1、AN-04-1 -3 Two-Phase Refrigerant Distribution in Round Tube Manifolds Sivert Vist ABSTRACT Two-phase refrigerantflow distribution in round tube heat exchanger manifolds has been investigated. Experimental data have been acquired in a heat exchanger test rig specially made for measurement of mass flo
2、w and phase distribution in the manifolds of compact heat exchangers. Horizontal round tube manifolds with 8 and 16 mm inner diameter and ten upward oriented heat exchanger tubes were used in the experiments, und CO, und HFC-134a were used as refrigerants. The phase distribution results show that th
3、e vapor phase is preferentially distributed to the$rst heat exchanger tubes, while most of the liquidphase leaves through the tubes at the end ofthe manifold. The experimental data are compared to existing semi-empir- ical models for phase split in Tjunctions. INTRODUCTION Uneven two-phase distribut
4、ion reduces the thermal performance of compact heat exchangers with parallel flow- circuits. Due to the separation of the two phases in the mani- folds, the vapor and the liquid are distributed unequally into the heat transfer tubes. The understanding of two-phase disri- bution in the manifold syste
5、ms is therefore of great importance for the design and optimization of compact heat exchangers based on parallel flow technology. Mueller and Chiou (1988) stated that many variables affect the two-phase distribution, including geometric factors (manifold cross-section design, branch couplings, locat
6、ion, and orientation of the tubes) and operating factors (flow rate, flow structure and vapor fraction at the inlet of the manifold, and heat load on the tubes). Due to this complexity, no general physically based method has been developed to describe the flow conditions in heat exchanger manifolds
7、and predict the two-phase flow distribution. Nagata et al. (1988) made experiments on a horizontal round tube manifold with four vertical upward tubes. Asoh et al. (1 99 1) studied the two-phase flow in three vertically down- ward tubes and found that refrigerant maldistribution appeared due to two-
8、phase fluid dynamics and non-uniform thermal load. Watanabe et al. (1995a) studied distribution of refrigerant R-1 1 in a horizontal manifold with four upward heat exchanger tubes. In a later study, Watanabe et al. (1 995b) investigated how heat load on the branches affected the two- phase distribut
9、ion. Horiki and Osakabe (1 999) studied water distribution in a horizontal manifold with four vertical branches with and without a small amount of gas-phase. The branch pipes could be protruded into the header, and the effect of prorusion length on the water distribution was studied. Yo0 et al. (200
10、2) studied aidwater distribution in a manifold with fifteen microchannel heat exchanger tubes. Visual observa- tions of the flow regime in the manifold both in vertical and horizontal position were carried out. Some experimental work has also been done on two-phase distribution in plate heat exchang
11、er manifolds (Rong et al. 1996; Bemoux 2000; Bemoux et al. 2001). In Vist and Pettersen (2002), two-phase distribution of R-l34a (an HFC) in round tube manifolds with ten parallel tubes was investigated. The results showed severe maldistribution of both the vapor and the liquid phase. In the upward
12、flow experiments, the vapor phase flow was distrib- uted much easier into tubes near the inlet, and the liquid was preferentially distributed to the last tubes ofthe heat exchanger manifold. Similar results were reported for distribution of CO2 in a manifold with ten parallel microchannel heat excha
13、nger tubes (Vist and Pettersen 2003). Due to the large number of Sivert Vist is a Ph.D. student in the Department of Energy and Process Engineering, Norwegian University of Science and Technology, Trond- heim, Norway. 02004 ASHRAE. 307 both geometry and flow parameters that are of importance for the
14、 distribution of two-phase flow in manifolds, there is an evident need for additional experimental data. Although extensive work on the phenomena of phase splitting in single T-junctions has been published, no generally applicable model for prediction of two-phase flow split has been presented. Such
15、 models are used in one-dimensional codes for calculation of two-phase-flow phenomena in large hydraulic systems, and a number of models have been devel- oped. These models are based on empirical closure relations, and, therefore, are applicable only within the ranges of oper- ational parameters and
16、 geometries that they are based upon. Several authors have performed experiments with two-phase split in T-junctions with horizontal stratified flow and vertical branchtubes, e.g., Reimann et ai. (1988), Seeger et al. (1986), and Maciaszek and Micaelli (1 990). In the present study, CO, is used as t
17、he refrigerant in two different round tube manifolds with ten upward heat exchanger tubes. The inner diameters of the manifolds are 8 mm and 16 mm. First, an outline of the experimental test rig and the measurement procedures is given. Secondly, the experimental two-phase distribution data are repor
18、ted. Together with data using R-134a from Vist and Pettersen (2002), the results are compared to the purely empirical model for phase split in T-junctions of Seeger et al. (1986) and the semi-empirical correlation of Castiglia and Giardina (2002). EXPERIMENTAL METHODS In order to measure two-phase f
19、low distribution under different operating conditions and with different refrigerants, an experimental rig was built in the laboratories of the Institute of Energy and Process Engineering. Refrigerant flow is supplied to the inlet manifold of a heat exchanger in a variety of thermodynamic states, ra
20、nging from subcooled liquid to superheated vapor. The test section was designed to model a car air-conditioning evaporator of approximately 5 kW capac- ity. The test heat exchanger consists of an exchangeable inlet manifold, ten parallel evaporator tubes, and a system of two outlet manifolds. The ev
21、aporator tubes are heated by a closed water circuit. The test facility is depicted in Figure 1, where the flow in the refrigerant loop is driven by a variable-speed gear pump (1). Unlike a compressor, the gear pump requires no lubrica- tion, so the loop can be operated oil-free. The refrigerant flow
22、 is measured by a Coriolis flow meter (2), and an electrical heater (3) is used to heat the subcooled liquid refrigerant to the desired vapor fraction at the inlet of the test section. In the inlet manifold (4) of the evaporator, the flow is divided into ten parallel tubes that are heated by counter
23、flowing water in water jackets. The flow in a single tube in the evaporator can be redi- rected by three-way valves (5) to the outlet manifold of the tap- off circuit (9). The refrigerant in the tap-off circuit is condensed in the condenser (ll), and the mass flow is measured by a Coriolis flow mete
24、r (12). The heat added to the water in the condenser is calculated using the temperature difference and the water mass flow rate, which is measured by a Coriolis mass flow meter (1 3). The main refrigerant flow is condensed in the main condenser (8) before it is mixed with the tap-off flow and led b
25、ack to the pump (1). The operating pressure in the refrigerant loop is controlled by regulating the pressure in the refrigerant tank (19). More details on the test rig were presented by Vist and Pettersen (2002). The test section consists of an inlet manifold (4) and ten parallel 4 mm inner diameter
26、 (ID) evaporator tubes (see Figure 2). At the outlet of each evaporator tube, there is a three-way valve (7), which allows one tube at a time to be redirected to a separate tap-off circuit. The pressure drop in the symmetric three-way valves is equal when the flow is diverted in either of the two di
27、rections. The inlet manifold is exchangeable, and different manifold geometries can easily be used in the I s Wiigumt crwpomtw tuber. 16 1s 1 -Refrigerant pump 2-Mass flow meter 3-Electrical preheater 4-Inlet manifold 5-Three-way valves (1 O) 6-Main outlet manifold 7-Valve 8-Main condenser 9-Tap-off
28、 outlet manifold 1 O-Valve 1 1 -Tap-off condenser 12-Tap-off mass flow meter 13-Water mass flow meter 14-Water tank 15-Water pump 1 : + 50 - 75e (13) where A, is the dimensionless vapor cross-sectional area (Avd =Add?) and h, is the dimensionless liquid height in stratified flow. The ratio of the li
29、quid Weber number to the liquid Froude number is given by The nondimensional empirical exponents Fl(q) and F2(q) include the effect of heat flux on the onset of dryout of the annular film. In the case of adiabatic flow, the values become Fl(q) = 1 .O and F2(q) = 1.023. An analysis using the transiti
30、on criterion defined by Zrcher et al. (1999) and the local flow properties (G, x) along the manifold show that all data series for the ID 16 mm mani- fold are in the gravity-dominated stratified and stratified-wavy region of the flow map. For the CO, and R-l34a data series of the IDS mm manifold, th
31、e flow at the inlet to the first three to eight junctions in the manifold are in the inertia-dominated intermittent flow pattern area. The models of Watanabe, Seeger, and Castiglia were all developed based on data obtained using stratified flow in the main tube. This is a likely explanation of the d
32、ivergence between the models and the experimental results in the region of inertia-dominated flow in the manifold (Figures 8 and 9). When using flow maps for two-phase flow in a manifold, one has to remember that the flow transition criterions are based on fully developed flow, which is not the case
33、 in a mani- fold. At the end of the manifold, downstream effects, such as liquid pooling, will give a higher liquid level than would be predicted by the transition criteria for fully developed two- phase flow. Also, disturbances due to radial flow in the mani- fold caused by the branch tubes will pr
34、obably affect the two- phase flow pattern in the manifold. However, the transition criteria between stratified-wavy and intermittent or annular flow (Equation 13), which is based on the relation between inertia and gravitational forces, is assumed to give a quantita- tive insight in important forces
35、 affecting the two-phase flow pattern in the manifold. The two-phase flow at the inlet of the manifold should be fully developed, passing through a straight pipe section of 15D before entering the ID16 mm manifold (30D for the IDS mm manifold). To bring the effect of flow regime change into the anal
36、ysis of phase split in the manifold, a transition criterion is defined. In Figure 12, the fraction of liquid taken off in the branch tube is plotted against a dimensionless manifold total mass flux, G, i Gm,wuvy Gm,wuvy is the mass flux at the transition between stratified-wavy and intermittent flow
37、 defined in Equation 13. Above a value of approximately G, / Gm,wuvy = 2.0, the liquid take-off is seen to be approximately constant. This effect is not taken into account in the earlier correlations outlined in the previous section, where the liquid take-off is zero above a certain value of vapor m
38、ass flux in the manifold. Another interesting observation is that the CO, experiments seem to give significantly higher liquid take-off in this region compared to the experiments with R-134a. This results in a better two-phase distribution for CO, than for R- 134a when the mass flux at the inlet of
39、the manifold is as high as it is in the ID8 mm manifold experiments. According to Beaver et al. (2000) and Azzopardi (2000), a likely reason for this obser- vation is the small density ratio of CO, (4.2 at 20C) compared to R-134a (34.7 at 2OoC), which allows more liquid to divert into the branch tub
40、e. Figure 11 Vapor fraction in tubes as function of vapor mass Jux in manifold at inlet of tube junctions. Series with tubes I to IO is shown from right to le$. ID8 mm manifold. Refrigerant: CO, at 55.6 bar. Horizontal manifold with upward heat exchanger tubes. 31 4 ASHRAE Transactions: Symposia 0.9
41、 0.8 O.? I I 0.6 f 0.4 0.1 “_ O O - - - - - - O 0.5 1 1.5 2 2.5 3 3.5 GmPln.i.waw -I L 4 J 4.5 Figure 12 Fraction of liquid take-off in tubes as function oj total manifold mass Jux divided by the mass Jux at transition from wavy to intermittent jlow dejned by Zrcher et al. (1999). ID8 mm manifold. V
42、alues from all measurement series are plotted. Horizontal manifold with upward heat exchanger tubes. CONCLUSION A special test rig to measure the two-phase distribution in compact heat exchanger manifolds was built. Flow rates of both liquid and vapor phases were measured in ten parallel heat exchan
43、ger tubes by a system of switching single tube flows into a tap-off measurement circuit. The CO, measurements showed severe maldistribution of both the liquid and the vapor phase in the horizontal manifolds with upward heat exchanger tubes. Especially in the ID16 mm manifold, the vapor phase distrib
44、uted much easier into tubes near the inlet, while the liquid flows down the manifold and is mainly distributed to the last tubes of the heat exchanger. Experiments with the ID8 mm manifold showed that the liquid was more evenly distributed, but still an overfeeding ofthe two last tubes of the manifo
45、ld was observed. Comparisons with existing models for prediction of phase distribution in T-junctions with a horizontal main tube and vertical branch tube showed good agreement for the CO, experiments and the earlier published experiments with R- 134a in the ID 16 mm manifold. Larger deviations were
46、 seen for the ID8 mm manifold, especially in the first part of the manifold experiencing inertia-dominated two-phase flow. Investigation of the phase split in the local junctions of the manifold revealed that there was a strong correlation between vapor fraction in the branch tube and the vapor mass
47、 flux in the manifold below a certain threshold mass flux. At higher mass flux, as observed in the ID8 mm manifold, the nature of the phase split changed to a situation with a constant liquid take- off fraction. Compared to earlier published results with R- 134a, CO, had a higher liquid take-off fra
48、ction in the high mass flux area, leading to improved two-phase distribution in the manifold. The low mass flux experiments in the ID 16 mm manifold did not show significant differences between distri- butions of CO, and R-l34a. Further work is underway to clar- ify the significance of the differenc
49、e between the fluids with regard to two-phase distribution. Also, the consequences of changing the geometry of the manifold will be explored. The results of the phase distribution analysis showed that the two-phase split in the manifold could be regarded as a series of T-junctions (Figures 11 and 12). This finding is important for the development of models to be used in heat exchanger and system simulation tools. By including such phase distribution models, it may be possible to take into account the performance-reducing effects of uneven phase distribution in the manifolds. ACKNOW
