Available online at www.sciencedirect.

com

Solar Energy 82 (2008) 1209–1219

www.elsevier.com/locate/solener

Potential of solar heat pipe vacuum collectors in the desiccant

cooling process: Modelling and experimental results
P. Bourdoukan a,*, E. Wurtz b, P. Joubert a, M. Spérandio a
a
LEPTAB Laboratoire d’Etude des Phénomènes de Transfert Appliqué au Bâtiment, Université La Rochelle, Avenue Marillac, 17042 La Rochelle, France
b
LOCIE Laboratoire Optimisation de la Conception et Ingénierie de l’Environnement, Campus Scientifique Université de Savoie,
73376 Le Bourget du Lac, France

Received 10 August 2007; received in revised form 6 June 2008; accepted 11 June 2008
Available online 11 July 2008

Communicated by: Associate Editor P. Gandhidasan

Abstract

Desiccant cooling is an alternative technique to vapour compression systems. When thermally driven at moderate temperatures, it can
be coupled to solar collectors. The use of flat-plate collectors and air collectors has demonstrated low efficiency in the coupling process
and so a low potential of solar energy use in desiccant cooling. In this paper the use of heat pipe vacuum tube (HPVT) collectors in a
solar desiccant cooling set up is investigated. First, a model for the collectors is proposed and experimentally validated under various
operating conditions. A model of the storage tank taking into account thermal stratification is also validated. The experimentally eval-
uated efficiency of the HPVT collectors for one operating day varies between 0.6 and 0.7. Finally, simulation of the solar desiccant plant
cooling a building is performed for different climates over a summer season. The solar fraction and the overall efficiency of the solar plant
are calculated for this period and the potential of the vacuum tube collectors is evaluated for application to the desiccant cooling process.
Ó 2008 Elsevier Ltd. All rights reserved.

Keywords: Vacuum solar collectors; Desiccant cooling; Simulation; Experimental set up; Overall efficiency

1. Introduction another cooling stage by an evaporative process (4).

Solar energy is used for space heating, hot water pro-
duction and thermally driven air-conditioning systems
including desiccant cooling systems. This air-conditioning
technique would be an alternative to vapour compression
systems (Jurinak et al., 1984); it has a low environmental
impact since water is used as refrigerant and electrical
consumption is limited to the auxiliaries. A solar desic-
cant cooling plant is shown in Fig. 1. It is a thermally dri-
ven open cooling cycle based on evaporative cooling and
adsorption. With reference to Fig. 1, the cycle operates as
follows: first, outside air (1) is dehumidified in a desiccant
wheel (2); it is then cooled in the sensible rotary heat
exchanger (3) by the return, cooled, air before undergoing
*
Corresponding author. Tel.: +33 5 46458622; fax: +33 5 46458241.
E-mail address: paul.bourdoukan@univ-lr.fr (P. Bourdoukan).
Finally, it is introduced in the room (5). The operating
sequence for the return air (6) is as follows: it is first
cooled to its saturation temperature by evaporative cool-
ing (7) and then it cools the fresh air in the rotary heat
exchanger (8). It is then heated in the heat exchanger
fed by the solar energy tank (9) and if needed by a
backup (10) and finally regenerates the desiccant wheel
by removing the humidity before exiting the installation.
Depending on the outside conditions, the air handling
unit operates under two main modes: indirect evaporative
cooling (IEV) for small cooling loads (return air humidi-
fier and rotary heat exchanger) or desiccant mode (com-
plete operating system, including the regeneration heat
exchanger) for higher loads. For moderate summer condi-
tions, in the morning when the outside temperature is
low, the indirect evaporative cooling mode is able to keep
the room in a comfortable range; there is thus no need for

0038-092X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2008.06.003
1210 P. Bourdoukan et al. / Solar Energy 82 (2008) 1209–1219

Nomenclature

Symbols z coordinate in the fluid direction (m)

AHU air handling unit e emissivity (dimensionless)
C heat capacity (J kg1 K1) k conductivity (W K1 m1)
E energy (J) g efficiency (dimensionless)
e thickness (m) sa transmission-absorptance coefficient
G solar global radiation (W m2) r Stefan Boltzmann constant (W K4 m2)
OE overall efficiency (dimensionless) q density (kg m3)
H heat transfer coefficient (W K1 m2)
IEV indirect evaporative cooling Subscripts
K heat loss constant (W K1 m2) a ambiant
L latent heat of vaporization (J kg1) b buffer
m mass flow rate (kg s1) c condenser
M mass (kg) f storage fluid in the manifold
p power (W) g glass
RH relative humidity (%) H heat pipe
S area (m2) i inlet
SF solar fraction (dimensionless) p plate absorber
T temperature (K) o outlet
u fluid velocity (m s1) r return from regeneration
UA global heat exchange coefficient (W K1) reg regeneration
V volume (m3) sat saturation
W humidity ratio (gwater vapor kg1dry air)

regeneration and the solar heated water can be stored in
the tank (Stabat, 2003; Maalouf, 2006). During the day
with the rising outside temperature and increasing solar
gains the indirect evaporative cooling cannot provide
the cooling and so the desiccant mode is then required,
while solar energy is needed for the regeneration of the
desiccant wheel. The minimum temperature required for
regeneration depends on the nature of the desiccant mate-
rial. It varies from 50 °C for lithium chloride to 60 °C for
silica gel. We chose silica gel for its higher dehumidifica-
tion performances in spite of the higher temperature
needed for its regeneration.

Fig. 1. Solar desiccant cooling installation.

 P. Bourdoukan et al. / Solar Energy 82 (2008) 1209–1219 1211

The desiccant cooling process is well-suited to the 2. Modelling of the solar installation
requirements of non-residential buildings with high occu-
pancy needing high air exchange rates, e.g. seminar rooms In order to study the potential of HPVT collectors in the
and banks (Henning et al., 2001). In these buildings the solar desiccant cooling process, a model of the above
rooms are usually occupied during the day, so air-condi- described solar installation was developed.
tioning loads match solar energy availability, so coupling
desiccant cooling with solar collectors would seem a very 2.1. Heat pipe vacuum tube
interesting option but it is important to investigate the effi-
ciency of this coupling by studying the potential of solar In an evacuated heat pipe collector (Fig. 2), a sealed
energy in desiccant cooling. copper pipe containing a vaporizable fluid is bonded to a
Numerical models are an effective way of predicting the copper fin plate absorber located inside a glass tube. A
thermal behaviour of solar collectors and thus the potential small copper condenser is attached from one side to the
of solar energy in this process. Steady state models have top of the heat pipe and from the other side to the storage
certain advantages: they are simple, require little comput- working fluid. The heat pipe is an evaporating-condensing
ing time and are easily coupled to an installation model. device. As the sun shines on the absorber, the pipe is heated
On the other hand, these simple models generally have and some of the liquid inside evaporates. The vapour rises
the drawback of overestimating the solar potential, since toward the condenser at the top of the heat pipe and con-
the collectors do not usually operate in a steady state denses on being cooled by the storage water circulating in
regime due to the variability of the driving factors (Schnie- the manifold. The liquid then returns to the heat pipe.
ders, 1997; Isakson and Eriksson, 1991). In these cases, The vacuum tube ensures minimization of the heat losses
dynamic models are better at describing the collectors of the collector.
and at investigating the potential, as well as the control A model for the HPVT is proposed below, separately
strategy, of the process using solar energy (De Ron, considering each component of the tube, i.e. the glass
1979; Perers, 1993; Henning, 1995). In recent decades col- cover, the absorber, the heat pipe fluid, the condenser
lector modelling and the investigation of their potential and the storage fluid. The following assumptions are made
have aroused increasing interest. regarding the model:
Kamminga (1985) proposed a dynamic model by con-
sidering each part of a plate collector separately, i.e. the  The properties of the materials are independent of the
absorber, the glass cover and the fluid. Based on this model temperature.
Schnieders (1997) proposed a model for direct flow vacuum  The temperature gradient along the absorber and the
plate collectors, and compared the energy-saving perfor- condenser is negligible.
mance of five models. Praene et al. (2005) applied Schnie-  Due to the sufficient quantity of fluid in the heat pipe
ders’ model to direct flow vacuum tube collectors, more (more than 90% of the heat pipe volume), the vapour
accurately taking into account the heat transfer by radia- is not superheated (Gidas, 1971; Bricard and Chaudo-
tion inside the tube. urne, 1997).
Henning et al. (2001) investigated the potential of flat-  The liquid returning from the condenser to the heat pipe
plate collectors and air collectors for the summer season is saturated.
in desiccant cooling using simulation and experimental
results. Wurtz and Maalouf (2006) compared the operation
of autonomous and solar-assisted desiccant cooling systems
using a model based on the Hotel–Whillier–Bliss equation.
Few works concerning the modelling and the potential
of an evacuated heat pipe vacuum tube collector using an
evaporating-condensing device have been published. And
yet these collectors are very well-suited for high tempera-
ture processes such as thermally driven air-conditioning
systems. Due to vacuum, their losses are limited and the
tube can operate at a high efficiency even for elevated inlet
water temperatures. In this paper a dynamic model of these
collectors and also of a stratified tank is proposed and val-
idated experimentally under various operating conditions.
The experimental efficiency of the HPVT collectors in des-
iccant cooling operations is calculated and the potential of
these collectors for desiccant cooling applications is evalu-
ated for three different locations (varying from hot and
humid to moderately humid) over a summer season and
for typical summer days. Fig. 2. Heat pipe vacuum tube.
1212 P. Bourdoukan et al. / Solar Energy 82 (2008) 1209–1219

 The glass cover is clean and completely transparent to dT n kS i

qn V n C f ¼ m1 C f ðT n1  T n Þ þ ðT n1  T n Þ
solar radiation. dt en;n
 The part of radiation reflected by the absorber leaves the þ m2 C f ðT i2  T n Þ þ S n K n ðT a  T n Þ ð7Þ
tube.
 Due to the vacuum, convection does not occur inside the dT i kS i
qi V i C f ¼ m1 C f ðT i1  T i Þ þ ðT i1  T i Þ
tube. dt ei
 Conduction in the storage fluid direction (z) is consid- þ m2 C f ðT iþ1  T i Þ þ S i K i ðT a  T i Þ
ered to be negligible. kS i
 ðT i  T iþ1 Þ ð8Þ
ei
The glass cover exchanges heat by convection with the
outside air and by radiation with the sky and the absorber. For the heat exchanger model, a counterflow configura-
The absorber receives solar radiation and heats the fluid tion and an NTU efficiency model were considered (Kays
inside the tube. The vapour rising from the heat pipe enters and London,1984; Incropera and Dewitt, 1996). The trans-
the condenser and releases energy to the circulating storage fer coefficient (UA) for the heat exchangers was evaluated
fluid, then exits the condenser as a saturated liquid. The from the experimental installation.
heat pipe liquid is heated to saturation before the evapora- The whole set of equations describing the solar instal-
tion process begins. lation (Eqs. (1)–(8)) is introduced into SPARK (SPARK,
The governing equations for each component (charac- 2003), a general simulation environment that supports the
terized by heat capacity C and temperature T) then read definition of simulation models and permits the solution
Glass cover: of these models via a robust and efficient differential/alge-
braic equation solver (Sowell and Haves, 2001). In the
dT g SPARK environment, the modeller describes the set of
M gCg ¼ eg rS g ðT 4sky  T 4g Þ þ ep rS p ðT 4p  T 4g Þ
dt equations defining a model as equation-based objects.
þ S g hg ðT a  T g Þ ð1Þ At the lowest level, an atomic object characterizes an
equation and its variables. Then macroscopic objects
Absorber: can be created as an assembly of various atomic or mac-
roscopic objects. The whole model is built by connecting
dT p the different elementary objects. Any class of objects can
M pCp ¼ ep rS p ðT 4g  T 4p Þ þ GsaS p þ S H hH ðT sat  T p Þ
dt be reused as many times as necessary, without any addi-
ð2Þ tional effort.
At this stage, we should note that a model imple-
Vapour flow rate: mented in the SPARK environment is input/output-free.
So the particular problem to be solved is described by
mv L ¼ S H hH ðT p  T sat Þ ð3Þ imposing adequate input data (boundary and initial con-
ditions) and by specifying the variables that are to be
Condenser: solved. In fact, it is not necessary to order the equations
  or to express them through assignment statements (algo-
dT c
M c Cc ¼ mv L þ S c hc ðT f  T c Þ ð4Þ rithms), unlike in conventional modular environments
dt
(Wurtz et al., 2006).
SPARK uses a mathematical graph of the model in
Storage fluid:
order to break it down into it in strong components to
 
oT f oT f be solved independently. Within each component, SPARK
m1 C f þu ¼ S c hc ðT c  T f Þ ð5Þ finds the appropriate calling sequence for obtaining the
ot oz
solution. If a direct sequence is not possible, as indicated
by a cyclic problem graph, a small ‘‘cut set” is determined
to minimize the number of variables involved in the New-
2.2. Storage tank ton-like iterative numerical solution process. As a result,
this decreases the size of the Jacobian matrix involved in
The investigated solar installation is shown in Fig. 1. In the Newton iteration within the component. Consequently,
order to take into account thermal stratification in the stor- the way SPARK handles the solution of coupled nonlinear
age tank, it is divided into n nodes (Klein, 1976). The fol- equations makes it a fast and robust solver for the simula-
lowing equations apply to the first node, the last node tion of energy problems in buildings.
(n = 12) and the ith current node: After the ‘‘cut sets” of variables have been identified, the
problem specification file is automatically converted into a
dT 1 kS i
q1 V 1 C f ¼ m1 C f ðT i1  T 1 Þ þ ðT 2  T 1 Þ C++ program which is then compiled, linked and executed
dt e12 for solving the problem for a given boundary and initial
þ m2 C f ðT 2  T 1 Þ þ S 1 K 1 ðT a  T 1 Þ ð6Þ conditions set.
 P. Bourdoukan et al. / Solar Energy 82 (2008) 1209–1219
3. Validation of the model
3.1. Experimental setup
An experimental solar installation (Fig. 3) similar to
that shown in Fig. 1 was set up and used to validate the
solar installation model. The installation consists of
40 m2 of heat pipe vacuum collectors (200 tubes), a 2500 l
storage tank with reinforced insulation and a plate heat
exchanger. Temperatures and flow rates are measured at
the inlet and outlet of each component (solar collectors,
heat exchanger and storage tank). The overall solar radia-
tion is measured in the plane of the collectors.
3.2. Validation of the collector model
Three different recorded days (the data (T, m and G) is
that measured on the experimental installation in La
Rochelle Tsky is given by Meteo France for these recorded
days) were used to validate the collector model. The first
day is representative of typical summer conditions, the sec-
ond of atypical solar radiation and the third of desiccant
cooling load conditions. Once the solar radiation in the col-
lectors’ plane, the outside temperature, and the inlet tem-
perature were recorded, the computed and measured
collector outlet temperatures were compared. Fig. 5 plots
the calculated and measured collector outlet temperature
for the three studied days. Solar radiation is also plotted
on the figure. Comparison between the computed and the
measured temperatures for the typical summer day condi-
tions (day 1) shows the model’s high performance in pre-
dicting collector outlet temperature: the absolute error of
Fig. 3. The experimental solar installation at La Rochelle.
1213
the model varies between 0 °C and 0.7 °C, with a corre-
sponding maximum relative error of less than 1%. This
accuracy is due to the fact that each component is taken
into consideration by the model and the calculations are
performed in each vacuum tube simultaneously (see Fig. 4).
While collector outlet temperature can thus be predicted
accurately in normal radiation conditions, it is very impor-
tant to study the performance of the model for atypical
conditions. By comparing the calculated and the measured
outlet temperatures for the atypical conditions day (day 2),
when the overall solar radiation fluctuated constantly, it
can be seen that the model very accurately predicts the out-
let temperature for the whole day, whatever the amplitude
of the solar radiation fluctuations. The maximum absolute
error in these conditions is 1.8 °C, i.e. a relative error of
2%.
In these first two cases, the model correctly predicts the
collector outlet temperature under different radiation con-
ditions. The next step in the validation procedure is to
study the model’s response to a desiccant cooling load,
i.e. storage in the morning, with regeneration when the des-
iccant mode is enabled. This is obtained as follows, (with
reference to Fig. 1): in the morning, while the outside tem-
perature is below 30 °C, the desiccant mode is not required,
so the heated water is stored in the tank (m1 > 0 and
m2 = 0). When the outside temperature exceeds 30 °C, the
desiccant mode is needed and the outlet valve of the buffer
is opened (m2 > 0). Flow rate m2 is regulated to obtain the
desired regeneration power.
Under typical desiccant cooling load (day 3) there is
remarkable agreement between the predicted and the mea-
sured outlet temperatures, for both storage and regenera-
1214 P. Bourdoukan et al. / Solar Energy 82 (2008) 1209–1219

Solar global radiation -day 1- Solar global radiation -day 2- Solar global radiation -day 3-
1000 1200 1000
900 1100 900
1000
800 800
900
Radiation (W/m )

700

Radiation (W/m )
700

Radiation(W/m )
2

2
800

2
600 700 600

500 600 500

400 500 400

400
300 300
300
200 200
200
100 100 100

0 0 0
0 100 200 300 400 500 0 100 200 300 400 500 0 100 200 300 400 500
Time (min) Time(min) Time (min)

Collector outlet temperature -day 1- Collector outlet temperature -day 2- Collector outlet temperature -day 3-
90 100 90
Experimental Experimental Experimental
Model Model Model
80
90
Temperature(°C)
Temperature(°C)

Temperature(°C)
70 80

60 80

50 70
70
40

30 60 60
0 100 200 300 400 500 0 100 200 300 400 500 0 100 200 300 400 500
Time (min) Time(min) Time (min)

Fig. 4. Calculated and measured collector outlet temperature for different radiation and load conditions.

tion periods. The absolute error is less than 0.5 °C except is significantly below the temperature of the bottom of the
very early in the day, when it is 1.4 °C. buffer.
In both cases, the maximum error between the calcu-
lated and the measured temperatures of the tank is 2 °C
3.3. Validation of the tank model except during the transition period between storage only
and regeneration, where the error can reach 3 °C.
The temperature and the flow rates at the two inlets and The potential of the HPVT collectors in solar desiccant
outlets of the tanks are measured, so the calculated temper- cooling is evaluated in the following section.
atures at the top and the bottom of the buffer (node 1 and
node n) can be compared to the measured ones. 4. Potential of HPVT in desiccant cooling
Two different scenarios were used to validate the tank
model. The first concerns only storage (m1 > 0, m2 = 0) 4.1. Experimental efficiency of HPVT
and the second considers typical desiccant conditions, e.g.
storage in the morning and both storage and regeneration The efficiency of the HPVT is defined as the ratio
in the afternoon (m1 > 0, m2 > 0). between the energy delivered by the collectors and the solar
Fig. 5 shows a comparison between the calculated and energy they receive. With reference to Fig. 1, this efficiency
the measured temperatures at the top and the bottom of reads
the buffer for these two operating modes. From this figure,
Edelivered m1 C f ðT o  T i Þ
it can be seen that the tank model is able to take into g¼ ¼ ð9Þ
Ereceived G:S
account the thermal stratification occurring in the buffer.
When operating under storage only, the stratification The experimental efficiency of the HPVT collectors for the
remains fairly constant, but increases when storage and day of desiccant cooling operation is shown in Fig. 7. We
regeneration are combined. This may, at first, appear inco- see that the efficiency of the tubes varies between 0.6 and
herent, as more mixing occurs in the latter case. The fact 0.7. The higher value is obtained for the minimum temper-
that stratification is greater with the regeneration load is ature of the collectors. It can be seen that even at high
due to the return temperature from regeneration, Tr, which operating temperatures, the efficiency never falls below
 P. Bourdoukan et al. / Solar Energy 82 (2008) 1209–1219 1215

Storage Desiccant load

90 90

80 80

Temperature (°C)
Temperature (°C)

70 70

60 60

50 50

40 40

30 30

20 Top experimental 20 Top experimental

Top model Top model
10 Bottom experimental 10 Bottom experimental
Bottom model Bottom model
0 0
0 100 200 300 40 0 100 200 300 400
Time (min) Time (min)

Fig. 5. Calculated and measured temperature for the top and the bottom of the buffer under storage and desiccant load.

0.6. This represents the major advantage of vacuum collec- La Rochelle (France), Bolzano (Italy) and Berlin
tors compared to traditional flat-plate collectors, whose (Germany).
efficiency significantly decreases with increasing operating
temperature. 4.3. Potential for desiccant cooling applications

4.2. Overall efficiency and solar fraction 4.3.1. Simulation conditions

In order to study the potential of HPVT collectors for
An important property of a solar installation is its over- solar desiccant cooling applications, simulations are per-
all efficiency (OE), which is the ratio of the solar energy formed for different locations, of the solar desiccant cool-
consumption of the system for regeneration to the solar ing installation similar to that of the Fig. 1 that cools an
energy received by the collectors. By integrating the regen- office building of a total gross area of 1800 m2 and a con-
eration power and the solar power received over a particu- ditioned area of 1350 m2. It is occupied from 8 am to
lar period we obtain the energy consumed for regeneration 6 pm, 5 days a week. The structure is thermally insulated
(E reg,solar) and the energy received (Ereceived) and so the (U = 0.52 W/m2) and the windows are double glazed. The
overall efficiency can be calculated internal gains are lightings (15 W/m2), .equipments
Z (14 W/m2) and occupants (5 W/m2). The building is mod-
Ereg;solar ¼ P reg;solar  dt ð10Þ elled by zonal method (Wurtz et al., 2006). For the air han-
regeneration
dling unit, the desiccant wheel model is that proposed by
P reg;solar ¼ m2 C f ðT 1  T i2 Þ ð11Þ Maclaine-cross and Banks (1972) and improved by Stabat
Z
and Marchio (2008), for the sensible regenerator Kays and
Ereceived ¼ Greceived  dt ð12Þ
day
London (1984) model is considered while the humidifiers
Ereg;solar are modelled by constant wet bulb temperature. The loca-
OE ¼ ð13Þ tions chosen are La Rochelle in France (hot and humid
Ereceived
with a mean humidity ratio of 13.5 g/kg) Bolzano in Italy
Another important property of a solar installation coupled (moderately humid with a mean humidity ratio of 11.7 g/
to a thermal process is the solar fraction of the installation, kg) and Berlin in Germany (relatively dry in comparison
defined as the ratio of the energy provided by the collectors with La Rochelle and Bolzano with a mean humidity ratio
for regeneration to the total energy needed for regenera- of 8.7 g/kg). Locations with humidity over 14.5 g/kg are
tion, including the auxiliary. The solar fraction is expressed not considered since it becomes inefficient to control the
as humidity inside when the desiccant system is coupled to
Ereg;solar solar collectors since the needed regeneration temperature
SF ¼ ð14Þ tends to increase significantly (see Fig. 6).
Ereg;solar þ Ereg;backup
The control strategy is following (Bourdoukan et al.,
In the following section, the potential of HPVT collectors 2008): In the morning the system starts with the indirect
in solar desiccant cooling is evaluated by determining the evaporative cooling (IEV) mode (return humidifiers and
solar fraction and the overall efficiency of the collectors the sensible regenerator) and when the temperature of the
over a summer season and for typical summer days for room exceeds 26 °C or the humidity ratio exceeds the
1216 P. Bourdoukan et al. / Solar Energy 82 (2008) 1209–1219
Collector efficiency
1 90
0.9
Efficiency (dimensionless)
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1 Efficiency
Outlet temperature
0 40
0 100 200 300
Time(min)
Fig. 6. Experimental efficiency of the HPVT collectors for one day under
desiccant cooling load.
10.8 g/kg the system switches to the desiccant (DEC) mode
(all the components are operational). The temperature and
the humidity of the conditioned space are controlled by the
mean of the regeneration temperature. By increasing the
regeneration temperature the dehumidification capacity is
increased and thus more sensible cooling is possible by
increasing the efficiency of the supply humidifier. In rever-
sal to increase the latent capacity of the system, by increas-
ing the regeneration temperature and by maintaining the
same efficiency of the supply humidifier the sensible cooling
capacity is maintained and the latent capacity is increased
due the increase of the dehumidification capacity. If the
temperature of the water at the outlet of the buffer (T1 in
Fig. 1) is high and there is no need for high regeneration
temperature a part of the water flow rate is bypassed to
keep the regeneration temperature close to 60 °C. The
backup is used if the solar regeneration temperature (posi-
tion 9 in Fig. 1) falls below 60 °C which is the required
regeneration temperature for the silica gel, or if there is a
need to increase the regeneration temperature with the
increasing load.
4.3.2. Simulation results
The first step in the potential estimation is the evalua-
tion of the overall efficiency and the solar fraction of HPVT
installation coupled to a desiccant unit. Simulations of the
solar desiccant plant, cooling the described building pow-
ered by HPVT collectors for the three mentioned locations
are performed. Since the cooling load differs for a location
to another the collector area for each location depends on
this load so for La Rochelle the collector area is 300 m2, for
Bolzano it is 245 m2 and for Berlin 205 m2.
Fig. 7 shows the simulation results of the hottest days of
the season for the three locations. For each case is plotted
in the left series of Fig. 7 the outside temperature (T out-
side), the supply temperature (T5), the inside temperature
(T6), the inside relative humidity (RH6) and the operating
mode of the AHU (20 for IEV and 40 for the DEC). For
the right series of Fig. 7 are plotted the temperature at
80
the top of the buffer (T buffer), the temperature of the
regeneration air before the backup (T9), the temperature
Temperature (°C)
70 of the regeneration air after the backup (T10), the position
of the 3 way valve (valve position) and the fraction of the
backup (backup fraction).
60 Fig. 7 shows that for all the cases the temperature and
the humidity are both controlled and maintained at an
acceptable level (T < 26.5 and w < 11.8 g/kg), and the col-
50 lectors can provide the major part of the needed regenera-
tion energy.
For La Rochelle conditions (humidity ratio for the hot-
test day is 13.4 g/kg) the systems operates only one hour
400
in the early morning under the IEV mode since the sensi-
ble load is high with elevated outside temperature that
limits the use of the IEV beside the high outside humidity
ratio that reduces the potential of the IEV when elevating
the inside humidity by introducing outside air at high
humidity ratio. The system switches to the DEC mode
and is able to keep the inside temperature near to 25 °C
for the whole day and the relative humidity close to
55% (humidity ratio is then 11.7 g/kg). Even that the out-
side temperature reached 34 °C during the day the system
was able to provide the cooling load for the whole day.
The supply 3 way valve for the regeneration heat exchan-
ger was completely open since during the DEC mode with
constant need for regeneration and since the DEC mode
started early in the morning it limited the buffer tempera-
ture. The HPVT collectors were able to provide most of
the regeneration energy and the backup was operational
for limited periods during the day, first in the morning
when the collectors were not yet operational and during
the peak load in the early after noon. When the backup
was operational the fraction of the backup power was
always below 28%.
For this day in La Rochelle the calculated OE (Eqs.
(10)–(13)) is 63% and the SF (Eq. (14)) is 93%, while for
the whole summer season the calculated OE is 54% and
the SF is 96%. The difference in the OE on the day and sea-
son basis is that for the season basis takes into account the
received energy during weekend where the HPVT are not
efficiently used (only for storage) which lowers the OE
beside that during this particular day the DEC mode was
operational during 9 h which increases the efficiency with
the increase of the time of use (Bourdoukan et al., 2007).
The increase in the SF from the day to the season basis
is related to the fact that during the hottest days the backup
is needed more than the other days of the season.
For Bolzano (outside humidity ratio is 11.4 g/kg) in the
early morning the outside temperature was low and the
solar global radiation was not typical of a summer period
which gave a small cooling load in the beginning of the
day and the IEV was very efficient which delayed the
DEC mode to the afternoon. The inside temperature was
 P. Bourdoukan et al. / Solar Energy 82 (2008) 1209–1219 1217

Backup power fraction or valve position(%)

La Rochelle (France) -outside humidity ratio 13.4 g/kg- La Rochelle (France)- HPVT collectors
40 100 80

Relative humidity(%) or AHU mode

90 100
35
90
80
30 70 80
70
Temperature(°c)

Temperature(°c)
25 T1 buffer 70
60
T9 reg solar 60
20 50 60 T10 reg backup
backup fraction 50
40 valve position
15 40
T outside 30
50 30
10
T5 supply 20
T6 inside 20
5 RH6 inside 10 10
AHU mode 20=IEV, 40=DEC
0 0 40 0
8 9 10 11 12 13 14 15 16 17 18 8 9 10 11 12 13 14 15 16 17 18
Time (hours) Time (hours)

Backup power fraction or valve position(%)

Bolzano (Italy)-outside humidity ratio 11.4 g/kg- Bolzano (Italy)- HPVT collectors
35 100 90
T1 buffer

Relative humidity(%) or AHU mode

90 T9 reg solar 100
30 T10 reg backup
80 80 90
backup fraction
Temperature(°c)

Temperature(°c)
25 valve position 80
70
70
60 70
20
60
50
50
15
40 60
40
10 T outside 30
30
T5 supply 20 50
T6 inside 20
5
RH6 inside 10 10
AHU mode 20=IEV, 40=DEC
0 0 40 0
8 9 10 11 12 13 14 15 16 17 18 8 9 10 11 12 13 14 15 16 17 18
Time (hours) Time (hours)

Backup power fraction or valve position(%)

Berlin (Germany)-outside humdity ratio 9.5 g/kg- Berlin (Germany)- HPVT collectors
35 100 90
Relative humidity(%) or AHU mode

90 100
30
80 80 90
Temperature(°c)

Temperature(°c)

25 80
70
70
60 70
20
60
50
50
15
40 60
40
10 T outside 30 T1 buffer 30
T5 supply 20 50 T9 reg solar
T6 inside T10 reg backup 20
5
RH6 inside 10 backup fraction 10
AHU mode 20=IEV, 40=DEC valve position
0 0 40 0
8 9 10 11 12 13 14 15 16 17 18 8 9 10 11 12 13 14 15 16 17 18
Time (hours) Time (hours)

Fig. 7. Performance of HPVT collectors coupled to a desiccant system for three different locations.

kept below 26 °C for the whole day and the inside relative it was operational the fraction of the backup power was
humidity was fluctuating depending on the inside tempera- below 20% of the total needed regeneration power. For this
ture (but the inside humidity ratio was always below day the OE is 50% and the SF is 87% while for the season
11.8 g/kg). Since the DEC mode was delayed more storage the OE is 51% and the SF is 97%. For this day the OE is
was possible in the buffer which increased its temperature. relatively low due to the use over a short period of solar
In this case the supply 3 way valve of the regeneration heat energy due to the delay of the DEC mode, which reduces
exchanger was regulating the flow rate depending on the the OE (Bourdoukan et al., 2007).
load demand. For Bolzano and for this particular day For Berlin (outside humidity ratio is 9.5 g/kg) the sce-
the backup was needed only at the end of the day. When nario of the hottest day is similar to Bolzano. The DEC
1218 P. Bourdoukan et al. / Solar Energy 82 (2008) 1209–1219
was delayed due to the low cooling load in the morning and
the low outside humidity ratio. In this case the inside tem-
perature was below 25 °C for almost the whole day and the
humidity ratio below 11 g/kg. The 3 way valve was com-
pletely opened only on peak load at noon and at the very
end of the day with the decrease of the buffer temperature
which enabled the backup with a fraction below of 10%. In
this case the OE is 56% and the SF is 96.7%, while for the
season the OE is 51%.
For the studied cases the control strategy was able to
keep the temperature and the humidity below the comfort
constraint. For all the cases if the DEC mode is enabled
when the inside temperature reaches a value below 26 °C
(for example 24 °C instead of 26 °C), this will produce
more cooling during the day and the inside temperature
will be a bit lower but in reversal this strategy will consume
early the energy and so the backup will be used very early
in the afternoon which will decrease the SF significantly
with a negligible increase in the OE and in the thermal
comfort. Same, if the flow rate (at the outlet of the buffer)
for the regeneration heat exchanger is not controlled by the
3 way valve the regeneration temperature will be higher
and thus more cooling will be produced but the storage will
be consumed rapidly and the backup will be used early and
will decrease the SF.
For the three cases on a season basis the SF is greater
than 96%, and the OE is greater than 51%. Coupling the
HPVT collectors with a desiccant cooling plant thus
appears as a very interesting option.
4.4. Comparison with flat-plate collectors
To compare the performance of HPVT and flat-plate
collectors, simulations were carried on the same system
that cools the same building, this time powered by flat-
plate collectors. For the same day in La Rochelle consid-
ered in the previous paragraph, Fig. 8 shows the energetic
performance of the flat-plate collectors installation coupled
to the desiccant system. When using these collectors the
same inside conditions can be achieved but in this case
the backup is used for the whole day and the backup power
reaches 50% of the total regeneration power in the early
morning. The OE of flat-plate collectors for this day is
44% and the SF is 72%. For the season the OE is 36%
and SF is 75%. For Berlin and Bolzano the flat-plate collec-
tors operate with an OE of 33% and a SF 77%.
In order to attain the same performance as for the
HPVT collectors we must increase the flat collector’s area
by 20–25% which will cause a space constraint, especially
for multi-storey building. Beside the collectors which are
used for cooling during summer will be used for heating
during winter, or HPVT are more efficient in handling
the diffuse radiation and operates almost independently
of outside temperature due to vacuum while the losses of
flat-plate collectors increase significantly with the decrease
of outside temperature. For the investment issue the HPVT
are still more expensive than the flat-plate collectors but
Backup power fraction or valve position(%)
La Rochelle (France)- flat plate collectors
80
100
70
90
60
80
Temperature (°C)
50 70
T1 buffer
T9 reg solar 60
40 T10 reg backup
backup fraction 50
30 valve position 40
30
20
20
10
10
0 0
8 9 10 11 12 13 14 15 16 17 18
Time (hours)
Fig. 8. Performance of flat-plate collectors in desiccant cooling applica-
tion for La Rochelle (France) weather conditions.
since it is a new technology with the coming years the price
will decrease with the increasing demand.
This clearly indicates the importance and the advantage
of HPVT collectors in solar desiccant cooling.
5. Conclusion
In this paper, a model of heat pipe vacuum tubes using
an evaporating-condensing device was proposed and
implemented in the SPARK simulation environment. The
model has been validated for different conditions by com-
parison with an experimental solar installation and has
accurately predicted collector outlet temperature with a
maximum error of 2% in irregular solar conditions and
an error of less than 1% in normal conditions. A tank
model was also developed and validated for predicting
the temperature at the top and bottom of the buffer.
The experimentally observed efficiency of the collectors
for desiccant cooling conditions varies between 60% and
70%. Finally, the potential of heat pipe vacuum tube collec-
tors for a solar desiccant plant that cools an office building
was investigated. A simulation over a summer season
showed the capacity of a desiccant plant to maintain the
building temperature and humidity at a comfort level for
three different locations going from relatively dry to hot
and moderately humid. These simulations render a sea-
sonal overall efficiency greater than 51% and a seasonal
solar fraction of 96% while the daily OE varies from 50%
to 64% and the daily solar fraction varies from 87% to 97%.
Acknowledgements
This work was supported by ADEME (French Agency
for Environment and Energy Management) and the Regio-
nal Council of Poitou-Charentes. The authors would like
 P. Bourdoukan et al. / Solar Energy 82 (2008) 1209–1219 1219

to thank Mr Michel Burlot for his valuable technical sup- Bricard, A., Chaudourne, S., 1997. http://www.techniques-ingenieur.fr/
port on the experimental installation. dossier/caloducs/. Sciences et Techniques, T.I, ed.
De Ron, A.J., 1979. Dynamic modeling and verification of a flat-plate
solar collector. Solar Energy 24, 117–128.
Appendix A. A1-Values of the physical properties for each Gidas, N.K., 1971. Champs de vitesse et de temperature à l’intérieur d’un
vacuum tube caloduc. Revue Generale de Thermique 118, 843–863.
Henning, H.-M., 1995. A collector hardware simulator theoretical analysis
and experimental results. Solar Energy 55, 39–48.
Symbol Value Henning, H.-M., Erpenbeck, T., Hingenburg, C., Santamaria, I.S., 2001.
The potential of solar energy use in desiccant cooling cycles.
Mg (kg) 2.94 International Journal of Refrigeration 24, 220–229.
Cg (J kg1 K1) 815 Incropera, F., Dewitt, D.P., 1996. Fundamentals of Heat and Mass
eg (–) 0.9 Transfer. John Wiley & Sons, New York.
ep (–) 0.115 Isakson, P., Eriksson, L.O., 1991. A Dynamic Solar Collector Model for
Sg (m2) 0.793 TRNSYS. Nordic Solar Energy R&D Meeting, Borlange.
Jurinak, J.J., Mitchell, J.W., Beckman, W.A., 1984. Open-cycle desiccant
hg (W K1 m2) 4 air conditioning as an alternative to vapor compression cooling in
Cp (J kg1 K1) 390 residential applications. Journal of Solar Energy Engineering 106,
Mp (kg) 1.6465 252–260.
Sp (m2) 1.85 Kamminga, W., 1985. The approximate temperature within a flat plate
SH (m2) 0.0953 solar collector under transient conditions. International Journal of
Heat Mass Transfer 28, 433–440.
hH (W K1 m2) 41.8 Kays, W.M., London, A.L., 1984. Compact Heat Exchangers. McGraw-
MC (kg) 0.15 Hill, New York.
CC (J kg1 K1) 390 Klein, S.A., 1976. A Design Procedure for Solar Heating Systems. Ph.D.,
hC (W.K1 m2) 21.4 University of Wisconsin-Madison.
SC (m2) 0.015 Maalouf, C., 2006. Etude du potentiel de rafraichissement passif par
dessiccation avec régénération par panneauxsolaires. Ph.D., University
sa (–) 0.8835 of La Rochelle.
Maclaine-cross, I.L., Banks, P.J., 1972. Coupled heat and mass transfer in
regenerators – prediction using analogy with heat transfer. Interna-
tional Journal of Heat Mass Transfer 15, 1225–1242.
A2-. List of components of the La Rochelle solar installation Perers, B., 1993. Dynamic method for solar collector array testing and
evaluation with standard database and simulation programs. Solar
 40 m2 of heat pipe vacuum tube collectors. Energy 50, 517–526.
Praene, J., Garde, F., Lucas, F., 2005. Steady state model of a solar
 Plate heat exchanger. evacuated tube collector based on sensitivity analysis. In: ASME
 Storage tank (2500 l) with reinforced insulation. International Mechanical Engineering Congress and Exposition,
 Three water flow meters to measure m1, m1 and m2 Orlando, Florida USA.
(Fig. 1). Schnieders, J., 1997. Comparison of the energy yield predictions of
 7 Pt100 temperature sensors to measure the temperature stationary and dynamic solar collector models and the models’
accuracy in the description of a vacuum tube collector. Solar Energy
TO, Ti, T1, Ti1, Ti2, Tn, and Toutside (Fig. 1). 61, 179–190.
 Two radiation meters to measure the solar global radia- Sowell, E.F., Haves, P., 2001. Efficient solution strategies for building
tion G. energy system simulation. Energy and Buildings 33, 309–317.
SPARK, 2003. Simulation Problem Analysis and Research Kernel. LBNL,
California, Berkeley.
Stabat, P., 2003. Modélisation de composants de systèmes de climatisation
References mettant en oeuvre l’adsorption et l’evaporation d’eau. Ph.D., Ecole des
Mines, Paris.
Bourdoukan, P., Wurtz, E., SpÃÓrandio, M., Joubert, P., 2007. Global Stabat, P., Marchio, D., 2008. Heat and mass transfer modelled for rotary
efficiency of direct flow vacuum tube collectors in autonomous solar desiccant dehumidifiers. Applied Energy 85, 128–142.
desiccant cooling: simulation and experimental results. In: proceedings Wurtz, E., Haghighat, F., Mora, L., Mendoncßa, K.C., Zao, L., Maalouf,
of 10th International Building Performance Simulation Association C., Bourdoukan, P., 2006. An integrated zonal model to predict
IBPSA, vol. 2, Beijing, China, pp. 342–347. transient indoor humidity distribution. ASHRAE Transactions, 175–
Bourdoukan, P., Wurtz, E., Joubert, P., Spérandio, M., 2008. A control 186.
strategy to prevent the impact of outside and load conditions on the Wurtz, E., Maalouf, C., 2006. Analysis of the operation of a solar
hygro-thermal performance of a desiccant cooling system, In: desiccant cooling system. In: 61st ATI-National Congress – Interna-
Proceedings of Indoor Air, Copenhagen, Danemark. tional Session Solar Heating and Cooling, Perugia, Italy, pp. 57–62.