Energy and Buildings 38 (2006) 1230–1239

www.elsevier.com/locate/enbuild

Modeling and simulation of desiccant wheel for air conditioning

Fatemeh Esfandiari Nia a,*, Dolf van Paassen b,1, Mohamad Hassan Saidi a,2
a
Mechanical Engineering Department, Sharif University of Technology, Tehran 11365-8639, Iran
b
Mechanical Maritime and Materials Engineering, Delft University of Technology, 2628 CD Delft, The Netherlands
Received 29 January 2006; received in revised form 3 March 2006; accepted 4 March 2006

Abstract
This paper presents the modeling of a desiccant wheel used for dehumidifying the ventilation air of an air-conditioning system. The simulation
of the combined heat and mass transfer processes that occur in a solid desiccant wheel is carried out with MATLAB Simulink. Using the numerical
method, the performance of an adiabatic rotary dehumidifier is parametrically studied, and the optimal rotational speed is determined by examining
the outlet adsorption-side humidity profiles. The solutions of the simulation at different conditions used in air dehumidifier have been investigated
according to the previous published studies. The model is validated through comparison the simulated results with the published actual values of an
experimental work. This method is useful to study and modelling of solid desiccant dehumidification and cooling system. The modeling solutions
are used to develop simple correlations for the outlet air conditions of humidity and temperature of air through the wheel as a function of the
physically measurable input variables. These correlations will be used to simulate the desiccant cooling cycle in an HVAC system in order to define
the year round efficiency.
# 2006 Elsevier B.V. All rights reserved.

Keywords: Desiccant; Adsorption; Cooling; Modeling; Simulation; Correlation

1. Introduction fuels. The electrical energy requirement can be very low

comparing with conventional refrigeration systems. The source
In desiccant cooling processes, fresh air is dehumidified and of thermal energy can be diverse (i.e., solar, waste heat, natural
then sensibly and evaporatively cooled before being sent to the gas) [2]. Having low coefficient of performance (COP) can be
conditioned space. Since this technique works without considered as the main disadvantage for desiccant cooling
conventional refrigerants, such as fluorocarbons and also since systems. COP values of 0.8–1 are commonly predicted for this
it allows the use of low-temperature heat (low-temperature cycle. COP or coefficient of performance is defined as the space
industrial waste heat or solar energy) to drive the cooling cycle, cooling load divided by thermal energy required to regenerate
it attracted increased attention especially in America, Japan, the desiccant. Some investigators use the heat removed from the
Europe and China [1]. Desiccants remove moisture from the processes air stream divided by the thermal energy required to
surrounding air until they reach equilibrium with it. This regenerate the desiccant.
moisture can be removed from the desiccant by heating it to Kang and Maclain-Cross [3] showed that the dehumidifier is
temperatures around 60–90 8C and exposing it to a regenerative the key component of a desiccant cooling system and the
air stream. The desiccant is then cooled so that it can adsorb cooling COP (coefficient of performance) can be significantly
moisture again. Desiccant cooling cycles are particularly useful improved by improving the performance of this component.
if they are used in humid regions. Since the introduction of this technology, much research on the
The major advantage of desiccant cooling is significant solid desiccant dehumidifiers has been accomplished.
potential for energy savings and reduced consumption of fossil Maclaine-Cross and Banks [4] developed an analogy method
for predicting the coupled heat and mass transfer process in
desiccant dehumidifier wheel. Consecutively, Banks [5–7]
analyzed the coupled heat and mass transfer processes in a
* Corresponding author. Tel.: +31 15 2783115; fax: +31 15 2782460.
E-mail address: [email protected] (F.E. Nia).
porous medium using a nonlinear analogy method and
1
Tel.: + 31 15 2786675; fax: + 31 15 2782460. predicted the performance of a silica gel air dryer. Neti and
2
Tel.: + 98 21 6165522; fax: + 98 21 6100002. Wolf have reported [8] that the analogy method appears to be
0378-7788/$ – see front matter # 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.enbuild.2006.03.020
 F.E. Nia et al. / Energy and Buildings 38 (2006) 1230–1239 1231

resistance model fits the experimental data much better than the
Nomenclature pseudo-gas-side controlled model. Zheng and Worek [13]
discussed the effect of rotary speed on the performance of the
A adsorption potential (kJ/kmol)
desiccant wheel by numerical simulation using an implicit
Ac interface area in a channel (m2) finite differential method. Dai et al. [14] used a finite difference
Ad cross-section area for desiccant layer in a channel model to analyze and explain the ‘‘concentration’’ wave,
(m2) ‘‘thermal’’ wave and middle zone point in the desiccant wheel
Ag cross-section area for air flow (m2) in detail, and the rules to improve the performance of
C isobaric specific heat (J/kg K) dehumidification were discussed using psychrometric charts.
COP coefficient of performance Zhang and Niu [15] presented a two-dimensional coupled heat
dt thickness of the desiccant coating (m)
and mass transfer model which takes into account the heat
Dh hydraulic diameter of a channel (m)
conduction, the surface and gas diffusion in both axial and
h heat transfer coefficient (W m
2 K
1) radial directions and compared the performances of desiccant
hm mass transfer coefficient (kg m
2 s
1) wheels as dehumidifier and enthalpy recovery. Chauch et al.
H enthalpy (kJ/kg) [16] studied the relative importance of gas phase and solid-side
k thermal conductivity (W m
1 K
1) resistance using two film models and showed the importance of
L depth of the rotor (m) solid-side resistances parametrically in small gap passages and
Le Lewis number when the gel size is large.
N wheel speed
The objective of this paper is to develop a model and
Nu Nusselt number
obtaining the model solutions for heat and mass transfer in a
P pressure (Pa) desiccant wheel to study performance of a desiccant air-
qst adsorption heat (J/kg) conditioning system. The numerical solutions for coupled
R gas constant (J/kg K) nonlinear heat and mass transfer equations require significant
t time (s) computational efforts to calculate outlet air states. Correlating
T temperature (8C) functions of temperature and humidity with input conditions are
U velocity (m s
1) developed which allow simulation of the performance of
w water content of the desiccant material (kg/kg)
components in air-conditioning cycles. Yearly simulation runs
Greek letters can then be made. This requires simple and fast correlation
e efficiency functions to limiting the computing time to an acceptable level.
w relative humidity These functions are developed and can be used to model the
r density (kg m
3) cycle.
v humidity ratio (kg/kg)
vs humidity ratio of air in equilibrium with the 2. Governing equations
desiccant or in saturation (kg/kg)
The dehumidifier is a rotating cylindrical wheel of length L
Subscripts and radius R with small channels which walls are adhered
d desiccant with an adsorbent such as silica gel. For simplicity it is
e enthalpy divided into two equal sections: the adsorbing section and the
g gas regeneration section (desorption of water vapor). The
i inlet regeneration and adsorption air streams are in a counter
l latent flow arrangement. The schematic of a balanced rotary
m moisture dehumidifier is illustrated in Fig. 1 and the analysis is based
o outlet on the following assumptions:
R regeneration
s saturation 1. axial heat conduction and water vapor diffusion in the air
st sensible are negligible;
v water vapor 2. axial molecular diffusion within the desiccant is negligible;
3. there are no radial temperature or moisture content
gradients in the matrix;
good for only a small range of conditions. Van Den Bulck et al. 4. hysteresis in the sorption isotherm for the desiccant coating
[9,10] introduced a wave analysis to establish a one- is neglected and the heat of sorption is assumed constant;
dimensional transient model of the rotary heat and mass 5. the channels that make up the wheel are identical with
exchanger with infinite transfer coefficients, and with definite constant heat and mass transfer surface areas;
ones employing the e-NTU method. Pesaran and Mills [11,12] 6. the matrix thermal and moisture properties (support
established a solid-side resistance model and pseudo-gas-side material/desiccant and adsorbed water) are constant;
controlled model to study the vapor transfer process in the silica 7. the channels are considered adiabatic and impermeable;
gel packed bed, their studies revealed that the solid-side 8. the mass and heat transfer coefficients are constant;
1232 F.E. Nia et al. / Energy and Buildings 38 (2006) 1230–1239

Heat transfer equation for the air stream:

dðrg Ag LCg Tg Þ
¼ rg Ug Ag Cg ðTgi
Tgo Þ þ hAc ðTs
Tg Þ (7)
dt
dTg
¼ C1 ðTgi
Tgo Þ þ C3 ðTs
Tg Þ (8)
dt
hAc
C3 ¼ ; C3 ¼ Le C2 (9)
rg LAg Cg

Mass transfer equation for solid desiccant layer:

dðrd wAd LÞ
¼ hm Ac ðv
vs Þ (10)
Fig. 1. Desiccant wheel. dt

9. the adsorption heat per kilogram of adsorbed water is where w is the water content of desiccant material,
constant; rvd
10. the carry over between two air flows is neglected. ¼w (11)
rd
Based on the above assumptions, the model used in this So
analysis is transient and one-dimensional.
One of the channels is divided into to a number of equal step dw h m Ac
¼ ðv
vs Þ (12)
discrete elements or channels as shown in Fig. 2. For each dt rd Ad L
discritized channel that identified in Simulink model as a
framework with inlet conditions for air and storing outputs and For having the equations according to the variables vs, Ts v,
initial condition for solid, as shown in Fig. 2, the energy and T, it can be written as
mass conservation equations can be written as follows. Mass  
@w @’ @w @’ @w
transfer equation for the air stream: dw ¼ dvs þ þ dTs (13)
@’ @vs @’ @v @TS
 
d rg rrvg Ag L or
¼ Ug Ag rg ðvi
vo Þ þ hm Ac ðvs
vÞ (1)
dt dw ¼ S1ðvs ; Ts Þdvs þ S2ðvs ; Ts ÞdTs (14)
dv Ug h m Ac
¼ ðvi
vo Þ þ ðvs
vÞ  
dt L rg Ag L @w @’ @w @’ @w
S1ðvs ; Ts Þ ¼ ; S2ðvs ; Ts Þ ¼ þ
¼ C1 ðvi
vo Þ þ C2 ðvs
vÞ (2) @’ @vs @’ @Ts @Ts
(15)
where
rv So mass transfer equation for desiccant layer will become
v¼ (3)
rg
dvs S2ðvs ; Ts Þ dTs hm Ac
¼
þ ðv
vs Þ
Ac 2L dt S1ðvs ; Ts Þ dt rd Ad LS1ðvs ; Ts Þ
¼ (4)
Ag D2h S2ðvs ; Ts Þ dTs C4
¼
þ ðv
vs Þ (16)
S1ðvs ; Ts Þ dt S1ðvs ; Ts Þ
Ug
C1 ¼ (5)
L where
h m Ac
C2 ¼ (6) h m AC
rg LAg C4 ¼ (17)
rd LAd

Ac 4Dh L
¼ (18)
Ad ðDh þ dt Þ2
D2h

h
hm ¼ (19)
Cg Le

and Le is Lewis number that here is assumed equal 1 for air

Fig. 2. Schematic of descritization of one of the channels. stream.
 F.E. Nia et al. / Energy and Buildings 38 (2006) 1230–1239 1233

Heat transfer for solid desiccant layer:

dðrd Ad LCd Ts Þ dw
¼ qst rd Ad L þ hAc ðTg
Ts Þ (20)
dt dt
dTs hm Ac qst hAc
¼ ðv
vs Þ þ ðTg
Ts Þ
dt rd Ad LCd Cd rd Ad L
¼ C4 C5 ðv
vs Þ þ C6 ðTg
Ts Þ (21)
qst
C5 ¼ (22)
Cd
hAc
C6 ¼ (23)
Cd rd LAd
Relative humidity and saturation pressure can be calculated
by [15]
v s P0
’¼ (24)
ð0:622 þ vs ÞPs
 
6 5294 ð1 þ 1:61vs Þ
Ps ¼ 10 P0 exp (25)
Ts ð0:622 þ vs Þ
For first channel (element) in adsorption period the initial
conditions are known as
Tgi ¼ Tprocess air inlet ¼ Tg initial (26)
vgi ¼ vprocess air inlet (27)
vsi ¼ vs regarding to inlet air Temperature (28)
Ts initial ¼ Tgprocess air inlet (29)

and initial conditions for regeneration period:

Tgi ¼ Tregeneration air inlet (30)
vgi ¼ vregeneration air inlet (31)
vsi ¼ vs regarding to inlet air Temperature (32)
Ts initial ¼ Tgprocess air inlet (33)
Fig. 3. Simulink representation of the model.

3. Simulation platform and subsystems

The model used in this simulation has been build up based time step and the time each channel of wheel passes through the
on a main framework for each element shown in Fig. 2 that air stream can be considered to evaluate of the output
can be regarded as an interaction of four main subsystems. In conditions in different wheel angels.
Fig. 3, a block diagram shows the set-up of the simulation.
Each subsystem is composed of a series of other, more 4. Evaluation of the coefficients and step sizes
detailed, subsystems. The calculation blocks are storing the
response of an output. For simulation of the rotation of the Regarding the derived nonlinear equations of combined
wheel between two counter flow adsorption and regeneration mass and heat transfer we have many coefficients that are
air flow, the convenient user defined functions regarding functions of Tair, vair that can be considered constant with
the time step of revolution has been built up and is jointed considering the average values of them. It improves the
to the main framework for each element. There are a accuracy and reduces the calculation time. Some of the
number of techniques to solve the derived differential coefficients are functions of Ts, vs.
equations. The selected scheme of solutions in MATLAB for The present model was validated by comparing the
initial value problems is Runge–Kutta methods with variable simulated results with the published experimental values
time steps. [17]. The adsorption capacity of a desiccant material is not only
The output conditions of an individual channel can be dependent on the relative humidity of the surrounding air but
considered as average values for each section of the wheel. The also on its temperature. Thus, the relation of capacity to both
1234 F.E. Nia et al. / Energy and Buildings 38 (2006) 1230–1239

temperature and relative humidity must be known. It is this laminar flow arrangement and the hydraulic diameter of the
convenient to find a function that involves both of these hexagonal passes is estimated to be 2.33  10
3 m. The heat
parameters and allows prediction of the desiccant adsorption transfer coefficient becomes
capacity at any given temperature–humidity combination. A
correlation for silica gel given in the literature [18] is Nu kair
h¼ ¼ 43:8 W m
2 K
1 (37)
  2    2  Dh
A A
W ¼ 0:106 exp
þ 0:242 exp
;
8590 3140
A ¼
RT ln ’ (34) 5. The behavior of solutions

So for Eq. (15) it can be written: The rotational speed of a rotary desiccant dehumidifier is
   2  inversely proportional to the sorption time. The rotational
RTs A
S1 ðvs ; Ts Þ ¼ ð2:9e
9ÞA exp
speed of a rotary desiccant dehumidifier is optimum when
’ 8590 the average outlet humidity ratio of the process air flow is the
RTs minimum. A rotary desiccant dehumidifier operates such that
þ ð4:9e
8ÞA exp
’ the moisture absorbed during the adsorption process must be
  2   desorbed during the regeneration process, so when the
A 0:622P0

(35) average outlet humidity ratio in the adsorption stream is
3140 ð0:622 þ vs Þ2 Ps
optimized, the average outlet humidity ratio of the
@w @’ @w regeneration stream is maximized. In other words, the
S2 ðvs ; Ts Þ ¼ þ adsorption and regeneration processes will be optimum at the
@’ @Ts @Ts
   2  same rotational speed. According to the published studies
RTs A
¼ ð2:9e
9ÞA exp
[13] the behavior of the Simulink model output can be
’ 8590 estimated.
  2 
RTs A 5294’ When a desiccant wheel rotates much faster than optimum
þ ð4:9e
8ÞA exp

’ 3140 Ts2 speed, the adsorption and regeneration processes are too short,
   2  which results in poor performance. Also, when the rotational
A
þ ð2:9e
9ÞAR ln ’ exp
speed is low, the adsorption and regeneration processes are too
8590 long and less effective. The outlet humidity ratio profiles on the
  2 
A adsorption-side are shown in Fig. 4A–C. In Fig. 4C the outlet
þ ð4:9e
8ÞAR ln ’ exp

8590 humidity profile at faster rotational speed than optimum speed
(36) (the case for the adsorption time is insufficient, is plotted versus
the wheel angle). This profile shows that the outlet humidity
In the experimental study [15], desiccant wheel has been ratio at the last point, or at an angle of 180, is less than the
used as a dehumidifier and the wheel has 20 cm width and average value. This means that at the end of adsorption process,
the silica gel wall thickness is 0.2 mm. The pitch of the the desiccant wheel can still efficiently dehumidify the air
honeycomb shaped rotor used for experiments has been stream.
3.2 mm  1.8 mm. The heat of adsorption is calculated Therefore the rotational speed should be slower to allow
between 2100 and 2300 kJ/kg. The air velocity 2 m s
1 for more adsorption time. In Fig. 4A the outlet humidity profile is
air flow in both adsorption and regeneration periods have in slower speeds than optimum rotational speed. In this case the
been reported. The important parameter in the modeling of outlet humidity at the end of adsorption is larger than the
the desiccant wheels is rotational speed or the time step average value; this implies that the last portion of the adsorption
which can be determined regarding to the time passed during process is inefficient. The rotational speed of the desiccant
each revolution of the wheel. As a simplifying assumption, wheel should increase to remove this last ineffective portion of
the wheel in this modeling is considered a balanced wheel the adsorption process. Therefore, to improve dehumidification
with two equal cross area and air flow rate for each part. performance, the rotational speed should be in between the
Therefore the time step is two times more than the time cases in Fig. 4A and C. As shown in Fig. 4B, at the optimum
needed for one complete matrix rotation. For desiccant rotational speed, the outlet humidity ratio at the end of the
dehumidifier wheel, the rotational speeds are in the range 1– adsorption process, or at an angle of 180, is equal to the average
30 revolution per hour (RPh). The number of time elements value. This result [13] is used to investigating the optimum
is determined by the maximum time of running the program rotational speeds of desiccant wheel, as the most important
and the time increments that are variable from 0.001 to 0.1 s. design parameter of the wheel, when operating conditions
The number of the elements is 50 for 0.2 m length of the change.
wheel. Fig. 5 shows the prediction of the optimal rotation speed
The heat transfer coefficient is calculated by a correlation for at various regeneration temperatures and compares the
forced convection in internal flows. As stated by Bejan [19], the accuracy of the simulation results with experimental work
Nusselt number is approximately between 3.63 and 4.364 for [17]. As it can be seen in Fig. 5 the model can predict the
 F.E. Nia et al. / Energy and Buildings 38 (2006) 1230–1239 1235

Fig. 4. (A) A simulation solution for adsorption-side outlet humidity ratio profile for wheel with 12 Rph; (B) adsorption-side outlet humidity ratio profiles for wheel
with 15.5 Rph; (C) adsorption-side outlet humidity ratio profile for wheel with 25 Rph.

optimum speed of the wheel with very good accuracy in most be examined. The evaluation is carried out for a dehumidifier
conditions. wheel with silica gel as desiccant material and the sorption
In addition the model has very good accuracy regarding to curve used in the simulation is a correlation for silica gel given
the temperatures and humidity of the outlet air in different in the literature [18] or Eq. (34). The dehumidifier effectiveness
conditions. The result of modeling presented here and is defined as
the published experimental data agree within the vinlet air
voutlet air
acceptable margin of error. The user friendly and flexible e¼ (38)
vinlet air
programming environment of Simulink provides an excellent
alternative. The advantages of such modeling platform Fig. 6a shows that the dehumidifier effectiveness decreases
include notion of adaptability, tenability to generate accurate with the inlet air temperature increases. As shown in Fig. 7a,
solutions that are significantly concise with minimum outlet air temperature increases with inlet air temperature. The
calculation time as well. wheel speed is one of the most important parameters in the
dehumidification itself and in many other applications. The
6. Sensitivity of the individual parameters results in Figs. 6c and 7c suggest that the wheel speed affects
the effectiveness and outlet air temperature and therefore on the
In this section the effect and the importance of the individual outlet air relative humidity and enthalpy considerably. The
parameters such as mass flow rate or air velocity, wheel speed, dehumidification effectiveness for a desiccant wheel depends
diameter of air passages, desiccant thickness and inlet on the regeneration temperature as indicated in Fig. 6b. Fig. 7b
conditions of air temperature and humidity for process air shows the increase of outlet air temperature with regeneration
stream as well as inlet air temperature of regeneration stream on air temperature increases. It can be proposed a simple linear
the outlet air conditions and dehumidification effectiveness will function to show the variation of outlet air temperature with
1236 F.E. Nia et al. / Energy and Buildings 38 (2006) 1230–1239

large [16]. Therefore for constant air passage gap for the
solutions here (2.33 mm) there is limitation to select the range
for size of wall thickness. The initial conditions of humidity for
inlet air affects on the effectiveness and outlet temperature as it
has been shown in Figs. 6g and 7g. In the dehumidifier the inlet
air for both process air and regeneration air have same absolute
humidity but different temperatures. The simulation data for
different mass flow rate according to different air velocity are
shown in Figs. 6f and 7f. The variation of outlet air conditions
with mass flow rate is considerably high and it can be
considered as an important parameter. According to the
simulation data, the effect of air passages on the wheel
effectiveness and outlet temperature on effectiveness and
temperature shown in Figs. 6d and 7d. Regarding to modeling
and the mass transfer equations in this paper, the moisture
diffusion in solid layer has been neglected. And as mentioned
before in this section, for constant wall thickness in the
solutions here (0.2 mm) there is limitation to select the range
for the size of air passages.

7. The correlations for effectiveness and outlet air

conditions

In this section, the correlations for dehumidification

effectiveness and outlet air temperature are presented according
to the sensitivity of individual parameters presented in Figs. 6
Fig. 5. Prediction of the optimal rotation speed at various regeneration tem-
and 7. The correlations presented in this section were
peratures.
determined by minimizing the maximum difference between
over 500 simulated effectiveness and temperature values. The
inlet air temperature. Regarding to modeling in this paper and coefficients of the correlations were determined using
the mass transfer equations, the moisture diffusion in solid layer optimization routine in Matlab.
has been neglected. This simplification is not good assumption The correlation for outlet air temperature is proposed by
when the gap or air passage diameter is small or the thickness is Eq. (39) and in Fig. 8 it has been shown comparing with the

Fig. 6. Effect of individual parameters on effectiveness for constant conditions of air and wheel: vi = 8 g/kg dry air; Ti = 30 8C; N = 15 RPh; U = 2 m s
1;
dt = 0.2 mm; TR = 90 8C; Dh = 2.33 mm and different individual parameter as shown with profiles a–g.
 F.E. Nia et al. / Energy and Buildings 38 (2006) 1230–1239 1237

Fig. 7. Effect of individual parameters on outlet air temperature for constant conditions of air and wheel: vi = 8 g/kg dry air; Ti = 30 8C; N = 15 RPh; U = 2 m s
1;
dt = 0.2 mm; TR = 90 8C; Dh = 2.33 mm and different individual parameter as shown with profiles a–g.

outlet air temperatures from the simulation results:

Tout ¼ g1 ðNÞg2 ðTi Þg3 ðdt Þg4 ðTR Þg5 ðvi Þg6 ðDh Þg7 ðUÞ;
g1 ðNÞ ¼
0:0002N 2 þ 0:0112N þ 0:4201;
g2 ðTi Þ ¼
0:0001Ti2 þ 0:0275Ti þ 0:7993;
g3 ðdt Þ ¼
18:79dt2 þ 7:92dt þ 1:75;
g4 ðTR Þ ¼
0:0004TR2 þ 0:1255TR þ 0:6757;
g5 ðvi Þ ¼ 594:48v2i þ 26:76vi þ 3:79;
g6 ðDh Þ ¼
0:039D3h þ 0:026D2h þ 0:603Dh þ 0:0912;
g7 ðUÞ ¼
0:060U þ 0:7973 (39)

The maximum difference between simulated and correlated

outlet air temperature is 2%. There is a significant error for
some of the conditions of wheel and air. It is because the
correlation can be proposed for only a limited range of
individual parameters selected to obtain simulated data. The
coefficients calculated from optimization routine are different
in different range of parameters. Therefore the correlations
proposed here can be considered for only the range of the
conditions that covers the selected data to make that
correlations and the modeling of cycle needs more correlations.
The correlation for outlet air humidity is derived through the
following relation:

vout ¼ vin
evin (40)

The optimization can be proposed by polynomial fitting of

the variables and propose a correlation as given in Eq. (41).
Fig. 9b shows the outlet air absolute humidity calculated with
Eqs. (40) and (41) that is approached by polynomial equations.
The maximum difference between simulated and correlated Fig. 8. Outlet air temperature calculated with Eq. (39) compared to the
outlet air humidity is 2%. These equations are for silica gel simulated outlet air temperature.
1238 F.E. Nia et al. / Energy and Buildings 38 (2006) 1230–1239

Fig. 9. (a) Wheel efficiency calculated with Eq. (40) compared to the simulated efficiency; (b) outlet air absolute humidity calculated with Eqs. (40) and (41)
compared to the simulated outlet air humidity.

wheel with the sorption curve given in Ref. [18]: parameters. The maximum difference between the results
from the proposed correlations and simulation results are
e ¼ f1 ðNÞ f2 ðTi Þ f3 ðdt Þ f4 ðTR Þ f5 ðvi Þ f6 ðDh Þ f7 ðUÞ; 2%, provided that the limited range of the variables are
f1 ðNÞ ¼
0:0001N 2 þ 0:0042N þ 0:4474; considered.
f2 ðTi Þ ¼
0:0001Ti2
0:0031Ti þ 0:8353;
References
f3 ðdt Þ ¼
21:67dt2 þ 6:93dt þ 1:34;
[1] J.J. Jurinak, J.W. Mitchell, W.A. Beckman, Open-cycle desiccant air
f4 ðTR Þ ¼
0:0001TR2 þ 0:0355TR
0:4924;
conditioning as an alternative to vapor compression cooling in residential
f5 ðvi Þ ¼ 592:77v2i
41:23vi þ 1:283; applications, Transactions of the ASME 106 (1984) 252–258.
[2] D.G. Waugaman, A. Kini, C.F. Kettleborough, A review of desiccant cooling
f6 ðDh Þ ¼
0:0572D3h þ 0:0933D2h þ 0:6139Dh
0:0922; systems, Journal of Energy Resources Technology 115 (1993) 1–8.
f7 ðUÞ ¼
0:0611U þ 0:8376 (41) [3] T.S. Kang, I.L. Maclain-Cross, High performance solid desiccant cooling
cycles, Transactions of ASME 111 (1989) 176–183.
[4] I.L. Maclaine-Cross, P.J. Banks, Coupled heat and mass transfer in
regenerators––prediction using an analogy with heat transfer, Interna-
8. Conclusions tional Journal of Heat and Mass Transfer 15 (1972) 1225–1242.
[5] P.J. Banks, Coupled equilibrium heat and single adsorbate transfer in
In this paper a heat and mass transfer model has been fluid flow through a porous medium-2: predictions for a silica-gel air
presented which predicts the temperature and humidity states drier using characteristic charts, Chemical Engineering Science 27
(1972) 1157–1169.
of the outlet air from a desiccant wheel and the optimum speed
[6] P.J. Banks, Prediction of heat and mass regenerator performance using
of the wheel when used as a dehumidifier. Furthermore, it has nonlinear analogy method. Part 1. Basis, ASME Journal of Heat Transfer
shown that the results of simulation and the published 107 (1985) 222–229.
experimental data are agreeable within an acceptable margin [7] B.P.J. Banks, Prediction of heat and mass regenerator performance using
of error. The Simulink code provides the possibility to couple nonlinear analogy method. Part 2. Comparison of methods, ASME Journal
the simulation model of one component such as this case, the of Heat Transfer 107 (1985) 230–247.
[8] S. Neti, E.I. Wolfe, Measurements of effectiveness in a silica gel rotary
drying wheel with models of total system. Another advantage exchanger, Applied Thermal Engineering 20 (2000) 309–322.
is the tools for model reduction. Here the reduced physical [9] E. Van Den Bulck, J.W. Mitchell, S.A. Klein, Design theory for rotary heat
module could be applied into simple correlation functions. and mass exchangers. I. Wave analysis of rotary heat and mass exchangers
This makes the simulation of complex systems such as the with infinite transfer coefficients, International Journal of Heat and Mass
HVAC systems faster when year round analyses are needed. Transfer 28 (1985) 1575–1586.
[10] E. Van Den Bulck, J.W. Mitchell, S.A. Klein, Design theory for rotary heat
The modeling approach is adapted for a commercial cooling and mass exchangers. II. Effectiveness—number-of-transfer-units’
cycle, through development of simple correlation functions, to method for rotary heat and mass exchangers, Inernational Journal of Heat
calculate the outlet air states according to individual and Mass Transfer 28 (1985) 1587–1595.
 F.E. Nia et al. / Energy and Buildings 38 (2006) 1230–1239 1239

[11] A. Pesaran, A.F. Mills, Moisture transport in silica gel packed beds. I. [15] L.Z. Zhang, J.L. Niu, Performance comparisons of desiccant wheels for air
Theoretical study, International Journal of Heat and Mass Transfer 30 dehumidification and enthalpy recovery, Applied Thermal Engineering 22
(1987) 1037–1049. (2002) 1347–1367.
[12] A. Pesaran, A.F. Mills, Moisture transport in silica gel packed beds. II. [16] Y.K. Chauch, P. Norton, F. Kreith, Transient mass transfer in parallel
Experimental study, International Journal of Heat and Mass Transfer 30 passage dehumidifiers with & without solid side resistance, ASME Journal
(1987) 1051–1060. of Heat Transfer 111 (1989) 1038–1044.
[13] W. Zheng, W.M. Worek, Numerical simulation of combined heat and mass [17] A. Kodama, T. Hirayama, M. Goto, T. Hirose, R.E. Critoph, The use of
transfer process in a rotary dehumidifier, Numerical Heat Transfer Part A psychometric charts for the optimization of a thermal swing desiccant
23 (1993) 211–232. wheel, Applied Thermal Engineering 21 (2001) 1657–1674.
[14] Y.J. Dai, R.Z. Wang, H.F. Zhang, Parameter analysis to improve rotary [18] G. Steich, Performance of rotary enthalpy exchangers, MSc thesis in
desiccant dehumidification using a mathematical model, International Mechanical Engineering, University of Wisconsin-Madison, 1994.
Journal of Thermal Science 40 (2001) 400–408. [19] A. Bejan, Convection Heat Transfer, John Wiley and Sons, New York, NY,
1984.