International Journal of Thermal Sciences 48 (2009) 1664–1671

Contents lists available at ScienceDirect

International Journal of Thermal Sciences

www.elsevier.com/locate/ijts

Model validation and case study on internally cooled/heated

dehumidifier/regenerator of liquid desiccant systems
Yonggao Yin ∗ , Xiaosong Zhang, Donggen Peng, Xiuwei Li
School of Energy and Environment, Southeast University, 2 SiPaiLou Road, Nanjing 210096, China

a r t i c l e i n f o a b s t r a c t

Article history: Traditional dehumidifiers and regenerators of liquid desiccant systems often use packed columns sup-
Received 9 November 2007 porting adiabatic heat and mass transfer between air and liquid desiccant. As new-style equipment,
Received in revised form 18 December 2008 internally-cooled dehumidifiers can improve dehumidification performance due to restraining tempera-
Accepted 18 December 2008
ture increase of the desiccant. Similar to internally-cooled dehumidifiers, an idea of internally heating
Available online 15 January 2009
is imitated to put forward internally-heated regenerators. The uniform mathematical model for an inter-
Keywords: nally cooled dehumidifier and internally heated regenerator was presented and validated by comparison
Liquid desiccant of computation results with experimental data in this study. The case study focused on the parameters
Internally heated regenerator distribution comparisons of the internally cooled/heated dehumidifier/regenerator with adiabatic ones
Internally cooled dehumidifier and demonstrated coupled heat and mass transfer behavior. The results show that the internally-heated
Optimal design regenerator can produce higher regeneration efficiency than the adiabatic one to produce better energy
efficiency and eliminate the dehumidification possibility which would happen in adiabatic regenerators.
The internally-cooled dehumidifier can also provide better dehumidification performance comparing with
the adiabatic one; however its benefit would be not as good as the internally-heated regenerator. In ad-
dition, effect of the width of the air channel on internally cooled/heated dehumidifier/regenerator was
discussed and the results can help the optimal design of this kind of dehumidifiers and regenerators.
© 2008 Elsevier Masson SAS. All rights reserved.

1. Introduction adiabatic dehumidifiers and regenerators. All results indicated that

in the packed dehumidifiers and regenerators heat and mass trans-
Liquid desiccant system can utilize low grade heat such as so- fer wears away with the progress of processes due to the change of
lar energy [1] to bring remarkable potential for saving energy [2] local temperature and concentration gradients, which restricts the
and improving indoor air quality. Many new air conditioning sys- performance of them. Internally-cooled dehumidifiers like coil-type
tems using liquid desiccant have been proposed as alternatives to dehumidifiers are different from the adiabatic dehumidifiers using
the traditional cooling systems [3]. Dehumidifier and regenerator packing, and in the former ones the release latent generated by de-
are the most important components, since they determine the effi- humidification can be removed by the coolant to keep higher mass
ciency and thermal performance of liquid desiccant systems. In the transfer gradients. Khan [13] presented the performance analysis
dehumidifier, air is contacted with the concentrated desiccant so- of an internally cooled absorber using thin plate heat and mass
lution and dehumidified; in the regenerator, the diluted and heated exchanger cooled by direct evaporation with lithium chloride as
desiccant solution is directly contacted with ambient air, which can desiccant. Enthalpy effectiveness and humidity effectiveness were
take out the water from the desiccant solution. brought out to define the thermal performance of the absorber. It
In the past decades, the packed dehumidifiers and regenerators was found that the number of mass transfer units had great effect
were highly attractive since packed columns can provide long res- on the enthalpy and humidity effectiveness. Khan [14] investigated
idence time and much contacting area with simple structure and numerically on an internally cooled dehumidifier using tube-fin
low cost. Coupled heat and mass transfer between liquid desic- exchanger with the air crossing flow using one-dimensional defi-
cants and air happen simultaneously and adiabatically in the pack- nite difference model. Jain et al. [17] conducted experimental study
ing. Many researchers (Grossman [4], Gandhidasan [5], Stevens [6], on wetted wall column dehumidifier with lithium bromide as des-
Goswami [7], Chung [8], Liu [9], Yin [10,11], Ren [12]) have con- iccant. It was found that the improper wetness had important ef-
ducted some experimental and theoretical studies on this kind of fect on the performance of dehumidifier and regenerator and gave
the wetness factor. Theoretical model predicted the experimental
data with ±30%. Saman and Alizadeh [18] presented the experi-
* Corresponding author. mental results using cross-flow type plate heat exchanger (PHE)
E-mail address: [email protected] (Y. Yin). as a dehumidifier/cooler. Many flow passages were separated by

1290-0729/$ – see front matter © 2008 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ijthermalsci.2008.12.017
 Y. Yin et al. / International Journal of Thermal Sciences 48 (2009) 1664–1671 1665

Nomenclature

Cp specific heat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . kJ/(kg ◦ C) Sh Sherwood number, dimensionless

Comp. computation result T temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ◦ C
D diffusion coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . m2 /s Ua air velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . m/s
d width of air channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . m UE uncertainty in comparison
E error between simulation and measurement data UM uncertainty of measurement data
Exp. experiment data Us uncertainty of simulation results
Ga flow rate of air . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . m3 /s W width . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . m
Gs flow rate of desiccant . . . . . . . . . . . . . . . . . . . . . . . . . . . . kg/s xi input parameters of the model
H height . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . m X desiccant concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . %
ha enthalpy of air . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . kJ/kg z height of control volume . . . . . . . . . . . . . . . . . . . . . . . . . . . m
hC heat transfer coefficient between air and
desiccant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . W/(m2 ◦ C) Greek letters
hD mass transfer coefficient based on air humidity ratio β wetness coefficient, dimensionless
difference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . kg/(m2 s)
η efficiency, dimensionless
hs enthalpy of desiccant solution . . . . . . . . . . . . . . . . . . kJ/kg
ω humidity ratio of the air . . . . . . . . . . . . . . . . . . . . . . . . kg/kg
hw heat transfer coefficient between water and
ρ density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . kg/m3
desiccant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . W/(m2 ◦ C)
h v ,s enthalpy of moisture air on solution surface . . . kJ/kg Subscripts
ms desiccant flow rate in control volume . . . . . . . . . . . kg/s
a air
M measurement result
in inlet
k coefficient in correlations, dimensionless
out outlet
Le Lewis number, dimensionless
NTU Transfer Unit Number, dimensionless reg regeneration
Qh heat supplied . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . kW s desiccant solution
qv latent heat of vaporization . . . . . . . . . . . . . . . . . . . . . . kJ/kg Ts, sat saturation status under temperature of T
Re Reynolds number, dimensionless v vapor
S simulation result v, s vapor on solution surface
Sc Schmidt number, dimensionless w cooling water

thin plastic plates, and one side of each thin plastic plate was mass transfer driving force–vapor partial pressure difference be-
for the desiccant solution/air passage; the other side is for wa- tween liquid desiccant and air reduces; as a result, dehumidifica-
ter/air passage. So the indirect evaporation process in the water/air tion performance would decay with the progress of the process in
passage provided cooling to the dehumidification process. The ex- dehumidifiers. Similarly in regeneration processes of desiccant so-
perimental results indicated the effects of inlet parameters of air lution, temperature of liquid desiccant goes down in adiabatic re-
and solution on the PHE effectiveness. Considering the fact that generators because it is necessary for desiccant solution to provide
there were different velocities and properties in air layer and des- latent heat of vaporization to transfer water in the desiccant so-
iccant falling film in the cross-section normal to the flow, some lution to air. Hence, the regeneration performance would become
investigators [19–22] considered the coupled heat and mass trans- weaker and weaker with the process going in regenerators. In or-
fer as two-dimensional cases using Navier–Stokes equations, but der to keep more driving force in dehumidifiers and regenerators,
for solution of the equations, very complex computations have to heated or cooled sources are proposed to import to the regener-
be taken besides assumptions about boundary conditions, so sim- ation and dehumidification processes. Here plate heat exchanger
plified models need developing and validating to probe into the is used for the internally cooled/heated dehumidifier/regenerator.
heat and mass transfer behavior between air and desiccant in de- Fig. 1(a) shows the schematic of internally cooled/heated dehu-
humidifiers and regenerators. midifier/regenerator. Air and desiccant enter the plate exchanger
Similar to internally-cooled dehumidifiers, an idea of internally from the top, and desiccant solution (LiCl–H2 O) is sprayed to the
heating is imitated to put forward internally-heated regenerators.
walls by distributing device and flow down along walls to the bot-
Till now, little work has been conducted on internally-heated re-
tom by its gravity; there air and desiccant solution contact directly
generation processes. In addition, simultaneous heat and mass
to conduct the heat and mass transfer. Fig. 1(b) shows the fluid
transfer between liquid desiccant and air is very complex, and
flow configuration in a unit. In each unit there are two channels in
it need further analysis on the coupled heat and mass trans-
the dehumidifier/regenerator with 1 m height; one (channel II) is
fer processes of internally cooled/heated dehumidifier/regenerator.
for the liquid desiccant and air with the width of 12 mm, and the
This paper will put forward the idea of internal heating source
other (channel I) is for coolant or heating fluid with the width of
based on internally cooling, and investigate simultaneous heat and
2.5 mm. When dehumidification process is necessary, the coolant
mass transfer behavior of an internally-cooled dehumidifier and
enters into channel I. While in regeneration processes, the heating
internally-heated regenerator using a mathematical model vali-
dated by experimental data. fluid enters into the channel I.
Along the height of the dehumidifier and regenerator, finite
2. Methodology segments are divided to generate control volumes for describing
the heat and mass transfer processes, and the control volume is
2.1. Mathematical model shown in Fig. 1(c). To demonstrate the simultaneous heat and mass
transfer for the processes, following assumptions are made in the
In adiabatic dehumidifiers, temperature of liquid desiccant and derivation of the steady-state governing equations for the dehu-
air would rise because of the release of vaporization heat, and midifier and regenerator:
1666 Y. Yin et al. / International Journal of Thermal Sciences 48 (2009) 1664–1671

Fig. 1. Internally cooled/heated dehumidifier/regenerator.


(1) Due to very small width of channels and desiccant, air and dha = NTU · Le(h v ,s − ha ) + (Le − 1)(q v + Cp v · T a )(ω T s,sat − ωa )
water well mixing in the cross-section normal to their flow,
gradients for each fluid only exist in their respective flow di- 
h w (1 − β)( T w − T a ) dz
rections; + (8)
βh D H
(2) Considering the width of the channel is very small, the proper-
ties of water, desiccant solution and air are considered uniform The enthalpy of liquid desiccant is changed because of heat ex-
within the control volume; change with both the air and the heating or coolant fluid, defined
(3) The Lewis number keeps constant under a set of definite inlet as:
parameters;
(4) Local wall temperature of the heat exchanger is equal to the
G w · Cp w · dT w + d(G s · h s ) + G a · dha = 0 (9)
local water temperature.
Here the enthalpy change of the desiccant solution is defined as:
Jain et al. [17] showed that only a portion of total available area
was actually wetted and available for heat and mass transfer. So
dh s = Cps · dT s (10)
the wetness efficiency should be concerned in description of the
simultaneous heat and mass transfer. Mass transfer happens only Combining Eqs. (9) and (10), yield to:
at wetted area under the vapor partial pressure difference between
 
the liquid desiccant surface and air: 1 Ga Gw Ga
dT s = − · dha + Cp w · dT w + Cps · T s · dωa (11)
1 Cps Gs Gs Gs
G a · dωa = h D · (ω T s ,sat − ωa ) · W · β dz (1)
2
Considering the energy balance of control volume, the temperature
Here, of heating water or coolant can be defined as:
Ga = d · W · Ua (2)
dT w β( T s − T w ) + (1 − β)( T a − T w )
= hw · W (12)
Heat transfer conducts both wetted area and non-wetted area un- dz G w · Cp w
der the temperature difference between air and desiccant or walls:
According to the mass balance between the desiccant solution and
1 the air, the concentration change of the desiccant solution control
G a · Cpa · dT a = h C ( T s − T a ) W β dz
2 volume can be calculated as:
+ h w ( T w − T a ) W (1 − β) dz (3)
dG s
d Xs = − Xs (13)
The enthalpy of the air is changed resulting from the change of G s + dG s
temperature and humidity, and therefore the total heat exchange
amount of the air is defined as: Here,

dha = Cpa · dT a + (q v + Cp v · T a ) dωa (4) dG s = −G a dωa (14)

Substitute Eqs. (1) and (3) into (4) to yield: Eqs. (1) and (2) are the governing equations of heat and mass
 transfer processes of air, which is used to solve the temperature
dha = β h C · ( T s − T a ) + h w (1 − β)( T w − T a ) and humidity ratio of air in control volumes. Eq. (10) can infer the
 2W dz temperature change of solution in control volumes. Temperature
+ β h D · (ω T s ,sat − ωa ) · (q v + Cp v · T a ) (5)
Ga change of cooling or heating fluid is determined by Eq. (11). Af-
ter calculating air parameters, flow rate and concentration change
The Lewis number and NTU are defined as:
of desiccant solution are computed by Eqs. (12) and (13). During
hC above computation, the Lewis number is usually assumed as con-
Le = (6)
h D · Cpa stant, however the NTU number is greatly depended on the mass
hD · W · Hβ transfer coefficient between air and desiccant solution. Therefore,
NTU = (7) experimental data is necessary to determine the mass transfer co-
G a /2
efficient between air and desiccant solution and to validate the
Substitute Eqs. (6) and (7) into (5) to yield: model.
 Y. Yin et al. / International Journal of Thermal Sciences 48 (2009) 1664–1671 1667

2.2. Empirical correlations and model validation Table 1

Inlet parameters and uncertainties of air and desiccant solution in experiments.
Since many assumptions and parameter estimation have to be T a (◦ C) U a (m/s) ωa (g/kg) T s (◦ C) X s (%) G s (kg/s)
used to the model, it is necessary to validate the model by compar- Dehumidification 26–32 1.5–4 9.5–14.5 20–28 36–40 0.12
ison of the computation results with the experimental data. Here Regeneration 21–30 1.5–4 6–12 55–78 32–39 0.1–0.12
the model accuracy is verified by comparison of computation out- Uncertainty 0.3 2% 5% 0.3 0.6 5%
let parameters of air and desiccant with the experimental outlet
measurement data. Experimental data and more detailed informa-
tion about the setup can be referred to the paper by Yin et al. [24].
Parameters of air and liquid desiccant have important effect
on the heat and mass transfer in internally cooled/heated dehu-
midifiers/regenerators. Here we concentrated our attention to two
main parameters of them, which are desiccant temperature and air
velocity. Based on this consideration, some experiments were car-
ried out. The experimental parameters are shown in Table 1. The
mass transfer coefficient can be expressed by the Sh dimensionless
number. Based the experimental data, the nonlinear regression cor-
relation is drawn as follows in dehumidification and regeneration.

In dehumidification:

Sh = 4.513 × 10−3 k · Re1.56 Sc0.33 (15)

−2.991
Here k = 76.456T s (16)
In regeneration:

Sh = 4.6767k · Re1.55 Sc0.33 (17)

4 −3.36
Here k = 5.52 × 10 T s (18)
The mass transfer coefficient can be calculated by:
D · Sh
hD = (19)
d
Here d is the width of air channel, the value of which in experi-
mental is 0.012 m.

To determine the validation of the model, a comparison error,

E, is defined with respect to the measurement data, M and the
simulation result S. And the uncertainty U E in the comparison er-
ror is expressed as the follows proposed by Coleman and Stern
[15].

E=M−S (20)

UE = 2
UM + U 2S (21)

Here U M is the uncertainty in the measurement and U S is the un-

certainty in the simulation. The measurement uncertainties of dif-
ferent parameters are shown in Table 1. The simulation model un-
certainty U S is determined by the uncertainty propagation equa-
tion proposed by Taylor [16].
 2  2  2
∂U S ∂U S ∂U S
US = δ x1 + δ x2 + ··· + δ xn (22)
∂ x1 ∂ x2 ∂ xn
Here xi term is the input parameter to the model. The validation
process is a comparison of E and U E . If absolute value of E is less
than U E , then the combination of all the errors of measurement
and simulation is smaller than the estimated validation uncertainty
and the validation is achieved, the level of validation of the model
is U E , otherwise, the proposed model needs some improvement.
Based on Eq. (22) and Table 1, calculation results showed that the
uncertainty of the model for predicted outlet temperature of air,
desiccant and outlet air humidity were respectively 4.83%, 6.59%
and 8.32%.
Fig. 2 displays the comparison results of outlet air temperature,
humidity ratio and desiccant temperature under different oper- Fig. 2. Comparison of computation results with experimental data for dehumidifica-
ation conditions for internally-cooled dehumidification processes. tion.
1668 Y. Yin et al. / International Journal of Thermal Sciences 48 (2009) 1664–1671

Table 2
Comparison of the computation results with experimental data for regeneration.

G a,in T a,in ωa,in T w ,in G s,in T s,in X s,in Le ωa,out T a,out T s,out
(m3 /s) (◦ C) (g/kg) (◦ C) (kg/s) (◦ C) (%) Exp. Comp. Exp. Comp. Exp. Comp.
1 0.062 26.5 11.4 70.5 0 .1 70.0 37.5 0.3 24.5 24.8 37.1 37.3 66.3 65.4
2 0.067 26.5 11.5 70.5 0 .1 70.8 38.0 0.35 23.9 23.9 38.9 38.8 64.8 65.2
3 0.076 26.0 11.7 70.5 0 .1 71.0 38.0 0.35 24.0 23.7 39.8 39.5 64.9 64.1
4 0.051 21.0 6.3 65.0 0.12 64.5 32.2 0.39 20.8 21.1 29.4 29.4 58.6 59.1
5 0.050 21.7 6.3 74.0 0.12 74.3 33.6 0.45 28.0 28.4 34.1 34.4 67.7 67.8
6 0.050 22.4 6.3 78.0 0.12 77.1 34.8 0.5 27.3 28.2 35.3 35.6 73.1 71.4
7 0.050 22.9 6.7 76.0 0.12 74.0 36.2 0.5 25.5 25.3 36.8 36.8 70.3 69.3
8 0.050 23.1 6.6 76.0 0.12 74.5 37.3 0.5 24.3 23.7 37.1 37.0 69.9 69.9
9 0.050 25.1 7.2 78.0 0.12 75.1 38.3 0.5 23.5 22.4 37.7 37.2 70.3 70.3
10 0.050 26.6 7.2 78.0 0.12 73.0 39.2 0.5 21.4 21.6 38.4 38.7 69.9 70

The comparison results show the difference of 5% between pre- Table 3

dicted value and experimental data, which are very close or less Inlet parameters of air and desiccant for the case study.
than the uncertainties of the model for predicted results. The ac- G a,in T a,in ωa,in G s,in T s,in X s,in Gw T w ,in
curacy of the model is under influence of many factors, such as (m3 /s) (◦ C) (g/kg) (kg/s) (◦ C) (%) (kg/s) (◦ C)
measure error of air and desiccant parameters, approximation of Dehumidification 0.0131 28 11.5 0.008 26 39 0.04 24
parameters in the model, of which the Lewis number and effi- Regeneration 0.0144 29 11.5 0.008 60 37.5 0.04 65
cient contact area between air and desiccant are two important
parameters. It is found that Lewis number was changed between
of the hot water, and therefore the heating water temperature de-
0.7 and 1.5 under the experimental operation conditions, and also
creases from 65 to 61 ◦ C gradually. From the figure, it is found
we found the same concern described in the paper by Jain et al.
that in the internally-heated regenerator the temperature of the
[17], only part of the total available area was wetted by desiccant
desiccant solution decreases from 60 to 45.2 ◦ C firstly and then in-
solution and available for heat and mass transfer. Here the wet co-
creases. The reason is that at the beginning the more mass transfer
efficient used to describe the efficient contact area was around 0.7 would take place between solution and air, and furthermore the air
during the validation, which is greatly depended on flow rate of temperature is very low at the entrance; it means that desiccant
liquid desiccant and air. solution would expend much heat for that. With the progress of
Table 2 shows the comparison of the computation results of the process, the desiccant temperature would increase due to the
outlet parameters with experimental data for internally-heated domination of the internally hot water.
regenerations. The outlet humidity ratio and temperature of air Fig. 3(b) displays the distribution of the desiccant concentration
and desiccant temperature are mainly concerned. Comparing the in both regenerators. The concentration in the internally-heated re-
computation results and experimental data, it is found that the generator increases from 37.5% to 38.3% along the height, but in
differences between them are acceptable and the Lewis number the beginning increases and then decreases in the adiabatic regen-
changed between 0.3 and 0.5 under experimental conditions. The erator, which means the dehumidification has happened in it and
validations indicate that the model can give reasonable prediction air is dehumidified, shown in Fig. 3(d). The reason is that the des-
for the heat and mass transfer between air and liquid desiccant iccant temperature decreases sharply shown in Fig. 3(a) so that
and in the following section a case study conducted based on the the vapor partial pressure of desiccant solution surface is less than
validated model. that of the air, which means that no enough heat is supplied for
the whole regeneration process, however the internally hot water
3. Case study and discussion can restrain it to give the good regeneration performance. Thus, it
is very important to choose reasonable regenerator physical model
Here the main objective is to conduct a case study to com- and operation parameters.
pare the internally cooled/heated dehumidifier/regenerator with Fig. 3(c) shows the temperature distribution of air in both re-
the adiabatic ones, and obtain the distribution characteristic of the generators. From the figure it seems there is almost same air distri-
parameters of the liquid desiccant and air along the height of the bution in both regeneration processes. In our experiment data we
dehumidifier/regenerator, and to analyze the heat and mass trans- also found the result, which is almost same outlet air temperature
fer processes. The physical model of the dehumidifier/regenerator was obtained in both regenerators, and the detailed experimen-
is shown in Fig. 1, and the dehumidifier/regenerator is adiabatic if tal results could be referred to the paper by Yin et al. [24]. To
the coolant/heating fluid does not work, otherwise the dehumid- discuss energy efficiency of the two regeneration models, thermal
ifier/regenerator is internally cooled/heated if the coolant/heated efficiency of regeneration is often defined as follows:
fluid works. The operation conditions in per channel unit are Δωa · q v
shown in Table 3. The coolant/heating fluid is water. The desiccant ηreg = (23)
Qh
is aqueous lithium chloride (LiCl–H2 O). The properties of the desic-
cant are referred to the paper by Conde [23]. It was supposed that In the above equation, Q h is heat power supplied for the regenera-
wetness efficiency was 0.8 during the simulation and the Lewis tion. In desiccant regeneration process, suppose that the desiccant
number was 0.5 in the regeneration and 1.2 in the dehumidifica- from the regenerator is heated directly and circulated in the re-
tion according to the test results. generator, and therefore for the regeneration unit the energy con-
Fig. 3 demonstrates the comparison of the solution, air and sumption is used to increase the enthalpy of the air, i.e.
water parameters in both internally heated and adiabatic regen- Δωa · q v
erators. Fig. 3(a) indicates the distributions of temperature of the ηreg = (24)
G a ρa (Cpa Δta + Δωa · q v )
solution and water along the height of the regenerators. The tem-
perature of solution in the internally-heated regenerator is obvi- According to Fig. 3(c), we can suppose the change of air tempera-
ously higher than that in the adiabatic one due to the attendance ture is same in both regenerators. But the change of air humidity
 Y. Yin et al. / International Journal of Thermal Sciences 48 (2009) 1664–1671 1669

Fig. 3. Comparison of parameters distribution of internally-heated regenerator with adiabatic one.

Fig. 4. Comparison of parameters distribution of internally-cooled dehumidifier with adiabatic one.

ratio in the internally-heated regenerator is much more than that Fig. 4 shows the comparison of parameters of the solution and
in the adiabatic one shown in Fig. 3(d). Combining Eq. (20), it is air in the adiabatic and internally cooled dehumidifiers. Fig. 4(a)
concluded that the regeneration efficiency of the internally-heated displays that temperature distribution of the desiccant and water
regenerator is higher than that of the adiabatic one. in both dehumidifiers. In the adiabatic dehumidifier, the desiccant
1670 Y. Yin et al. / International Journal of Thermal Sciences 48 (2009) 1664–1671

Fig. 5. Effect of the Lewis number on the outlet parameter of dehumidification.

Fig. 6. Effect of the channel II width on the dehumidification with different air flow
temperature increases from 26.0 to 30.6 ◦ C gradually because of rates.
the release of vapor latent heat during dehumidification process,
which would make mass transfer between air and desiccant de-
cay, however internally-cooled dehumidifier can remove the latent
heat by internally coolant to keep lower temperature of desiccant,
which manage to give less outlet humidity ratio of the air, just as
shown in Fig. 4(b) humidity ratio of the dehumidified air is less
1 g/kg than that in the adiabatic one, anyway the difference is not
as obvious as that in the regenerators.
Fig. 4(c) depicts the comparison of temperature distributions of
air in both dehumidifiers. In this case, the figure shows that the air
temperature decreases at the beginning and then increases in both
dehumidifiers, which is because the temperature of the air is lower
than the desiccant at the beginning, and then with the progress of
the dehumidification, the latent heat release dominates the process
so that the air temperature increases. From the figure, it can be
seen that the outlet temperature of the air in the internally-cooled
dehumidifier is higher than that in the adiabatic one, although
the difference between them is very small, only 0.3 ◦ C, the rea- Fig. 7. Effect of the channel II width on the regeneration with different air flow
son for which is that in the internally-cooled dehumidifier more rates.

mass transfer causes more latent heat release from the dehumid-
ification, and as a result the air would gain more heat to be with mance goes down with increase of the channel width, which is
higher outlet temperature. The outlet temperature difference be- the result from that smaller width of the channel II can pro-
tween both dehumidifiers greatly depends on the Lewis number vide higher air velocity and mass transfer coefficient. However,
and operation conditions. During our experiments, the same fact too small width of the channel would bring another problem,
happened under some operation conditions, which was that the which is too high pressure resistance in the channel and more en-
outlet air temperature in the internally cooled dehumidifier was ergy consumption of pumps. Fig. 7 shows the effects of the air
slightly higher than that in the adiabatic one. In above computa- channel width on the regeneration with three different air flow
tion for the dehumidification, the Lewis number was assumed as rates (0.006, 0.012 and 0.018 kg/s) under the operation conditions
1.5. In order to check the behavior of effect of Lewis number on shown in Table 3. The results show that higher air flow rate can
the process, Fig. 5 demonstrates outlet temperature and humidity produce better regeneration performance, for better regeneration
ratio of the air at different Lewis numbers in both dehumidifiers. performance it is recommended that the width of the air channel
The results indicate that the Lewis number has little influence on should be less than 1.5 cm, however pressure resistance problem
outlet humidity ratio of the air, and however has a little influence should also be considered here as in the optimal design of the
on outlet temperature of air since the outlet temperature of air dehumidifier. Hence, the designs of the internally cooled/heated
changes less than 1 ◦ C with the obvious change of Lewis number. dehumidifier/regenerator need to consider roundly the dimension
The results also show that the outlet temperature of the air in the size, thermal performance and air pressure resistance.
internally-cooled dehumidifier is slightly higher than that of the
adiabatic one. 4. Conclusions
Usually the internally cooled/heated dehumidifier/regeneration
expected is with suitable dimensions and good thermal perfor- The idea of internal heat/sink source was put forward in order
mance. The width of the heat and mass transfer channel is an to keep high temperature of desiccant solution in the regenera-
important parameter for optimal design of this kind of dehumid- tor and lower temperature of the desiccant solution in the de-
ifier. Fig. 6 indicates the effects of the width of channel between humidifier for better mass transfer performance compared with
air and desiccant with three different air flow rates (0.0076, 0.0151 traditional packing, and a unitized model for the internally-cooled
and 0.0227 kg/s) under the typical operation conditions shown dehumidifier and internally-heated regenerator was presented. By
in Table 3. In this case, for better dehumidification performance utilizing experimental data, the correlations of mass transfer co-
it is recommended that the width of the channel should be less efficient between air and desiccant were developed for the dehu-
than 2 cm. The results show that the dehumidification perfor- midification and regeneration, which indicated the effect of tem-
 Y. Yin et al. / International Journal of Thermal Sciences 48 (2009) 1664–1671 1671

perature and flow rate of desiccant on mass transfer coefficients. [5] P. Gandhidasan, Prediction of pressure drop in a packed bed dehumidifier oper-
In addition, model validation was carried out by comparing the ating with liquid desiccant, Applied Thermal Engineering 22 (2002) 1117–1127.
[6] D.I. Stevens, J.E. Braun, S.A. Klein, An effectiveness model of liquid desiccant
computation results with experimental data. It was found that the
system heat/mass exchangers, Solar Energy 42 (1989) 449–550.
errors were within 5% and showed acceptable accuracy. Whereas [7] V. Martin, D.Y. Goswami, Effectiveness of heat and mass transfer processes in
further study is still necessary to develop more comprehensive a packed bed liquid desiccant dehumidifier/regenerator, HVAC&R Research 6
mass transfer coefficient correlations under wider range experi- (2000) 21–39.
mental conditions and deal more accurately with the model as- [8] T.W. Chung, H. Wu, Comparison between spray towers with and without fin
coil for air dehumidification using triethylene glycol solutions and development
sumptions, such as actual Lewis number, definite description of
of the mass transfer correlations, Ind. Eng. Chem. Res. 39 (2000) 2076–2084.
wetness. Anyhow, the present study highlights the importance of [9] X.H. Liu, Y. Zhang, K.Y. Qu, Y. Jiang, Experimental study on mass transfer per-
actual Lewis number and wetness, which should be paid more at- formances of cross-flow dehumidifier using liquid desiccant, Energy Conversion
tention in future study. and Management 47 (2006) 2682–2692.
A case of the internally cooled/heated dehumidifier/regenerator [10] Yonggao Yin, Xiaosong Zhang, Zhenqian Chen, Experimental study on dehu-
midifier and regenerator of liquid desiccant cooling air conditioning system,
using the plate exchanger with the parallel flow was studied nu- Building and Environment 42 (2007) 2505–2511.
merically to discuss the heat and mass transfer behavior between [11] Yonggao Yin, Xiaosong Zhang, A new method for evaluating coupled heat and
air and desiccant. The results show that the internally-heated re- mass transfer coefficients between air and liquid desiccant, International Jour-
generator can avoid the dehumidification possibility which would nal of Heat and Mass Transfer 51 (2008) 3287–3297.
[12] Chengqin Ren, Yi Jiang, Yinping Zhang, Simplified analysis of coupled heat and
happen in adiabatic one, and in addition it could offer higher re-
mass transfer processes in packed bed liquid desiccant–air contact system, So-
generation efficiency than the adiabatic one to produce better en- lar Energy 80 (2006) 1221–1231.
ergy efficiency. The internally-cooled dehumidifier can also provide [13] A.Y. Khan, Cooling and dehumidification analysis of an internally cooled liquid
better dehumidification performance comparing with the adiabatic desiccant absorber, Applied Thermal Energy 18 (1998) 265–281.
one; however its benefit would be not as good as the internally- [14] A.Y. Khan, J.L. Martinez, Modeling and parametric analysis of heat and mass
transfer performance of a hybrid liquid desiccant absorber, Energy Conversion
heated regenerator. Effect of the air channel width on the dehu-
and Management 39 (1998) 1095–1112.
midifier and regenerator was discussed and reasonable width is [15] H.W. Coleman, F. Stern, Uncertainties in CFD validation, Journal of Fluids Engi-
recommended to optimize the design of this kind of dehumidifier neering 119 (1997) 795–803.
and regenerator. [16] J.R. Taylor, An Introduction to Error Analysis—The Study of Uncertainties in
Physical Measurements, University Science Books, California, 1982.
[17] S. Jain, P.L. Dhar, S.C. Kaushik, Experimental studies on the dehumidifier and
Acknowledgements
regenerator of a liquid desiccant cooling system, Applied Thermal Engineer-
ing 20 (2002) 253–267.
The authors would like to thank the supports by National Natu- [18] W.Y. Saman, S. Alizadeh, An experimental study of a cross-flow type plate heat
ral Science Foundation of China (No. 50676018), China Scholarship exchanger for dehumidification/cooling, Solar Energy 73 (2002) 59–71.
Council and Foundation for Excellent Doctoral Dissertation of the [19] G. Grossman, Simultaneous heat and mass transfer in film absorption under
laminar flow, International Journal of Heat and Mass Transfer 26 (1983) 357–
Southeast University in China. In addition, the constructive com-
371.
ments of the reviewers and editor are appreciated greatly. [20] A. Ali, K. Vafai, A.R.A. Khale, Analysis of heat and mass transfer between air
and falling film in a cross flow configuration, International Journal of Heat and
References Mass Transfer 47 (2004) 743–755.
[21] M. Park, J.R. Howell, G.C. Vliet, J. Peterson, Numerical and experimental results
[1] K. Gommed, G. Grossman, Experimental investigation of a LiCl–water open ab- for coupled heat and mass transfer between a desiccant film and air in cross-
sorption system for cooling and dehumidification, Trans. of ASME, Journal of flow, International Journal of Heat and Mass Transfer 37 (1994) 395–402.
Solar Energy Engineering 126 (2004) 710–715. [22] Y.J. Dai, H.F. Zhang, Numerical simulation and theoretical analysis of heat and
[2] Y.J. Dai, R.Z. Wang, H.F. Zhang, et al., Use of liquid desiccant cooling to improve mass transfer in a cross flow liquid desiccant air dehumidifier packed with
the performance of vapor compression air conditioning, Applied Thermal En- honeycomb paper, Energy Conversion and Management 45 (2004) 1343–1356.
ergy 21 (2001) 1185–1202. [23] M.R. Conde, Properties of aqueous solutions of lithium and calcium chlorides:
[3] L.Z. Zhang, J.L. Niu, A pre-cooling Munters environmental control desiccant formulations for use in air conditioning equipment design, International Jour-
cooling cycle in combination with chilled-ceiling panels, Energy 28 (2003) nal of Thermal Science 43 (2004) 367–382.
275–292. [24] Yonggao Yin, Xiaosong Zhang, Gen Wang, Lei Luo, Experimental study on a new
[4] H.M. Factor, G. Grossman, A packed bed dehumidifier/regenerator for solar air- internally cooled/heated dehumidifier/regenerator of liquid desiccant systems,
conditioning with liquid desiccants, Solar Energy 24 (1980) 541–550. International Journal of Refrigeration 31 (2008) 857–866.