Pergamon P I I : S 0 0 3 8 – 0 9 2 X ( 0 0 ) 0 0 1 0 1 – 8 All rights reserved. Printed in Great Britain 0038-092X / 00 / $ - see front matter www.elsevier.com / locate / solener
DETERMINATION OF THE VOID FRACTION AND DRIFT VELOCITY IN A
TWO-PHASE FLOW WITH A BOILING SOLAR COLLECTOR
†
M. E. VIEIRA , P. O. O. DUARTE and H. L. B. BUARQUE
Applied Solar Energy Laboratory, Federal University of Ceara, Campus do Pici, 60455-760, Fortaleza, Brazil
Received 26 August 1998; revised version accepted 15 May 2000
Communicated by BRIAN NORTON
Abstract—This paper presents an approach to determine the void fraction and the drift velocity in a two-phase
flow with a boiling solar collector using easily obtained experimental data. The solar collector operates in a thermal siphon circuit, where the working fluid absorbs solar radiation mostly while boiling. The vapor bubbles release their latent heat in a condenser, while heating up a flow of water–glycol. Two numerical procedures are developed to calculate the void fraction because its experimental values cannot be easily measured. The use of a flow meter causes an additional pressure drop in the thermal siphon circuit and, consequently, changes the circulated mass flow rate. The first numerical procedure is based on a force balance in the thermal siphon loop and is used to estimate the total mass flow rate and the void fraction in the circuit. The second uses a drift flux correlation to estimate the void fraction and the drift velocity. Both procedures use the experimental values for the vapor mass flow rate, which is determined by an energy balance in the condenser. The volumetric flow rate of the water–glycol mixture and its temperature difference across the condenser are experimentally measured. The pipe length of the two-phase flow in the solar collector is experimentally determined using 44 thermocouples attached to the back of flow channels in the absorber plate. The results show that the two numerical models compare well and that either one can be used to estimate the void fraction in the two-phase flow solar circuit. 2000 Elsevier Science Ltd. All rights reserved.
1. INTRODUCTION 1988, 1989; Kaushika et al., 1982; Abramzon et
al., 1983; Neeper, 1985; Vieira, 1992) can be
Solar collectors operating in single-phase thermal
found in the literature about the operation of siphon loops, using either water mixtures or oils
boiling collectors in thermal siphon loops, in-as the working fluid, have found various
applica-formation about the void fraction at the collector tions in water heating, water desalination, and
exit, either experimental or numerical, was not solar cookers (Schwarzer and Krings, 1996;
found. Most models treat the boiling collector as a Hafner, 1993). These systems present the
advan-unit where the total mass in the collector starts tages of self-pumping and self-controlling, and the
boiling, that is, either sub-cooled or saturated low costs associated with maintenance and
opera-liquid enters the absorber plate, and either satu-tion. As disadvantages, a lower natural convection
rated or superheated vapor exits. This is not a heat-transfer coefficient in the flow channels is
correct assumption since only a small part of the found in comparison with the forced convection
mass flow rate boils and there is usually a two-coefficient. Boiling collectors in closed thermal
phase flow at the collector exit. This conclusion siphon loops present the advantages of
single-was shown by Price (1984), who developed two phase natural convection systems and, when well
models for use with TRNSYS (1983), considering designed, can operate at a lower absorber plate
the collector and condenser together as a single temperature due to the higher boiling heat-transfer
component. In the first model, it was assumed that coefficient. As disadvantages, these systems
re-saturated liquid entered the collector, re-saturated quire higher installation costs and tighter
con-vapor exited, and that there were no pressure struction, since most systems operate at low
losses in the connecting lines. In the second pressure.
model, these assumptions were removed. Using Even though various articles (Soin et al., 1979;
the homogeneous two-phase flow model and Al-Tamimi and Clark, 1983; El-Assy and Clark,
assuming that the mass flow rate of the working
† fluid through the collector was the same as
Author to whom correspondence should be addressed. Tel.:
through the condenser, a mass balance calculation
155-85-288-9599; fax: 155-85-288-9636; e-mail:
eugenia@ufc.br showed that the initial liquid length fraction in the
316 M. E. Vieira et al.
system should have been 0.05. For best per- second uses a drift flux correlation to estimate the void fraction and the drift velocity. Both pro-formance, this liquid length varies from 0.7 to 1.0
cedures use the experimental values for the vapor (Soin et al., 1979) and, therefore, the vapor
mass flow rate, which is determined by an energy quality, defined as the rate of the vapor mass rate
balance in the condenser. to the total mass rate, at the collector exit is much
less than 1.0.
The two-phase flow with a boiling circuit can
2. METHODOLOGY
be described using Fig. 1. In operation, liquid
This section is divided in two parts. The first refrigerant is fed from an accumulator tank to the
shows the development of the force balance collector by gravity. As it flows upwards through
model used to calculate the total mass rate and the the absorber plate channels, it is sensibly heated
void fraction through the system (Vieira et al., until boiling starts and vapor bubbles are formed.
1997). The experimentally measured vapor mass These bubbles rise, dragging liquid and initiating
rate and the pipe length of the two-phase flow in a two-phase flow. When the two-phase flow leaves
the collector are used in the calculation, and it is the collector, it moves through a standard T-type
considered that the two phases flow as the same connection where the liquid separates from the
velocity. The second part presents the drift flux vapor and returns to the collector inlet. The vapor
correlation (Chexal et al., 1991), which requires moves upward to the condenser, where it releases
knowledge of the vapor mass rate and the total its latent energy by heating a water–glycol flow
mass rate through the system. This correlation that circulates through this heat exchanger. Solar
emphasizes the relative motion of the two phases systems available in the market have the
con-instead of the separate flux of the liquid and vapor denser located right above the solar collector. This
phases individually. The relative motion is de-design was not used in the present work for two
termined by a few key parameters related to the reasons: first, some piping space was needed for
flux of each phase. instrumentation (pressure and temperature
sen-sors), and second, the results found can also be
2.1. Force balance model used when the condenser is installed inside the
To calculate the total mass rate and the void storage tank, avoiding the use of a water pump.
fraction, it is considered that the total frictional This paper presents an approach to estimate the
pressure drop equals the pressure gain due to the void fraction and the total mass flow rate in the
difference in gravity in the single phase and the two-phase flow thermal siphon system using
two-phase vertical parts of the system, easily obtained experimental data. Two numerical
procedures are developed to calculate the void g f
DPT1 DPT50 (1)
fraction because its experimental values cannot be easily measured. The use of a flow meter causes
Two models for the two-phase flow frictional an additional pressure drop in the thermal siphon
pressure drop (Owens, 1961; Marchaterre, 1961), circuit and, consequently, changes the circulated
based on the pressure drop for single-phase flow, mass flow rate. The first numerical procedure is
are used. To experimentally determine the pipe based on a force balance in the thermal siphon
lengths of liquid and two-phase flow, 44 ther-loop and is used to estimate the total mass flow
mocouples are mounted along the flow channels rate and the void fraction in the circuit. The
on the backside of the absorber plate. Fig. 2 shows the average wall temperature values at specific positions along the collector pipes. It is observed that from the inlet to about 1 / 3 of the flow channel, the working fluid is heated until boiling starts. In the boiling length, the tempera-ture remains approximately constant.
Using the pressure drop model presented by Marchaterre (1961) and Eq. (1), an expression for the total mass flow rate is expressed as,
s s 4 s
~
mT5a. (r 2 rl v)?Li?g?sinu ? p ?D ?rl
s 21 s 2
Fig. 1. Schematic view of the thermal siphon circuit. The full ?(128?ml) hLspf1(12a)?(rl)
line represents the working fluid loop and the dotted line the
s s 22 21
Fig. 2. Temperature along the flow channel on the back of the absorber plate.
~
where mT is the total mass flow, a, the void ties. A generic transient two-phase flow problem
fraction, u, is the angle of inclination, D, the can be presented either by the two-phase separate equivalent diameter of the tube, L , the length ofi model or by the drift flux model, depending upon tube in the two-phase flow, Lspf, the length of tube the degree of dynamic junction between the two in the single-phase flow (liquid), L , the lengthtpf phases. In the separate two-phase flow model,
of tube in the two-phase flow (liquid and vapor), each phase is considered separately and the
s s
rl, the saturated liquid density, rv, the saturated problem is formulated in terms of two groups of
s
vapor density, andml, the saturated liquid viscosi- conservation equations of mass, momentum, and
ty. energy for each phase. The use of two momentum
Eq. (2) has two unknowns, the total mass flow equations introduces considerable difficulties due and the void fraction. Using the definition of the to the mathematical complication and uncertain-void fraction and considering that the two phases ties in the specifications of the interaction terms flow at the same velocity, the total mass flow rate between the two phases. In the drift flux model,
is expressed as, one kinematics equation is used to specify the
relative motion between the phases. In other
s s 21
~ ~
mT5[11(12a) .rl. (arv) ] mv (3) words, the problem is simplified to five field
equations: two conservation of mass, two of ~
where m is the vapor mass flow rate.v
energy, and one momentum equation for the Solving Eqs. (2) and (3) simultaneously, an
mixture. The model that is used mostly employs expression that allows the calculation of the void
one conservation of energy equation. fraction can be written as,
In the drift flux model presented by Chexal et
s s 21 s al. (1991), the void fraction is estimated by a few
~
a 5[11(12a)r ?l (a rv) ]?mv?128?ml
key parameters, such as the superficial velocities
4 s s s 21
?[p ?D ?r ?l g?sinu ?(r 2 rl v)?L ]i of the liquid and vapor, bubble concentration
s 2 s parameter, and the slip velocity. The void fraction
?hLspf1(12a)?(rl) ?Ltpf?[(12a)?rl
is expressed as,
s 22
1a rv] j (4)
21
a 5Jg?[(Jg1J )f 1V ]g j (5)
2.2. Drift flux model
318 M. E. Vieira et al.
The superficial velocity for the vapor, Jg, and gained by the water–glycol flow in the condenser the superficial velocity for the liquid, J , aref to the working fluid’s latent heat.
defined as, A 1.22-m long sight glass is installed to
indicate the liquid level in the collector when
21
Jg5Wg?(r ?g A) (6) charging the system. The solar collector tilt angle
is 458 and the diameter of the copper piping is
21 12.7 mm. The working fluid circuit (solar
collec-Jf5Wf?(r ?f A) (7)
tor, copper piping, and condenser) is evacuated and then completely filled with liquid refrigerant where W and W are the liquid and vapor massf g
to a liquid level of 1. Ambient data are also flow rates, rf and rg are the liquid and vapor
obtained from the meteorological station. The densities, and A is the total flow area.
instruments are scanned every 20 s and are The bubble concentration parameter, C , for ao
averaged at each minute. two-phase flow mixture flowing through an
in-clined pipe, by angle u, is estimated using the
parameters for the vertical and horizontal direc- 4. RESULTS
tion, Cov and C .oh
Fig. 3 shows the values for the void fraction
Co5F Cr ov1s12F Crd oh (8) calculated using the two procedures. The void
fraction values are plotted throughout the day
where F is a flux orientation parameter.r (morning and afternoon hours). When the solar
The drift velocity, V , for a parallel up-flowg j radiation intensity is low, these values are also
parallel and inclination angle 08,u ,908 is de- low, since less vapor bubbles are formed. At
fined as, higher solar radiation intensity values, the void
fraction and the bubble concentration parameter Vg j5F Vr g jv1(12Fr) Vg jh (9) are higher. It is seen that the void fraction values
calculated using the drift flux model are lower where Vg jv and Vg jh are the drift velocities for a than those of the force balance model because the horizontal and vertical flow, respectively.
liquid phase does not flow at the same velocity as the vapor phase, as assumed in the latest model. However, the difference remains approximately
3. EXPERIMENTAL
constant throughout the day.
Fig. 4 shows the drift velocity, V , throughout
Experimental measurements are performed to g j
the day. At low solar radiation intensity, the vapor determine the incident solar radiation on
horizon-bubbles flow at a higher velocity than the liquid tal and tilted planes, the temperature of the
phase. A smaller number of bubbles implies that absorber plate and of the outside walls of the
there is less surface area between the two phases copper piping, the working fluid pressure and
and, consequently, less drag. At high solar radia-temperature at the collector inlet, and the pressure
tion intensity values, there is more surface area at the outlet. Incident radiation measurements are
between the two phases, which flow at closer made with two precision pyranometers,
tempera-velocity values. ture transducers (60.58C or 4.0%, whichever was
greater) are copper–constantan, special limit type-T thermocouples, and pressure values are mea-sured with a diaphragm transducer. The pipe length of the two-phase flow in the solar collector is experimentally determined using 44 thermocou-ples attached to the back of flow channels in the absorber plate.
On the storage-tank side, the volumetric flow rate through the condenser and the temperature difference between the inlet and outlet are mea-sured. A five-junction thermopile (61% of the reading10.058C) is used to measure this tempera-ture difference across the condenser, and a tur-bine-type meter measures the volumetric flow of
the water–glycol solution. The vapor mass rate is Fig. 3. Void fraction (drift flux and force balance models) versus solar radiation intensity throughout the day.
REFERENCES
Abramzon B., Yaron I. and Borde I. (1983) An analysis of a flat-plate solar collector with internal boiling. J. Solar
Energy Engineering 105, 454–460.
El-Assy A. Y. and Clark J. A. (1988) Thermal analysis of a flat-plate collector in multiphase flows including superheat.
Solar Energy 40, 345–361.
El-Assy A. Y. and Clark J. A. (1989) Thermal analysis of a flat-plate boiling collector having sub-cooled inlet and saturated exit states. Solar Energy 42, 121–132.
Al-Tamimi A. I. and Clark J. A. (1983) Thermal analysis of a solar collector containing a boiling fluid. Progress in Solar
Energy 6, 319–324.
Chexal B., Lellouche G., Horowitz J., Healzer J. and Oh S. (1991) The Chexal-Lellouche Void Fraction Correlation for
Generalized Applications, NSAC-139, USA.
Hafner B. (1993) Modellierung und Optimierung eines solar Fig. 4. Drift velocity, V , versus solar radiation intensityg j
¨
betrieben Prozeßwarmesystems, Verlag Shaker, Aachen.
throughout the day for a solar collector at an inclination angle
Kaushika N. D., Bharadwaj S. C. and Kaushik S. C. (1982) of 458. Analysis of a flat plate collector with fluid undergoing phase
change. Applied Energy 11, 233–242.
Marchaterre J. F. (1961) Two-phase frictional pressure drop prediction from Levy’s momentum model. Trans. ASME,
Series C. Journal of Heat Transfer, 503–505. 5. CONCLUSION
Neeper D. A. (1985) Efficiency of a solar collector with internal boiling. ASHRAE Trans. 93(v.1.), 91–99.
The void fraction values calculated using the
Price, H. W. (1984) Analysis and Modeling of Boiling Fluid
drift flux correlation are lower than those calcu- Solar Collector, M.S. Thesis, University of Wisconsin, Madison.
lated using the force balance model. The
differ-Owens W. L. (1961) Two-phase pressure gradient. In ASME
ence between these values at each point is about International Developments in Heat Transfer, Part II, pp.
12 to 13%, and this small difference gives some 363–368.
Schwarzer K. and Krings T. (1996) Demonstrations- und
assurance about the calculated results. Fig. 3
¨
Feldtest von Solarkochern mit temporarem Speicher in shows that, throughout the day, the void fraction Indien un Mali, Shaker Verlag, Germany.
values vary from 0.12 to 0.4 (drift flux correla- Soin R. S., Sangameswar Rao K., Rao D. P. and Rao K. S. (1979) Performance of flat plate solar collector with fluid
tion). This is a reasonable result considering the
undergoing phase change. Solar Energy 23, 69–73.
low rates of heat flux in solar radiation (0.9 TRNSYS (1983) A Transient Simulation Program,
Engineer-2
kW/ m ), and the low values for the wall tempera- ing Experiment Station Report 38-12. University of Wiscon-sin, Madison.
ture, as presented in Fig. 2. Using the drift flux
Vieira, M. E. (1992) Experimental and Analytical Study of a
model, other important parameters are estimated. Boiling Collector in Thermal Siphon Operation, Ph.D. The superficial liquid and vapor velocities, the Thesis, C.E. Department, Colorado State University, Fort
Collins, CO.
drift velocity, and the bubble concentration
pa-Vieira, M. E., Buarque, H. L. B., Duarte, P. O. O. (1997) Mass
rameter can be used to study the two-phase flow Flow Rate in a Two-Phase Flow with Boiling Thermal and to design absorber plates for boiling collec- Siphon Loop, XIV Brazilian Congress of Mechanical
En-˜