 1

)

( ms

_ 1

)

1

6

2 4

condenser fans) No 31 34

15.70 19.55 15.70 <

1.91

2

0

2 4

condenser fans)

Yes

38.58

18 26

18 26

2.23

Figure

10 that all the air flow discharged

from

the f low

channels of the upper 21 board layers of the stack,

wi th

some local recirculation

in

the space above the dehumidifier,

enters the

k i l n

fans

directly.

Figure 11 shows that, although

most of the air flow discharged from the lower 39 board

layers enters the condenser and then passes

across

the dehu

midifier fans, part of the air flow discharged from the

lower

part of the staGk

goes

to the

k i l n

fans

directly.

Th e mass flow rate, 6.15 kg s~', of the air discharged

from

the upper 21 board layers of the stack is 39.2 of

the

total

flow rate of the

k i l n

fans. Hence

less

than 60.8

o f

the total flow rate of the

k i l n

fans comes from the

dehumidifier. Thus, using the data shown in Table 1 for

test 1, the mass flow rate entering the k i l n fans from the

dehumidifier is

less

than 9.55 kg s~' and the

mass

f low

rate of air

recirculation

from the exit of the dehumidifiers

to

its inlet is more than 10.0 kg s_ 1 , representing 51.2 of

the

total

flow rate of the condenser fans. This

airflow

recir

culation would raise the temperature and reduce the humid

it y

of air at the inlet of the dehumidifiers, which would

reduce the efficiency of the dehumidifiers. In

addition,

energy used by the condenser fans to maintain this large

amount of recirculation f low, approximately 51.2 of the

total energy used by the condenser fans, does not make

any

contribution

to the performance of the

k i l n .

MODIFIED CONFIGUR TION

To

illustrate the effect of changing the dehumidifier kilns,

a modified configuration, test 2 in Table 1. was investigated

using the CFD turbulence model. In test 2, air recirculation

from

the

exit

to the

inlet

of the dehumidifier was e l im i -

nated, using an air duct connecting the dehumidifier to

the top ceiling space of the

k i l n ,

shown by the dashed

lines

in Figure 1(a).

The vertical profiles of the average x-velocity in test 2

are shown by the dashed lines in Figure 8. The air velocities

in the three packets of the k i l n in test 2 are significantly

larger than those in test 1. In addition, the velocity distri-

butions

from

the

first

packet to the

th i rd

packet are more

uniform than those in test 1 and the air velocity profiles

in

the

vertical

z-direction in test 2 are

similar

to those in

test 1. This indicates that, by using appropriate ducting

and

only the condenser fans, the drying performance of

the

k i l n would

be improved.

Overall

calculated results for test 2 are also

listed

in

Table 1. It is seen that, without using

k i l n

fans but wi th

an air duct, the calculated average

x-velocity

(2.23 m s

_

)

in test 2 is larger than that (1.91 m s- 1 ) in test 1. Since

no

k i l n

fans are used, the energy consumption of the

k i l n

fans would be eliminated in test 2. As shown in Table 1,

compared w i th test 1 the

overall mass

flow rate of the

Figure 10

Pathlines of the particles discharged

f rom

the rear-end surface

o f the upper

21

board layers of the stack. Particles are coded by the grey

colour.

Figure

11

Pathlines of the particles discharged

f rom

the rear-end surface

o f the lower

39

board layers of the stack. Particles are coded by the grey

colour.

Trans I C h emE

Part A, Chemical Engineering Research and Design 2004, 82 A10): 1344-1352


8/9

A IR FLOW PATTERNS

1351

condenser

fans in

test

2 has decreased by 1.29 kg s~ and

the stack pressure drop in

test

2 has increased by 7.4 Pa.

The volume flow

rate

of

each condenser

fan has

decreased

from 4.07 to 3.80 m3 s~ , and the power of the

condenser

fans

should

increase

by approximately 1 . On the other

hand, the adverse implications of the air recirculation on

the performance of the dehumidifier, which are

discussed

below, do not occur in test 2.

INFLUENCE OF RECIRCUL TION ON DRYER

PERFORM NCE

The CFD turbulence model cannot be

used

to simulate

the thermodynamic cycle and mass and heat transfer pro

cesses in the dehumidifier or in the wood boards. Hence,

to

assess

the influence of

airflow

recirculation at the dehu

midifiers on the performance of the

k i ln

system, a dynamic

dehumidifier wood drying model developed previously

(Sun et al. 2000) has been used. The model solves the inte

gral form

of the

unsteady

state mass, momentum and

energy balance equations

for both the air flow and the wood

boards.

The distributions of the

average mass

fractions,

temperature, pressure and velocity of the air

stream,

as

wel l as the average moisture content of the wood boards

and their temperature, can be estimated using this model.

For illustrative purposes, the timber is Pinus

radiata

sap-

wood

wi th an

i n i t i a l

moisture content 140 . The final

moisture content is 13 , and the volume of timber is

70 m3 . There are two dehumidifier modules in the k i ln ,

and the external heat and

mass

transfer

rates

between air

flow

and the surfaces of wood boards were calculated

using correlations established on the basis of modified

boundary layer

theories

which

take account

of the

separ

ation and reattachment flows (Sun, 2002). The maximum

condensing temperature, maximum evaporating

tempera

ture, and minimum

stack

inlet relative humidity are

l imi ted

to be 75 °C , 25=C and 40 respectively (Carrington et al

1995). An air

preheater

of 30 kW is

used,

which is turned

of f when the stack inlet dry-bulb temperature reaches

50°C . The

stack

inlet air velocity is 2 m s ' , based on the

CF D results as shown in Table 1 for the

present

k i l n (test 1).

The calculated overall dehumidifier

energy

use,

k i ln

fan

energy, heater energy, and drying time are shown in

Figures 12 and 13, respectively. It is

seen from these

figures

1.6

1.5

f 1.4

Z

g

* 1.3

>,

at

CD

c 1 2

1 1


9/9

1352

SUN et al.

22

141 , , 1

14 16 18 20

Model Power (kW)

Figure

15. Comparison of

measured

dehumidifier power consumption

wi th

a model benchmark under conditions

w i t h

and witho ut recirculation.

the temperature and humidity when the measurements

were made. This difficulty is avoided by comparing the

mea

sured data

w i th

dehumidifier model predictions, the model

assuming no recirculation, to provide a common perform

ance

reference for the

measured data.

Accurate

agreement

wi th

the model is not necessary for this comparison. In the

results, shown in Figures 14 and 15, the

measured

drying

rate is consistently below the model prediction when there

is no baffle, but close to the model results when the baffle

is present. In addition the data indicates that the power

use is higher when the baffle is not present.

CONCLUSION

A i r flow

patterns in an industrial dehumidifier wood

drying k i ln

have been

investigated using a CFD model. In

order to solve the computational

difficulties

for simulation

of

a practical k i ln a simplified procedure has been devel

oped in which the spaces between the board layers are trea

ted as an isotropic porous medium. A momentum source

term describing the pressure drop w i th in the porous

medium

has been added to the standard

fluid flow

equations.

The results obtained show that, without suitable air duct

ing at the dehumidifier air discharge, a large fraction of the

dehumidifier

airflow recycles back to the inlet of the dehu

midifiers. Using a dehumidifier wood drying k i l n model, it

has been demonstrated that such air recirculation

reduces

the efficiency of the dehumidifiers and increases drying

time,

by 18 and 14%, respectively, for the example

presented.

A modified

k i ln

configuration, in which an air duct con

nects the dehumidifier

w i th

the upper

airflow

channel of the

k i l n has been analysed. The results show that this configur

ation

significantly improves the dehumidifier performance.

I t is concluded that, for dehumidifier wood drying kilns to

achieve high efficiency, it is important to (a) ensure the

airflow is properly ducted to prevent recirculation and

(b) avoid using two

sets

of air fans in series.

NOMENCL TURE

k turbulent kine tic energy (m s~~)

M

mass flow

rate (kg s )

P

pressure (N trT

-

)

A f s pressure drop of air

flow

passing through stack (N m~

2

)

S/

momentum source term for porous media (N m ~ )

r time (s)

it veloci ty component or velocity in jr-direction (m s ')

s average x-velocity in stack (m s~ )

v velocity (m s

_ l

)

x coordinate (m)

coordinate (m)

z coordinate (m)

Greek symbols

ak inverse effective Prandtl number for turbulent kinetic

energy

ct€ inverse effective Prandtl number for turbulent dissipa

t ion rate

e turbulent dissipation rate (m

2

s

- 3

)

iu eff effective viscosity p., + p.) (kg m

-

' s

- 1

)

fj. y turbulent viscosity (kg m _ 1 s_ l )

p

density (kg m~

3

)

Subscripts

c,fan condenser fans

i /-direction

j /-direction

k k i l n

k,fan

k i l n

fans

REFERENCES

Carrington, C.G., Bannister, P. and Li u, Q., Performance analysis of a

dehumidifier using HFC-134a, Int J Refrig, 18: 477-485.

Fantech Pty Ltd, 1993.

Fans

by

Fantech,

1st edn (The Craftsman Press,

Australia).

Fluent Incorporated, 1997, Fluent/Uns and Rampant 4.2 User s Guide

(Fluent Incorporated, Lebanon, NH).

Fluent Incorporated, 1998,

GAMBIT

Modeling Guide (Fluent Incorporated,

Lebanon, NH).

Keey. R.B. Langrish, T A G . and Walker. J.C.F.. 2000, Kiln-drying of

Lumber

(Springer, B e r l i n Germany).

Langrish,

T . A .G . and Keey, R.B. 1996, The effects of air bypassing in

timber kilns on fan power consumption, in Proceedings of

CHEMECA

1996, Sydney, 30 September-2 October, pp. 103-108.

Nijdam. J.J. and Keey, R.B. , 2002. New timber

k i l n

designs for promoting

uniform airflows w i t h i n the wood stack. Trans IChemE, Part A, Chem

En g

Res Des, 80(A7): 739-744.

Sun. Z.F., 2001. Numerical simulation of

flow

in an array of in-line blunt

boards: mass transfer and flow patterns. Chem Eng Sci. 56(5): 1883-1896.

Sun, Z.F., 2002, Correlations for

mass

transfer coefficients over blunt

boards based on modified boundary layer theories, Chem Eng Sci,

57(11), 2029-2033.

Sun, Z.F. and Carrington, C.G., 1999, Effect of stack configuration on

wood

drying

processes,

in Proceedings of 6th International

IUFRO

Wood

Drying Conference. Stellenbosch, South Af r ica 25-28 January.

pp. 89-98.

Sun, Z.F. . Carrington, C.G. and Bannister, P., 2000, Dynamic modell ing of

the wood stack in a wood drying k i l n Trans IChemE. Pi A, Chem Eng

Res Des, 78: 107-117.

CKNOWLEDGEMENTS

The authors gratefully acknowledge the New Zealand Foundation for

Research Science and Technology for supporting this work under contract

UOOX0004.

The

manuscript was received I July 2002 and accepted for publication

after revision 18 June 2004.

Trans I C h emE

Part A,

Chemical Engineering Research and Design

2004, 82 A10): 1344-1352

air flow patterns

Documents